Perioperative Urinary Catheter Use and Association to (Gram-Negative) Surgical Site Infection after Spine Surgery

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

On page 1, line 29 the word "recur" doesn't make sense.  A possible substitution would be turn.  

On page 3, line 125 the authors indicate that 328 sampling of urinary cultures were done in 6% of all patients.  I suspect that they mean sampling of patients with urinary catheters which would be 328/2734 or 12%.  Or, if there was urinary cultures done on patients without catheters, that should be indicated.  

Comments on the Quality of English Language

English quality O.K. with the exception of line 29 which is commented above.

Author Response

Dear Reviewer 1,

Thank you for your valuable inputs and suggestions.

Reviewer Nr. 1:

Comment Nr. 1: On page 1, line 29 the word "recur" doesn't make sense.  A possible substitution would be turn.  

Ad comment Nr. 1: We substituted the word "recur" by the word "turn" in the revised manuscript, as you advised it.

 

Comment Nr. 2: On page 3, line 125 the authors indicate that 328 sampling of urinary cultures were done in 6% of all patients.  I suspect that they mean sampling of patients with urinary catheters which would be 328/2734 or 12%.  Or, if there was urinary cultures done on patients without catheters, that should be indicated.

Ad comment Nr. 2:  For more clarity, we modified the concerned sentence according to your comment as follows in the revised manuscript (page 3, line 129): In 328 episodes (328/5,485; 6%), the clinicians sampled a microbiological urinary culture analysis (patients with or without UC). 

Reviewer 2 Report

Comments and Suggestions for Authors

The work Perioperative urinary catheter use and association to (Gram-negative) surgical site infection after spine surgery is a study that analyzes spine surgeries performed over six years and surgical site infections. The work undoubtedly required great coordination among all participants, mainly due to the large amount of information that was reviewed. The statistical analysis was of great relevance to meet the objectives of the work. However, the study does not show different results from previous work carried out with the same objectives. On the other hand, the results to define the association between catheter use and surgical infection site, cannot be corroborated with the data collected in the study, mainly because they do not have conclusive information to ensure that the microorganism isolated from the wound is the same one that was in the catheter or in the hands of those who participated in the surgeries. The aforementioned is due to the fact that only the genus of the bacteria is described, but not specific characteristics of the identified microorganisms. For this, and as an example in the case of Escherichia coli, the serotype of the isolated microorganisms should have been identified, or the sensitivity profile to antimicrobials known, so it could be stated with greater certainty that it is the same microorganism. I should mention that in the same manuscript it is stated that the reported work has at least three limitations, a statement that suggests that you are aware that the work does not provide more information than that which has previously been reported. Finally, in the conclusión (SSIs after spine surgery seem associated with UC carriage, especially Gram-negative SSI’s in female patients. In other words, a UC carriage might not only be a risk for nosocomial UTI, but might also be associated with early SSIs after spine surgery. This observation indicates careful indication for UCs in the perioperative period around spine surgeries), the comments affirm that the study does not obtain conclusive information.

Author Response

Dear Reviewer 2,

Thank you for your valuable inputs and suggestions.

Reviewer Nr. 2:

Comment Nr. 1: On the other hand, the results to define the association between catheter use and surgical infection site, cannot be corroborated with the data collected in the study, mainly because they do not have conclusive information to ensure that the microorganism isolated from the wound is the same one that was in the catheter or in the hands of those who participated in the surgeries. The aforementioned is due to the fact that only the genus of the bacteria is described, but not specific characteristics of the identified microorganisms. For this, and as an example in the case of Escherichia coli, the serotype of the isolated microorganisms should have been identified, or the sensitivity profile to antimicrobials known, so it could be stated with greater certainty that it is the same microorganism. 

Ad comment Nr 1: The E. coli serotypes profiles to antimicrobials are not included in our retrospective database. In contrast, we have the antibiotic susceptibility pattern, which we highlight now on page 2, lines 74-76. Moreover, we now acknowledge in the “Limitation” section (page 9, lines 273-277) that we cannot perform retrospective serotyping.

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for the opportunity to review this hard-worked paper. It's clear to see that a lot of work and thought has been done.

The use of a UC is a perhaps one of the forgotten issues in orthopaedic surgeries, as it is considered  "peri-operative", but, as you show in your paper and as shown in the past can affect significantly the results of the surgical procedure.

The main question of the paper, whether surgical site infections in spine surgery is associatred with urinary catheter, is a practical and important question. This study adds another valuable proof that UC should be used carefully, and that the surgeon should be responsible for it and not only the anesthesiologist. While common sense would say that UC is associated with only UTI and not SSI, this paper proves otherwise.

The introduction is concise. The conclusions are appropriate to the results.

I was also intrigued about the length of stay,  days seems to me like a long time, especially in the non UC group which is younger, healthier and with shorter surgeries. I am wondering if the results are not biased by high DM rate, older patients, high BMI Etc. The next step is to perform a smaller scale study, but a prospective one, to examine the same question.

Overall, I recommend to accept the paper in its present form

 

Author Response

Dear Reviewer 3,

Thank you for your valuable inputs and suggestions.

Reviewer Nr. 3:

Comment Nr. 1: I was also intrigued about the length of stay, days seems to me like a long time, especially in the non UC group which is younger, healthier and with shorter surgeries

Ad comment Nr. 1: These are data for Switzerland and not surprising in our surgical context. Both groups included patients with similar health status, as both groups had equivalent median ASA scores. This fact and the fact that the UC group still included 22% of revision surgeries and 25% of implant-related surgeries, might explain the median length of hospital stay of 5 days.

 

Comment Nr. 2: I am wondering if the results are not biased by high DM rate, older patients, high BMI Etc.

Ad comment Nr. 2: No. The case mix is certainly huge, but after a case-mix adjustment by multivariate Cox regression analyses, we failed to associate specifically Gram-negative SSI to the patients’ underlying co-morbidities (in contrast to the occurrence of SSI in general). We re-run the analyses in many ways, and found no major detectable biases in terms of unreported co-morbidities. Indeed, only the duration of UC use and the female sex remained statistically associated with Gram-negative SSI, making a clinical relationship plausible. We report this in several parts of the manuscript and Tables.

Reviewer 4 Report

Comments and Suggestions for Authors

I have read with interest the manuscript submitted by Ansorge et al.

I have some comments to be addressed in order to improve the quality of the manuscript:

Tables 1,2 - for each parameter evaluated, it should be included the total number (when applicable). for female patients, for example, 46% were urinary catheterized, while 56% were not. Please correct it (it cannot be 102%). Also, the values for p do not seem correct. How can they (almost) all be 0.01, even though the variables ranged widely?

- all Latin bacterial names should be italicized

- row 149 - Enterobacter spp. is not a non-fermenting GNB.

- the authors mentioned that there were 67 SSIs, yet according to Table 2, the total number of isolates was 134. Please clarify. Also, considering that the study also focuses on UTIs and urinary catheterization, it would be indicated to add further information about those types of infections.

If possible, further information about the antibiotic susceptibility of the isolated microorganisms (especially since you covered this topic in the discussion section), the treatment received, and the outcome of the patients would add value to the study.

- The material and methods section should include further information about the microbiological methods used.

- The conclusion section should be rephrased, including statistically based information.

Comments on the Quality of English Language

moderate English and punctuation needed.

Author Response

Dear Reviewer 4,

Thank you for your valuable inputs and suggestions.

Reviewer Nr. 4:

Comment Nr. 1: Tables 1,2 - for each parameter evaluated, it should be included the total number (when applicable). for female patients, for example, 46% were urinary catheterized, while 56% were not. Please correct it (it cannot be 102%). Also, the values for p do not seem correct. How can they (almost) all be 0.01, even though the variables ranged widely?

Ad Comment Nr. 1: In tables 1 and 2, the total number is shown in the second line. We added a black vertical line between the second and the third column of table 1 for more clarity. One should for example read in Table 1 for the parameter "female sex"  that  within the group without UC (n=2,751), 1,265 or 46% (1,265/2,751) had a female sex and that within the group with UC (n=2,734), 1,541 or 56% (1,541/2,734) had a female sex. The addition of 46% of female sex patients in the no UC group to 56% of female sex patients in the UC group do not correspond to 100% because the no UC and the UC groups are of different size (n= 2,751 versus 2,734). Concerning the p-values, we checked them and found no error. However, the reporting is rounded up to 0.01. Instead of writing 0.0078, 0.0092, etc, we all rounded them up to 0.01, which is much more pleasant to read. We now say in the Statistical Analyses that we round up the p-values to no more than two decimal places (page 3, lines 108-109).

 

Comment Nr. 2: - all Latin bacterial names should be italicized

Ad comment Nr. 2: We modified the revised manuscript according to this comment.

 

Comment Nr. 3: - row 149 - Enterobacter spp. is not a non-fermenting GNB.

Ad comment Nr. 3: Yes. Thank you. We modified the concerned sentence in the revised manuscript (see page 4 lines 150-152) according to this comment as follows: Those most frequent prophylaxis-resistant bacterial group in the urine were Gram-negative rods such as Enterobacter spp. or Pseudomonas spp., followed by enterococci. This is now easier to read.

 

Comment Nr. 4: - the authors mentioned that there were 67 SSIs, yet according to Table 2, the total number of isolates was 134. Please clarify. 

Ad comment Nr. 4: Nostra culpa. Thank you. The title of Table 2 was not enough precise. We modified it according to this comment as follows: Table 2. Group comparisons with all own surgeries (left); and between Gram-positive and Gram-negative surgical site infections (right), including all own SSIs (n=67) and 97 external SSIs.

At the lines 90-94 and 109-110 (first submitted manuscript version) it is further explained that sixty-seven SSIs occurred among the included 5,485 own spine surgeries. It is also stated there that for the evaluation of risk factors for Gram-negative SSI, the sample size was increased by adding 97 transferred patients with infections that have been operated previously elsewhere.

 

Comment Nr. 5: Also, considering that the study also focuses on UTIs and urinary catheterization, it would be indicated to add further information about those types of infections.

Ad comment Nr. 5: At the lines 145-150, the first submitted manuscript reported a symptomatic UTI rate of 3%, of which 40% of the urinary pathogens were resistant to the prior prophylaxis (cefuroxime) during the index surgery. The genus of the most frequently encountered pathogens are also reported. However, our database only focuses on UTI and catheter use only in operated spine patients. As for the UTI, we already mention that they are clinically symptomatic UTIs (for example on page 4, line 148). We also say now that we did not detect clinical pyelonephritis (page 4, line 150). In contrast, we cannot retrospectively analyse for every case the exact UTI symptoms on the day of diagnosis (e.g. the intensity of pollakiuria, the degree of bladder or urethral pain, discomfort etc.), which we equally acknowledge in the “Limitation section” (page 9, lines 277-281).

 

Comment Nr. 6: If possible, further information about the antibiotic susceptibility of the isolated microorganisms (especially since you covered this topic in the discussion section), the treatment received, and the outcome of the patients would add value to the study.

Ad comment Nr. 6: Here, we beg to differ. This is beyond the word limitation of the manuscript and would substantially complicate the message. Please bear in mind that our article is about the postoperative epidemiology of UTI and SSI after spine surgery, and not targeting the treatment success of postoperative UTI’s per se. But we understand Reviewer 4 and have another solution. We propose to resume the fate of the UTIs by saying “All symptomatic UTI have been treated by a targeted systemic antibiotic therapy (range of duration, 3 to 10 days) and were ultimately classified as cured from UTI. No patient died of urosepsis or UTI.” (page 4, lines 152-155).

 

Comment Nr. 7:  The material and methods section should include further information about the microbiological methods used.

Ad comment Nr. 7: Yes. We agree. There was nothing special with the microbiological cultures, which were all processed according to standard CLSI criteria (page 2, lines 59-61).

 

Comment Nr. 8:  The conclusion section should be rephrased, including statistically based information.

Ad comment Nr. 8: We modified the conclusion section according to this comment as follows (page 9, lines 283-287): Gram-negative SSIs after spine surgery seem to be firmly associated with UC carriage beyond one day (OR 5.5, 95%CI 1.5-20.3) and with the female sex (OR 3.8, 95%CI 1.4-10.6), according to the performed multivariate analysis. A weaker association might also exist between SSIs (any Gram) and a prolonged UC carriage (OR 1.1, 95% confidence interval 1.1-1.1). In other words, a UC carriage might not only be a risk for nosocomial UTI, but might also be associated with early SSIs after spine surgery. This observation indicates careful indication for UCs in the perioperative period around spine surgeries[32,33].

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I appreciate the clarifications to the previously mentioned comments, I understand that this is a retrospective study and only information is available from the patients' clinical history, which does not allow to evaluate whether there is a correlation between the microorganisms present in the surgical wound and those isolated of the urinary catheter.

Author Response

Dear Reviewer 2,

Thank you for your time and your comment.

Comment Nr. 1 (2^nd round):

I appreciate the clarifications to the previously mentioned comments, I understand that this is a retrospective study and only information is available from the patients' clinical history, which does not allow to evaluate whether there is a correlation between the microorganisms present in the surgical wound and those isolated of the urinary catheter.

Ad comment Nr. 1:

We agree with Reviewer 2 that we cannot prove with certainty that those microorganisms found in the surgical wounds are genetically the same as those found in the urinary catheters. However, there was a clear congruence of the urinary pathogens with those of SSI, based on bacterial species and the corresponding antibiotic susceptibility patterns, which is an accepted method of comparison in daily clinical life. In the revised manuscript (major revision), it is already mentioned that the microbiological congruences between UTI and SSI pathogens formally base on the identification of the bacterial species and their antibiotic susceptibility patterns, which is an accepted method of comparison in daily clinical life; and that serotyping cannot be retrospectively performed (page 9, lines 273-277). No additional modifications were done according to this comment.

Reviewer 4 Report

Comments and Suggestions for Authors

I appreciate the author's efforts in addressing my comments.

My only remark is that in the conclusion section, the statistical numbers should not be included. Also, even though rounding p values makes them easier to read, in my opinion, the complete and accurate numbers would have been a better choice.

 

Author Response

Dear Reviewer 4,

Thank you for your time and your comments.

Comment Nr. 1 (2^nd round):

My only remark is that in the conclusion section, the statistical numbers should not be included.

Ad comment Nr. 1:

Thank you for this comment. We agree with it and modified the revised manuscript version according to it as follows:

Gram-negative SSIs after spine surgery seem to be firmly associated with UC carriage beyond one day and with the female sex, according to the performed multivariate analysis. A weaker association might also exist between SSIs (any Gram) and a prolonged UC carriage.

 

Comment Nr. 2 (2^nd round):

Also, even though rounding p values makes them easier to read, in my opinion, the complete and accurate numbers would have been a better choice.

Ad comment Nr. 2:

Unless, if you have to round up equally for four decimal points. Initially, we would have preferred to keep the p values rounded, for two reasons.

First, there are many of them. Giving a value, let’s say to the fourth decimal point, does not give much more information, as the values still remain estimates. For example, if only one SSI episode would be different, then the values would be also different. The p value 0.0067 would switch to 0.0054. However, in rounded form, both would still remain 0.01. It is the significance which is interesting (yes and no at an arbitrary level of 0.05), and not the degree in four decimal places (which is a pseudo-precision due to the number of cases analysed). For another analysis of (tens of) thousands of episodes, or a mathematical manuscript, things might be different. Second, p values at two decimal points are much easier to read and limit spaces in the manuscript.

However, upon the explicit demand of Reviewer 4, we rounded up the p-values shown in the Tables 1 and 2 to four decimal points in the revised manuscript. Hence, for highly significant p-values, the p of 0.01 now becomes 0.0001.