Title: Antibiotic susceptibility of Clostridium difficile is unchanged despite widespread use of broad spectrum antibiotics
Version: 1
Date: 13 May 2014
1. General: Terminology has been changed from CDAD to Clostridium difficile infection (CDI).
Reply: We amended it throughout the manuscript.
2. Line 68: Introduction: "CDADs" = Colloquial - suggest changing to CDAD cases (or CDI cases).
Reply: we amended it.
3. Line 69: As above
Reply: we changed it correspondingly.
4. Table 2: It would also be more informative to illustrate how many patients received combination of antibiotics, as well as DDD data
Reply: We have added the required information about combination antibiotic therapy as Suppl. Table
2.
Indeed we have used the DDD of our cohort to calculate the Odds ratios for CDI by comparing the intake of the various antibiotics examined in our study with the defined daily intake of the antibiotics prescribed in the University Hospital of Zurich during the study period (Fig. 1). In addition we provide the numbers of DDD in the discussion:
In our cohort, most cases of CDI occurred in patients treated with penicillins, followed by patients treated with cephalosporins, quinolones and carbapenems. However, we must take into account the rank order of prescribing the various antibiotics. Indeed, the defined daily dose/100 bed days was the highest for pencillins, followed by cephalosporins, quinolones and carbapenems (defined daily dose/100 bed days (DDI/100bd) in 2006, 2007, and 2008 for
1

penicillins were 23.22, 25.77, and 26.10, for cephalosporins were 9.36, 10.45 and 9.67, for quinolones 8.78, 9.82, 9.9, for carbapenems 3.66, 4.45 and 4.08, respectively).
5. Line 187 - some evidence that antifungals may predispose to CDI. Worth commenting on, and even informing reader what antifungals were used?
Reply: We have added this information as footnote to Suppl. Table 1. We are not aware that antifungals predispose to CDI – we even checked the PubMed with the key words: (clostridium difficile OR CDAD OR CDI) and fung*: no papers came up hinting to the possibility that antifungals may predispose to CDI.
6. Line 212 – “In the 1980s….” - Can this be referenced?
Reply: We added the corresponding references.
7. Line 269 – remove “lucky”
Reply: We did so.
1. Line 78: introduction: Colloquial - "Strong" microbial activity? Classification of antibiotic activity into strong and weak is too simplistic. Recommend altering this.
Reply: We entirely see the point Dr. Kanna is making: considering the complexity of the pathogenesis of CDI, the strength of the interaction between the antibiotic and Clostridium difficile is most likely not the key element. Furthermore, we will discuss the MICs of the various antibiotics in detail later in the manuscript. Thus, we changed the wording to:
Line: 86-90: Various antibiotics have an antimicrobial activity against C. difficile. However, it is largely unknown whether the inherent activity of the antibiotics against C. difficile has any protective effect on the development of CDI.
2. Line 166: Do you have lab records to prove that stool samples were diarrhea when tested for C.diff toxin? Testing of non-diarrhoeal samples will confound results.
Reply: We have focused our analyses on patients with diarrhea, and stated it explicitly. However, as it is the case for retrospective studies based on chart reviews, we had to deal with insufficient chart documentation in a minority of cases. We stated this in the manuscript:
Diarrhea was explicitly documented in the charts in 81.9% of the cases. Based on the directives of the hospital, there is no C. difficile testing without diarrhea. Thus, most likely all patients included in this study suffered from diarrhea even when documentation was missing.
3. Table 2: spelling mistakes in Suppl Table 2 – eg ceftazidime/metronidazole/glycopeptide
Reply: We highly appreciate the sharp eye of Dr. Khanna, and corrected the spelling mistakes. We also checked the other submitted documents for spelling mistakes and amended them if present.
4. Suppl table 2: Anti-mykotic should be antifungal
Reply: We corrected anti-mykotic to antifungal in Suppl. Table 2 as well as throughout the main manuscript – as we combined Suppl. Table 1 and 2, the corrections will be found now in Table 1.
5. Line 187 – Antifungal instead of anti-mycotic
Reply: see above.
2

6. Line 196 – metronidazol spelling
Reply: thanks for indicating us.
7. Line 199 – “Strains were in>90%....” – Does not make sense
Reply: We thank the reviewer for pointing to this unclarity; we changed the wording as follows:
Over 90% of all strains were susceptible to amoxicillin/clavulanate. In contrast more than
90% of the strains were intermediate or resistant to clindamycin, ceftriaxone and ciprofloxacin
8. Line 202 – “where” should be “were”
Reply: thanks for indicating us.
9. Line 233 – spelling of cephalosporins
Reply: We corrected the spelling mistake.
10. Line 238 The patients suffering from CDADs – colloquial
Reply: We changed the wording to:
The patients with CDI….
11. Line 248 – remove additional words “reviewed in”
Reply: We did so.
12. Line 258 -The OR 95% CI for Vanc and Septrin crossed the 1 point suggesting that the effect may not be stat significant?
Reply: We entirely agree with Dr. Khanna and changed the wording:
Glycopeptides and sulfonamides are most likely also a risk factor for CDI, however the OR for these two antibiotic classes was no statistically significant.
13. Line 265 - Not all penicillins have activity versus C. difficile? What about Pen V/Amoxicillin?
Reply: We changed the wording here (see below) to define exactly the antibiotics prescribed to the patients and which were the objects of this work. No patient was prescribed Pen V and thus, we feel it would not be appropriate to discuss Pen V as related to the question we ask here. Furthermore, C. difficile has no beta-lactamase activity – thus, activity of amoxicillin will not differ from the activity of amoxicillin/clavulanate:
Line 279-286: It remains open whether the penicillins prescribed, i.e., amoxicillin, amoxicillin/clavulanate and piperacillin/tazobactam, due to their inherent activity against C.
difficile and aminoglycosides and macrolides due to their lack of activity against anaerobes, are less likely associated with the development of CDI. Our data are mostly congruent with a recently published meta-analysis which reported the strongest association for cephalosporins and clindamycin with CDI and a lesser one for carbapenems, trimethoprim/sulphonamides, fluoroquinolones and penicillin combinations [29]. The differing results we observed for clindamycin, penicillins and for fluoroquinolones are surprising.
1. Table 2: It would have been better to list certain individual antibiotics as opposed to classes as eg:
Augmentin is more likely to pre-dispose to CDI than Amox. Also antibiotics in preceding 3 months have been shown to be able to pre-dispose patients to CDI, even a single prophylactic agent - Do you
3

have antibiotic history for each patient that goes back 3 months before index toxin positive? If not, this needs to be acknowledged.
Reply: We entirely agree with Dr. Khanna that certain antibiotics even within one class may have a different risk of pre-disposition to CDI. Risk of pre-disposure of antibiotics of the same class, however, will more closely group together than antibiotics of different classes. For simplicity reasons, we initially decided to present the antimicrobial use grouping the individual antibiotics into classes.
Since in Table 3, in any case, we go into details concerning activity of individual antibiotics to C. difficile, Dr. Khanna is right and we present now a more detailed antimicrobial use according to the distinct antibiotics.
We indeed analyzed all antibiotics used back to eight weeks preceding the CDI. The time frame eight weeks was chosen on the work by Baxter R et al. [Case-conrol study of antibiotic use and subsequent
Clostridium-difficile-associated Diarrhea in hospitalized patients, Infection Control and Hospital
Epidemiolgy, January 2008, Vol 29, No 1]
We included this information in the “Material and Methods” part:
Line 117-119: Based on the fact that antibiotics may precede the development of CDI for up to eight weeks {Baxter et al., 2008}, we recorded all antibiotics for this time frame prior to the collection of the stool specimen.
2. Table 2: It would also be more informative to illustrate how many patients received antibiotic monotherapy, combination of antibiotics, as well as DDD data.
Reply: please see comment 4 of discretionary revisions.
3. Line 189 – As above, It would be useful to the reader to give more information what penicillins were used as each have different propensity to induce CDI.
Reply: We have added the requested information as footnote to table 2.
4. Materials and Methods: Any exclusion criteria? Are these index cases only with no patients with recurrent disease? Although this may not have an impact on MICs, it is important to state whether these are recurrent CDI cases or not.
Reply: We did not have formal exclusion criteria, but we did not include any patient with recurrent disease. We have added this information in the “material and method” part:
Line 107: There were no repeated stool specimens from patients with recurrent CDI
5. Line 171-173 and materials and methods: In these 17 patients, culture positive, cytotoxin neg, but
PCR positive - PCR detects presence of gene and not direct toxin production, so are you inferring that presence of toxin gene in patient with diarrhoea and c.difficile positive culture has CDI? As far as I'm aware Cytotoxin assay is still gold standard, so a negative Cytotoxin assay is concerning. Would it have been better to do toxin PCR first, followed by cytotoxin assay to confirm? Can you suggest reasons as to why the cytotoxin assay was negative?
Reply: There are different possibilities to define a gold standard for the detection of
Clostridium difficile toxin. One reference standard is the cytotoxicity assay. The advantage is, that the assay proves biological active toxin. However, sampling errors may occur and the assay can be false negative. In addition long transportation might have an impact on the stability of the toxin, or inhibitory substances might be present and result in a negative cytotoxicity assay. Another reason for a negative cytotoxicity assay is the lack of toxin gene expression of a strain harbouring the toxin genes. In addition the sensitivity of culture might be higher for some samples, as in theory one bacterium is sufficient to be detected positive
4

by culture. Therefore, another gold standard definition regularly used in studies is the toxinogenic culture, meaning a culture isolate which harbours toxin genes or is positive in the cytotoxicity assay. A third possibility is a combination of the two gold standards described above. In a meta-analysis performed by O’Horo et al (1) the sensitivity of molecular detection methods were found to be sensitive and specific. Again, studies included in this meta-analysis used the three different variants of “gold standard” to define C. difficile toxin positive samples. In daily routine laboratory practice, normally a stepwise approach is applied to guarantee for optimal sensitivity and specificity. In our study we decided to use the approach describe to achieve optimal sensitivity for the detection of C. difficile toxin positive samples.
Mayo Clin Proc.
2012 Jul;87(7):643-51. doi: 10.1016/j.mayocp.2012.02.024.Molecular techniques for diagnosis of Clostridium difficile infection: systematic review and metaanalysis. O'Horo JC 1 , Jones A , Sternke M , Harper C , Safdar N .
6. There should be a Limitations paragraph as part of the discussion.
Reply: We entirely agree with this legitimate request by Dr. Kanna. We think that the limitations of the study are the number of cases with CDI and the retrospective nature of the study.
We have already addressed the number of specimens in the discussion:
Lines 278-286: We are aware that the number of cases is rather limited for such an analysis, and thus, its interpretation must be taken cautiously. It remains open whether the penicillins prescribed, i.e., amoxicillin, amoxicillin/clavulanate and piperacillin/tazobactam, due to their inherent activity against C. difficile and aminoglycosides and macrolides due to their lack of activity against anaerobes, are less likely associated with the development of CDI. Our data are mostly congruent with a recently published meta-analysis which reported the strongest association for cephalosporins and clindamycin with CDI and a lesser one for carbapenems, trimethoprim/sulphonamides, fluoroquinolones and penicillin combinations [30]. The differing results we observed for clindamycin, penicillins and for fluoroquinolones are surprising.
Furthermore, we added a separate paragraph at the end of the discussion:
Lines 331-337: Limitations of the study are the number of cases and the retrospective nature of this work. The number of cases we analyzed is in the range of cases presented in published work (see Suppl. Table 3); nonetheless, we acknowledge that interpretations must be done cautiously and put in a bigger context in concert with published data. While we made a major effort to get hands on a homogenous cohort, the changes in microbiological assays over the time may be regarded as a confounder for diagnosing CDI. Irrespective of this confounder, we are pretty convinced that the retrospective identification of the cases is correct and made us believe in the data and conclusions presented.
7. Line 313-314 - The fact that cephalosporins and carbapenems are high risk for causing CDI is not new - This has been reviewed in other papers - most recently: Slimings and Riley, J Antimicrob
Chemother doi:10.1093/jac/dkt477, Dec 2013 - This paper provides a systematic review andmetaanalysis of antibiotics as risk factors for CDI
Reply: We have added the conclusion of this recent meta-analysis at the end of the paragraph discussing the association of antibiotics with CDI.
Line numbers 271-275: Our data are mostly congruent with a recently published metaanalysis which reported the strongest association for cephalosporins and clindamycin with
CDI and a lesser one for carbapenems, trimethoprim/sulphonamides, fluoroquinolones and
5

penicillin combinations [ 29 ]. The differing results we observed for clindamycin, penicillins and for fluoroquinolones are surprising.
Level of interest: An article whose findings are important to those with closely related research interests
Quality of written English: Needs some language corrections before being published
Statistical review: Yes, but I do not feel adequately qualified to assess the statistics.
Declaration of competing interests: I have received funding for research and fees for public speaking from AstraZeneca
Reviewer's report
Title: Antibiotic susceptibility of Clostridium difficile is unchanged despite widespread use of broad spectrum antibiotics
Version: 1
Date: 16 June 2014
Reviewer: Dongsheng Zhou
This is a totally descriptive study. The strain collection is limited since they came from a single location and from an old time range. The data presented are not sufficient to make the conclusions which the authors advance. There are numerous sentences in the text that the authors need to clarify and/or rephrase.
Reply:
It is indeed a descriptive study – surveys and analyses over time are by virtue of descriptive nature.
The aim of the study was to assess the local pattern and to compare it to the existing literature, thus we do not really catch why the referee made the point of the single location. We agree that there is some delay between the collection of the specimens and the submission of this manuscript – we, nonetheless, believe that the extensive analyses of specimens we did add to the knowledge of CDI.
We are aware that the number of specimens is limited but in the range most other studies published; thus, we formulated our conclusions cautiously and added a separate paragraph concerning limitations of the study to the discussion (see above).
Dr. Khanna did a very detailed review also concerning language and spelling mistakes. Since Dr. Zhou was not very elaborate in his review, we hope the corrections made incited by Dr. Khanna also cover the criticism brought up by Dr. Zhou.
Level of interest: An article of limited interest
Quality of written English: Needs some language corrections before being published
Statistical review: No, the manuscript does not need to be seen by a statistician.
Declaration of competing interests: I declare that I have no competing interests
6

Title: Antibiotic susceptibility of Clostridium difficile is unchanged despite widespread use of broad spectrum antibiotics.
Version: 1
Date: 21 June 2014
Reviewer: Elvira Garza-Gonzalez
Authors cannot assume that antibiotic susceptibility of Clostridium difficile is unchanged despite widespread use of broad spectrum antibiotics it they did not analyzed the antibiotic susceptibility in the same population.
Reply: We have not examined antimicrobial susceptibility in the same population. But indeed, we made an exhaustive comparison between all existing studies examining antibiotic susceptibility of C.
difficile since 1980 (Suppl. Table 3) published from various epidemiological setting and our data. In addition we compared our local MIC we got to the data provided by EUCAST (Fig. 2), usually including
MIC data from a large variety of strains. Therefore, we think, it is reasonable to draw the conclusion that the antimicrobial susceptibility has not changed.
• Review orthography and grammar. Example: some of the names of antibiotics are not well written.
Reply: Dr. Khanna also made this point and we went throughout the entire manuscript correcting the orthography and grammar. In addition, an American native professional editor checked the manuscript again for English.
• Most likely, the underlying diseases or 50 complications, in concert with the carbapenems’ activity on the GI-tract flora, rendered these 51 patients more susceptible to CDAD. This is acceptable as a conclusion. This is a hypothesis that is acceptable in the Discussion section.
Reply: We removed the last sentence in the abstract.
• The objective is unclear: Here, we investigated what antibiotics were primarily associated with CDI,
(treatment with some antibiotics?) the antimicrobial susceptibility pattern of C. difficile strains and whether some antibiotics with inherent antimicrobial (against C. difficile?) activity protects against
CDI.
Reply: We rephrased the objective and made in congruent with the aims described at the end of the introduction.
• The aims of abstract and the end of introduction are not the same. Please clarify them.
Reply: see comment above.
• The phrase “Notably, we have lived through a time in which antibiotics were increasingly used, and that usage may have resulted in unexpected collateral damage” it is completely out of place.
Reply: We have removed this sentence.
• In the introduction section there is more information needed about the resistance patterns of C. difficile isolates in the world and though the years.
7

Reply: We decided to provide an exhaustive review of all published studies covering this issue since
1980 (Suppl. Table 3). We discussed the resistance pattern in detail in the discussion. We would prefer not to give more information about the resistance of C. difficile in the introduction to avoid unnecessary repetition in the manuscript. We hope Dr. Garcia-Gonzalez will accept our point of view.
• Material and Methods, Laboratory analyses, including antimicrobial susceptibility testing: delete
Laboratory analyses.
We have split the laboratory part of materials and methods into three major parts and added new headlines to improve the structure of the materials and methods part. We deleted the headline
“laboratory analysis”. However, we consider detailed description of laboratory procedures very important for data interpretation. In particular, when we screened the literature for studies examining resistance pattern of C. difficile, we encountered a number of studies where the methodology was rather scarcely described which rendered the study results difficult to interpret.
• They describe in material and method section that The MIC results were categorized in susceptible, intermediate and resistant, following the guidelines of EUCAST and if not available with the values of
CLSI were used. Table 3 shows the results using the criteria of CLSI and of EUCAST when there is not value by CLSI.
Reply: We are very gratefully for this valuable comment. Of course, we interpreted the MIC data according to the CLSI criteria, and in case no breakpoints were available, we used the EUCAST breakpoints. We have changed the section in the materials and methods section.
• Supplementary table 2. These are not risk factors or the authors are not proven that these are risk factors. There is not a statistical analysis to prove it. These are only percentages. The 26.6 % of patients had been in ICU? This is not a risk factor if is not compared to patients without infection by
C. difficile.
Reply: We highly appreciate this comment and we agree entirely. Thus, we combined Suppl. Table 1 and 2 and changed the wording of the new Table:
Patient characteristics prior to first positive culture or toxin of Clostridium difficile.
Level of interest: An article whose findings are important to those with closely related research interests
Quality of written English: Needs some language corrections before being published
Statistical review: Yes, but I do not feel adequately qualified to assess the statistics.
Declaration of competing interests:
8