The Role of Regulation and Emotional Eating Behaviour in the Early Development of Obesity

Round 1

Reviewer 1 Report

The paper is concisely written and talks about the important topic of the role of regulation and emotional eating behavior in the early development of obesity.

The following points should be clarified/improved

1. Statistical methods stated that an ANOVA model was used in the analysis. However, it is not stated whether the assumptions of using this model, such as equality of variance or normality of model residuals, were checked.
2. It is not clear from the description of the analyses used whether a three-factor ANOVA or a three times one-way ANOVA was used. If a three-factor ANOVA was used, then there is no need to apply the Bonferroni correction. If three times  one-way ANOVA was used, then Bonferroni correction should be left in place.
3. "ANOVAs (BMI group x gender x age)." This notation in the language of statistics means that interactions between variables were examined. However, Section 2.3 does not describe that an interaction test would be performed. Such an examination of three-factor interaction or smaller i.e. two-factor interaction seems necessary here. If the interaction results turn out to be statistically significant they can greatly enrich the work and the conclusions drawn here.
4. The results for comparisons of COBQ measures between age and for comparisons of COBQ measures between gender are missing from the outcome tables. The results of these analyses should also be included in the tables if the researchers refer to the results of such analyses in the discussion section (lines 211-213)

Author Response

Response to Reviewer 1 Comments

Comments and Suggestions for Authors

The paper is concisely written and talks about the important topic of the role of regulation and emotional eating behavior in the early development of obesity.

The following points should be clarified/improved

Point 1: Statistical methods stated that an ANOVA model was used in the analysis. However, it is not stated whether the assumptions of using this model, such as equality of variance or normality of model residuals, were checked.

Response 1: Authors’ answer: The Levene test was applied, although the reviewer is right as nothing is mentioned about it. We have edited the text on Line 157 to include this information. In addition we note that z-scores were used (so that part of the residual variability is normalised) and the sample is relatively large (more than 120 participants), so checking normality was not strictly necessary.   

Point 2: It is not clear from the description of the analyses used whether a three-factor ANOVA or a three times one-way ANOVA was used. If a three-factor ANOVA was used, then there is no need to apply the Bonferroni correction. If three times  one-way ANOVA was used, then Bonferroni correction should be left in place.

Response 2: A 2x2x2 factorial ANOVA was carried out. That is, it is a three factor ANOVA and not a three times one-way ANOVA. Line 144 has been changed to include these details.

The reviewer is right that the Bonferroni correction was not necessary because there’s only two levels for each factor, and then only one possible comparison. Line 147 has been changed accordingly.

Point 3: "ANOVAs (BMI group x gender x age)." This notation in the language of statistics means that interactions between variables were examined. However, Section 2.3 does not describe that an interaction test would be performed. Such an examination of three-factor interaction or smaller i.e. two-factor interaction seems necessary here. If the interaction results turn out to be statistically significant they can greatly enrich the work and the conclusions drawn here.

Response 3: Interactions were checked as indicated by the degrees of freedom degrees (seven effects: 3 principals and 4 interactions plus the interception, 115+8= 123). For the scales considered, no significant interactions were found. Line 162 has been changed to include this information.

Point 4: The results for comparisons of COBQ measures between age and for comparisons of COBQ measures between gender are missing from the outcome tables. The results of these analyses should also be included in the tables if the researchers refer to the results of such analyses in the discussion section (lines 211-213)

Response 4: They were not included following APA’s recommendation of not repeating information in text and tables. There were not significant differences for these two scales between genders and between ages. Again it is correct that this was not specified in the results section but was in the conclusions. We apologise for this omission and have added this information in Lines 180-185 and changed the writing in line 225.  

Reviewer 2 Report

Valero-García et al address an important matter: the relationship between childhood overweight/obesity and some obesogenic behaviors such as emotional eating and the parent’s use of food to regulate their children behaviors. The study shows important results and are of interest for readers. However, major changes should be address prior to publication.

General review of the manuscript

The manuscript as written is difficult to follow. One of the major issues is the English writing. Please make a thorough review of the entire manuscript prior to the new submition.

Title:

Please see comment of the aim of the study and change your title. We don’t know if any of the behavioral components have a role in the early development of obesity.

Introduction

Authors did an extensive review of the literature.

Aim

Please rewrite your aim. The current study is design to study the relationship of some behaviors with obesity. It is a cross-sectional study and only associations can be obtained. You cannot state that obesogenic behaviors are influencing their weight.

If the aim was only to stablish any relationship between obesity and the scales 2 and 4 of the COBQ why did you asked caregivers to answer the rest of the test?

Methods

The study design is appropriate.

Please described if the socioeconomic status of all nurseries and schools participating was the same. Did you recruited the same amount of normoweight and overweight/obesity children in the different schools? High socioeconomic status tend to have lower incidence of overweight and obesity at the ages 4 and 7. I am concerned that children overweight and obese were from other schools with different socioeconomic environment. Please comment on the methods section and include some aspects in the discussion if needed.

Please describe scales used for height and weight.

If the category of normoweight, overweight and obesity were determined using percentile for age and sex, please do not mention z-score in the methods section unless is used for a specific purpose.

Please move the characteristics of participants at baseline to the results section.

Why only 9, 7-year-old normoweight boys, were included in the study vs more than 15 in any other category?

Line 108-109:  parents were asked to complete a brief battery of questionnaires. Please specify in this section the questionnaire. Example: parents were asked to complete The Childhood Obesogenic Behaviours' Questionnaire (COBQ) … and then go on describing this test as you did in line 118

Results

Include in this section the characteristics of study participants

Please describe the results of all the dimensions of the COBQ for all age/sex groups.

Please address if there is any particular are/question that was more relevant to separate normoweight from overweight/obese groups

Please state if there is any differences if the mother or the father answer the COBQ. Please mention if there are differences in schools from different socioeconomic status.

Discussion and conclusions

Content is excellent. However, the writing needs to be reviewed.

Abstract

Please review abstract to adapt all changes discussed for the manuscript. Please include real results in the abstract (higher COBQ score…) instead of describing only the conclusions.

Other minor issues

I suggest using the term normoweight instead of non-overweight as underweight children are excluded from the study.

Author Response

Response to Reviewer 2 Comments

Valero-García et al address an important matter: the relationship between childhood overweight/obesity and some obesogenic behaviors such as emotional eating and the parent’s use of food to regulate their children behaviors. The study shows important results and are of interest for readers. However, major changes should be address prior to publication.

General review of the manuscript

Point 1: The manuscript as written is difficult to follow. One of the major issues is the English writing. Please make a thorough review of the entire manuscript prior to the new submition.

Response 1: Thank you for this suggestion, it has now been edited by the MDPI translation service. We would also like to note that it had been reviewed by a professional native-English speaker before sending it to the journal.

Title:

Point 2: Please see comment of the aim of the study and change your title. We don’t know if any of the behavioral components have a role in the early development of obesity.

Response 2: Thank you - the title has been changed as we realised, based on this comment, that it could lead to confusion, given that the paper is focused on emotional regulation.

Introduction

Point 3: Authors did an extensive review of the literature.

Response 3: Thank you for this feedback.

Aim

Point 4: Please rewrite your aim. The current study is design to study the relationship of some behaviors with obesity. It is a cross-sectional study and only associations can be obtained. You cannot state that obesogenic behaviors are influencing their weight.

Response 4: We have reworded this aim to remove the use of the term “influence” to more accurately reflect the cross-sectional nature of the study.  Please see Line 91.  

Point 5: If the aim was only to stablish any relationship between obesity and the scales 2 and 4 of the COBQ why did you asked caregivers to answer the rest of the test?

Response 5: We have reworded this section to more clearly set up that just two subsacles of the COBQ were used in this study (see Lines 128-132).

Methods

The study design is appropriate.

Point 6: Please described if the socioeconomic status of all nurseries and schools participating was the same.

Response 6: As stated in line 108, families were middle-class. We recruited children from a limited area of a regional city so that all schools were in the same kind of neighbourhood.

Point 7: Did you recruited the same amount of normoweight and overweight/obesity children in the different schools?

Response 7: We asked parents for consent and once we had then, we matched the same amount of children who were non-overweight and overweight from the same classroom, to minimise differences between both groups.

Point 8: High socioeconomic status tend to have lower incidence of overweight and obesity at the ages 4 and 7. I am concerned that children overweight and obese were from other schools with different socioeconomic environment. Please comment on the methods section and include some aspects in the discussion if needed.

Response 8: There were not differences between schools in socioeconomic level as mentioned above. We have stated this in the Method (line 108).

Point 9: Please describe scales used for height and weight.

Response 9: The instruments have been detailed in line 116-117.

Point 10: If the category of normoweight, overweight and obesity were determined using percentile for age and sex, please do not mention z-score in the methods section unless is used for a specific purpose.

Response 10: Percentiles were used to classified children in the two categories of weight. z-scores were used, as in many other published papers, to minimise variability considering the great differences between children in growth. See, for example, the review by Must, A. & Anderson, S.E. (2006). Body Mass Index in children and adolescents: considerations for population-based applications, International Journal of Obesity, 30, 590-594.

Point 11: Please move the characteristics of participants at baseline to the results section.

Response 11: We included participants in the Method section according to APA’s recommendations as nothing is stated in the guidelines for authors of IJERPH (https://www.mdpi.com/journal/ijerph/instructions#preparation). We have checked and both styles are used within the journal. See, for example,  https://doi.org/10.3390/ijerph18010355 or  https://doi.org/10.3390/ijerph18052782 (in methods) while https://doi.org/10.3390/ijerph18010357 (In results). Our preference is therefore to leave this information where it is.

Point 12: Why only 9, 7-year-old normoweight boys, were included in the study vs more than 15 in any other category?

Response 12: As mentioned above, we matched children who were overweight with those who were not overweight from within the same classroom but we could only include children for whom we had parental consent. Sometimes, questionnaires were not fully completed and we had to discard subjects.

Point 13: Line 108-109:  parents were asked to complete a brief battery of questionnaires. Please specify in this section the questionnaire. Example: parents were asked to complete The Childhood Obesogenic Behaviours' Questionnaire (COBQ) … and then go on describing this test as you did in line 118

Response 13: Parents only completed the Childhood Obesogenic Behaviours' Questionnaire so we have removed the use of the phrase ‘battery’ to avoid confusion, and we have signposted from 2.2. the reader to below, for details of the measure when we first mention it. 

Results

Point 14: Include in this section the characteristics of study participants

Response 14: We explain above why we would prefer to keep the participants in the methods section.

Point 15: Please describe the results of all the dimensions of the COBQ for all age/sex groups.

Response 15: Only two of the COBQ subscales are the focus of this article. We have added the data for age and sex in line 179-184. As differences were not significant, a table was considered not necessary.

Point 16: Please address if there is any particular are/question that was more relevant to separate normoweight from overweight/obese groups

Response 16: Apologies but we don’t quite understand this question.

Point 17: Please state if there is any differences if the mother or the father answer the COBQ.

Response 17: Only mothers completed the COBQ in this study.

Point 18: Please mention if there are differences in schools from different socioeconomic status.

Response 18: We selected schools from the same geographical area so that differences in socioeconomic status were minimal.

Discussion and conclusions

Point 19: Content is excellent. However, the writing needs to be reviewed.

Response 19: All the paper has been reviewed as said above.

Abstract

Point 20: Please review abstract to adapt all changes discussed for the manuscript. Please include real results in the abstract (higher COBQ score…) instead of describing only the conclusions.

Response 20: We have added real results in the abstract (lines 19-22).

Other minor issues

Point 21: I suggest using the term normoweight instead of non-overweight as underweight children are excluded from the study.

Response 21: In view of the drive to avoid causing or perpetuating weight stigma and to use person-first language (https://www.worldobesity.org/what-we-do/our-policy-priorities/weight-stigma), our preference is not to use the term "normoweight" as this includes the term 'normal' which implies that anything outside of this is not normal, which can be deemed stigmatizing.  We therefore feel that our original term is non-stigmatizing and more factual.  We have changed our wording on line 107 for consistency.

We would like to clarify that with the term 'non-overweight' we refer to children with weights between overweight and underweight.