Effect of a Mediterranean Diet-Based Nutritional Intervention on the Risk of Developing Gestational Diabetes Mellitus and Other Maternal-Fetal Adverse Events in Hispanic Women Residents in Spain

Round 1

Reviewer 1 Report

• Line 4 and in the rest of the manuscript: Title: …maternal-fetal…
• Line 26 and in the rest of the manuscript: I don’t think the word “intervention” is appropriated. The authors should name the groups as: Mediterranean diet, low fat diet, and regular diet.
• Line 41: first paragraph has only one sentence.
• Line 44: this sentence is very confusing.
• Line 48: Our group has demonstrated …
• Line 49: I still don’t understand “real world”. Can the authors use a better way to describe it?
• Line 52: by appearance the authors mean incidence, right?
• I think the whole manuscript needs a complete review in terms of grammar. The English is very broken.
• Line 106: by visit 0 the authors mean visit 1 right? There is no way to have a visit number 0. Unless the authors are talking about time point, so it would be time point 0 and so on and so forth
• Figure 1: confusing, I lost track of the distribution of the participants. Also, the legend is very incomplete
• Line 162: this whole section here needs to be re written
• Please, review the whole manuscript.

Author Response

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

Line 4 and in the rest of the manuscript: Title: …maternal-fetal…

Thank you for your appreciation. This term has been corrected in the whole manuscript.

Line 26 and in the rest of the manuscript: I don’t think the word “intervention” is appropriated. The authors should name the groups as: Mediterranean diet, low fat diet, and regular diet.

We have reviewed the whole manuscript to clarify this concept, clarifying that the word intervention is accompanied by the word MedDiet-based or nutritional and in other cases changing the word to "nutritional therapy”. We hope that in this way it is better understood.

Line 41: first paragraph has only one sentence.

Thank you for your appreciation. We have modified the first paragraph and divided better all of the paragraphs in the introduction.

Line 44: this sentence is very confusing.

We have modified this sentence. We hope this time it is better understood.

Line 48: Our group has demonstrated …

Thank you for your appreciation. It has just been corrected in the text.

Line 49: I still don’t understand “real world”. Can the authors use a better way to describe it?

The term "real world" refers to the fact that this group was given nutritional recommendations based on what the usual clinical practice would be like, that is, what would happen in real life. This term is accepted. However, we have modified the text although the abbreviation “RW” remains. Instead of putting "extension-real world" we have put “study based on a real world context” or “usual clinical practice” in the whole manuscript to facilitate its understanding.

Line 52: by appearance the authors mean incidence, right?

We agree with that point. This word has been changed in the text.

I think the whole manuscript needs a complete review in terms of grammar. The English is very broken.

Thank you for your suggestion. The whole manuscript has been reviewed by an American native speaker.

Line 106: by visit 0 the authors mean visit 1 right? There is no way to have a visit number 0. Unless the authors are talking about time point, so it would be time point 0 and so on and so forth.

We agree. Following your suggestion the term “visit 0” has been changed to “visit inclusion” as well in the text as in figure 1.

Figure 1: confusing, I lost track of the distribution of the participants. Also, the legend is very incomplete.

Thanks for your suggestion. The figure has been modified to facilitate its understanding. The RCT and the "real world study" or study based on the usual clinical practice, are two different studies and were not simultaneous but successive. The development has been explained and the participants involved have been described in both studies. The participants who attended the first visit (12-14 GW) as the women who continued in the study and who were analysed after follow-up at visits 2 (24-28 GW), 3 (36-38 GW) and 4 (delivery) have been marked in bold. The legend has been also modified and completed.

Line 162: this whole section here needs to be re written

We have changed this paragraph. We hope this time it is better understood (section 2.3.1, line 175).

Please, review the whole manuscript.

Thank you for your suggestion. The whole manuscript have been reviewed.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

Abstract,

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

Table 1, The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

Lines 163-168, not clear what methods were used.

Lines 170-173, was height measured?

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

Line 208, use length (this is very minor)

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

Results

The results should have 95% CI for the incidence rates.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Total GWG should be shown, at least from study inception to delivery.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Author Response

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Thank you very much for your kind comments and constructive suggestions. We agree with you on most points. We have made the corresponding changes to improve the article.

The changes applied are as follows:

 

This paper is a summary of a trial on the Mediterranean diet and GDM in Hispanic women residing in Spain. In addition to the trial, a practice based convenience sample was followed after the trial.  Overall the work is interesting and important.  However, the way the results are presented is problematic, i.e. means vs cut-points.  In addition, there is very little information on diet, the scores are shown, but we still do not have a good sense in reading the report what the diet was like.  Some basic nutrient intake should be included as pertains to GDM.

 

Abstract

Is clearly written.

Add the dates of the trial. As written the “Extensively” is not clear on line 27, what does that mean in terms of a real world group?  Was it just another group, not really linked to the trial or were the real world participants randomized? It is not clear. Sample is not equal so maybe just a convenience sample after the trial; this becomes clear in the main methods of the paper.

Thank you for your comments. Following your suggestions, we have added this information in the abstract. Secondly, the word “extensively” has been changed to “successively” because this group is apart from the RCT. It was successive to the RCT and they received the same recommendations based on the Mediterranean diet principles as if it was a usual clinical practice including the consumption of EVOO and nuts (without supply in contrast to IG). We have modified this in the abstract so that it is better understood.

According to the sample, the RCT was randomized in two groups. From 1000 chosen participants after being categorized by age, BMI and parity 142 CG and 143 IG attended the visit 1. Lastly, after completed the four visits, 132 and 128 Hispanic women were analysed. In total 260 women were finally analysed. Whereas in the “Real world group” which was a unique group, 315 women attended visit 1 and after having completed the four visits, 284 were analysed. There is no vast differences between the participant who were finally analysed and it is within what is expected in the study.

It is interesting to see the high rate of GDM in the control group, but even the intervention groups have high rates. This should be included in the discussion/concluding lines.

Due to the fact that the method used to diagnose GDM is according to IADPS criteria, the results in relation to the rates of GDM can be different depending on the criteria used. Using this diagnosis criteria, the rates of women diagnosed with GDM tent to be higher, for this reason the IG presents also higher rates of GDM. However, the difference between the three groups continue being significant. Following your suggestion, we have added a line in the discussion clarifying this explanation (lines 303-305).

Introduction

The introduction has appropriate content. The structure of the paragraphs needs attention, all of them; some are too short, others too long, linking statements not always evident; these are easy to address.

Introduction structure has been modified. We hope in this way the paragraphs are clearer.

Methods

Some explanation for pistachio nuts would help the reader to appreciate the design further; no other options such as walnuts?

The recommendations given to women in IG and RW study were to increase the consumption of extra virgin olive oil (EVOO) and any kind of nuts, in the IG especially pistachio nuts, ─which were the type of nuts used in this study─ and were supplied. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts.

 

 

Table 1. The nutrition score and physical activity score are not clear in table 1. The footer might be adapted to help the reader given this table is early in the methods section.

We agree. We have added a concise line explaining where the nutrition score and physical activity score come from: the Diabetes Nutrition and Complications Trial (DNCT) which was extensively explained in PLoS ONE 2017, 12, e0185873. We have added this information as well in the section 2.3.3 (line 210-211) as in the table footer.

Line 118, how was the intervention strengthened? Do you mean more education or additional features?

During visit two, the OGTT was used to reinforce the recommendation based on the Mediterranean diet, encourage the consumption of nuts (GI and RW) and limit the fat-intake in CG in a meeting for about two hours. In turn, a personal visit was made where the information present in the questionnaires was taken. Clarifying this concept we have added this explanation on lines 125-128.

Lines 163-168, not clear what methods were used.

This paragraph has been newly rewrite. All this information related to clinical aspects were recorded at first visit within the interview through the questionnaire described in section 2.3.3 “Dietary and lifestyle assessment” (line 175).

Lines 170-173, was height measured?

Yes, height was measured (undoubtedly without shoes) in a height rod. To clarify this concept we have added this data on line 182-183.

Section 2.3.2, quality assurance data are needed; accuracy and precision as well as any certification programs.

We agree with your suggestion. According to Insulin is a immunoassays chemiluminescent in solid fase. It is carry out in Immulite 2000 xPi (Siemens) with an inter-assay accuracy  in concentrations of 11 uIU/mL of 6,3% and for insulin concentration of  21 uIU/ml of 5,91 %.

An External Quality Guarantee Program of the SEQC (Sociedad Española de Química Clínica) monthly evaluates the quality of the methods.

 

Inter-assay imprecision of HbA1c in the Hospital laboratory is of:

-For levels of 5.1 %, there is S.D of 0.06 and a C.V of 1.23%.

-For levels of HbA1c 10.39%, S.D = 0.11 and C.V = 1.04.

We have added this information on line 193-200.

Section 2.3.3, include information on the validation of the questionnaires for diet and activity.

This questionnaire was previously validated in the Diabetes Nutrition and Complications Trial (DNCT). It was previously explained in the prior study (PLoS ONE 2017, 12, e0185873). We have added this information on line 210-211.

Line 208, use length (this is very minor).

Thank you for your appreciation. This term has been changed.

Section 2.4, if the null hypothesis is “null”, why 30% reduction used? This is not clear. Why was time tested separate from the effect of the intervention? Seems like too many tests.

For sample size calculation, a primary end-point for conversion to GDM has been used assuming an expected conversion rate of 25% between 24-28 GWs and a reduction of at least 30% in the conversion to GDM after intervention. This intervention has shown a decrease of FPG -7 (-13; -1.3) mg/dl after 3 months in the PREDIMED Study (Ann Intern Med 2006 Jul 4; 145(1):1-11), representing a potential reduction of GDM rate higher than 30%.

We recalculated the sample size in this post hoc study in Hispanic women assuming an expected incidence about 15%, and for a predicted drop-out of around 20% in accordance with the expected response rate from initials results from our previous study ( PLoS ONE 2017, 12, e0185873), a sample size of 315 women (RW)  and 143 women (IG) the study provided a statistical power of 80% (2-tailed, α error of 0.05) detecting a relative risk (RR) reduction of at least 30% in IG and RW vs. CG.

The trial groups are compared, this is appropriate. Is it appropriate to compare the RW vs CG? These were not part of the same trial or time-frame.  Other methods might be more appropriate.

The only difference between IG and RW is that EVOO and pistachio nuts were supplied in the case of IG but both received the same recommendations and were encourage to follow the Mediterranean diet and increase their nuts and EVOO intake. In contrast, the CG was recommended to limit fat-intake. That is why we think they can be compared.

However, we are aware that these two groups were not analysed in parallel. For this reason we consider that this may be a limitation to our study and we have discussed it in the limitations.

Results

The results should have 95% CI for the incidence rates.

We agree. These data have been supplied.

The mean glucose for the OGTT is not really informative in the context of reducing GDM; the proportion above the cut-point for diagnosis of GDM should be shown. Just looking at the 2 h CG mean of 107.7 mg/dl is ~ 6 mmol/L which is normal; far below a cut-point for GDM.

We agree. These data have been supplied.

The HbA1C values are within the normal range, as a mean, how many were above normal?

Consider that the levels of glucose and HbA1c refer to the total population, that is, women with GDM and also with normal glucose regulation. For this reason, it is expected to find levels within the normal range. Consider that most of them are non-GDM and that GDM is not the same as pre-pregnancy DM, regarding HbAc1 levels and glycemic targets. Glycemic Targets During Pregnancy (Expert Recommendations) for Patients With Gestational Diabetes Mellitus (GDM):(LeRoith D, et. al. Endocrinol Metab Clin N Am. 2011;40(1): xii-919) is HbA1c <5% and Gluce < 90 mg/dl.

 

Trimester-specific reference intervals for HbA1c in pregnancy in non-diabetic caucasian pregnant women are:  Median (Q1-Q3) T1: 5.0% (4.3-5.4) T2: 4.9% (4.4-5.4) and T3:5.1% (4.7-5.7) (doi 10.1515/CCLM.2011.397). The cut-off point for HbA1C was considered 5.5% as reflected in this study (doi: 10.1136/bmjdrc-2018-000550).

Total GWG should be shown, at least from study inception to delivery.

These data are shown in Table 3 both in the assessment at 24-28 and 36-38 GW.

Discussion

While differences were observed, the mean values for many of the measurements were not different in a way that helps the reader to understand why GDM was essentially reduced by 50%. The authors do not comment on this. 

In the previous study (PLoS ONE 2017, 12, e0185873), the incidence of GDM was observed to decrease by 25%. In the present study, according to the rates obtained in the CG (25.8%) and comparing them with the IG (14.8%) and RW (13.4%), the decrease does not reach 50% but it does exceed 40%. This information can be seen in table 3. We have added a concise sentence on line (305-306) in the discussion explaining it.

It is also possible that the data are highly skewed and with unequal variances. It did not come out in the statistical section how this was addressed.  The very small differences in fasting blood glucose along with NS 2 h blood glucose makes the reader question, how was it that GDM was reduced by 50% then in the IG? The proportion below various targets needs to be added to the paper and then this discussed.

Both are based on means but the differences in the standard deviation mean that the variances are not equal. Table 3 shows how in FBG the typical deviation is 7, while in the assessment at 2 and 3 hours it exceeds 20 mg / dL.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I would suggest changing the names of the countries in the supplementary figure to English

Author Response

I would suggest changing the names of the countries in the supplementary figure to English

We agree. Following your suggestions, “Dominican Republic and Brazil” have been modified in the supplementary table.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript is much improved. The additional details in the methods are appreciated.

 

For the RW group, it is not really routine per se, it is advanced care or implementation of the research into practice. 

 

Lines 425 onward, it is still not clear why the intervention was that amount of olive oil and why pistachios were used vs other nuts.

 

Dietary information presented seems limited to scores; some information on choices or total intakes of key nutrients would be of value to the reader.

Author Response

Comments and Suggestions for Authors

The manuscript is much improved. The additional details in the methods are appreciated.

 For the RW group, it is not really routine per se, it is advanced care or implementation of the research into practice. 

We agree. We have added this term (“implementation/application of the research into clinical practice”) on lines 71-72, 84 and 87.

Lines 425 onward, it is still not clear why the intervention was that amount of olive oil and why pistachios were used vs other nuts.

We agree. We have added on lines 122-128 this concise paragraph:
“10 L of EVOO ( about 1 L each 10 days) and 2 Kg of roasted pistachios (about 160 grams a week) were provided at no cost to women in the IG at visits 1 and 2 in order to ensure the consumption of the minimum amount recommended throughout pregnancy. The pistachio was chosen as a dried fruit in the study since those used for the study were of Spanish production and it has also been seen that the fact that the pistachio does not come peeled but with the shell stimulates the cerebral cephalic phase, which is why it is interesting compared to others nuts”.

Dietary information presented seems limited to scores; some information on choices or total intakes of key nutrients would be of value to the reader.

Although your comments are interesting, from the application of the semi-quantitative questionnaires on the frequency of food consumption used in our study, the Med score and the Nutrition score, neither the caloric consumption nor its distribution in immediate principles can be obtained with precision. For example, in carbohydrate-rich foods, the frequency of consumption of whole grains instead of refined is estimated: bread, rice and pasta, but not each of them. In relation to fish, the consumption of fatty fish at least 3 days per week instead of white, but the consumption of white fish is not estimated. The same for fatty and/or processed meats instead of white meats.
The same applies in relation to other foods. The objective of these questionnaires is to evaluate the choice of "healthy" foods instead of those considered "less healthy" or "neutral". Therefore, it is not possible to estimate caloric consumption or its distribution.

The Nutrition score and physical activity score is as follows:

Score +1 ; 0; -1; respectively

Physical activity Score: 1. Daily walks (≥ 5 days/week)  >1 h; At least 30 min;  <30 min

.2. Stair climbing (no floors ⁄day, ≥ 5 day a week)>16;  4 -16; <4.  3. At least 30 min of more than moderate intensity >3 days ⁄week; 2 or 3 days ⁄week; <2 days ⁄week

Nutrition Score:  (Servings per week): 1. Vegetables >12; 6 -12; <6. 2 Fruits (pieces) >12; 6 -12; <6 .  3. Nuts >3; 1 -3; <1; 4. Extra Virgin Olive oil  Daily; 3-5 days; <3 days. 5. High-fat fish or Iberico ham >3; 1 -3; <1. 6. Bread and cereals (high fiber content) >6; 3 -6; <3. 7. Legumes >2; 1 -2; <1. 8.  Low-fat milk and cheeses >6; 3 -6; <3. 9. Red meat/processed <3; 3 -6; >6. 10. Sauces (except homemade “sofritos”)<2; 2 -4; >4. 11 Juices and sugar-sweetened beverages <2; 2 -4; >4. 12 Cookies <2; 2 -4; >4

We supplied the questionnaires in supplementary table 2 (online). You understand that it is not possible to supply caloric consumption or distribution data.

Author Response File: Author Response.pdf