Review Reports - Benchmarking: A Tool for Veterinary Practices to Improve Prudent Use of Antibiotics in Cats and Dogs in Switzerland

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear authors,

thank you for submitting this interesting and well-written paper. I have only a few comments in the following:

l. 95: register
acceptable: complicated term, not neutral; I would search for an alternative such as non conspicuous
Figure 1: Why use the log? The distribution changes, but the 75% and 95% percentiles are the same, and easier to communicate
I wonder if Figure 2b is really necessary. It is very similar to Figure 2a, which is due to the nature of the relationship. If it is not important to elaborate on the differences, I would refrain from Figure 2.
Table 1: What are the numbers in brackets?
Why did you choose 4,500 consultations as a breakpoint? And why did you make a classification in the first place?
Table 1: Green columns: Do you really mean animals or consultations? What is the percentage of treatments? Is it the percentage of consultations with treatments?
ll. 129/130: sometimes twice as high
ll. 132-134: This sentence would be a good sentence at the beginning of the results part.
ll. 145-146: Please check the numbers. If 17/35 were very high users, only 15/35 can be high users.
l. 149: None of the 32(?) practices?
l. 149-150: Am I wrong, or is there again a mistake in the numbers? If 21 out of 32 were very high users, how can 14 of 32 then be high users?
In Table 1 results per animal species and type of animal for the year 2024 for the 152 variables presented in the BM report. --> This sentence seems to be incomplete for me. Please check.
l. 155 etc: Please report p-values for statistical tests.
The subtitle and the paragraph ll. 151-158 don't fit together in my understanding. A general comparison of animal species does not belong to "Individual summary of AMU at practice level"
l. 162: require? You don't know that. I would say: "receive"
l. 289: witnessed
ll. 385 ff: Please provide more information on the quantile regression: package, regression parameters, level of significance etc.
Supplementary Material A: Maybe you should remove the equine part and translate it into English.

Author Response

thank you for submitting this interesting and well-written paper. I have only a few comments in the following:

l. 95: register

We changed the formulation, and the grammar of the sentence is now correct. Thank you for spotting this mistake. (now L114)

acceptable: complicated term, not neutral; I would search for an alternative such as non-conspicuous

Many thanks for your comment. We discussed vividly at our office how these categories should be named and settled with the ones described in the paper. We cannot change what has already been sent to veterinary practices in the previous years so we would advocate to use the same terminology in the paper. However, you raise a very valid point, and we will discuss it internally to try to agree to a less complicated term for the next years in the reports for veterinarians.

Figure 1: Why use the log? The distribution changes, but the 75% and 95% percentiles are the same, and easier to communicate

- Many thanks for your comment. I provided below the first version of figure 1 (never published) without the log transformation. While creating the first report, we consulted a data scientist specialised in data visualisation to improve this graph. It is meant to summarize the main information for the veterinary practices. Therefore, we quickly abandoned this first version because the categories red (very high users) and orange (high users) take almost 75% of the graph while they only represent 25% of the practices. We feared that this would mislead the readers by thinking that we classified a very big part of the AB prescribers as high or very high users. As the main information of this figure is the ranking and categorisation of the reader’s practice, we considered that a log transformation would provide a clearer guide for the reader.

I wonder if Figure 2b is really necessary. It is very similar to Figure 2a, which is due to the nature of the relationship. If it is not important to elaborate on the differences, I would refrain from Figure 2.

Many thanks for your comment. We agree that we underexploit these graphs and therefore do not need 2 figures. We deleted figure 2b with the number of animal treatments and focused on the number of practices. The manuscript is updated accordingly.

Table 1: What are the numbers in brackets?

Many thanks for your comment. The numbers in brackets are the values for critical antibiotics only. This is specified in the top left hand corner of the table, and we added a sentence in the legend to help the reader.

Why did you choose 4,500 consultations as a breakpoint? And why did you make a classification in the first place?

Guidelines for benchmarking recommend comparing units (e.g. practices, farms) that are comparable (same profile). Indeed, tertiary clinics face different challenges and receive peculiar patients, i.e. referral services, specialisation, surgery, more advanced cases, that would require more frequently and second- or first-line AB. Their pATI would then be higher in comparison to other practices due however to their normal activity. Tertiary clinics would be in the highest ranks, and “punished” for their normal activity. In discussion with the profession, veterinarians heavily advocate to separate practices according to their profile.
The Swiss veterinary profession does not have a clear definition to differentiate practices from clinics, supporting a categorisation based on other criteria. Initially, we defined the profile of practices/clinics based on expert opinion (self-declaration and peer recognition). However, the total number of clinics in Switzerland was too low to create a separate category; mentioned in chapter 4.1.3. Finally, we used the number of consultations to identify the biggest practices. All Swiss clinics are above the set thresholds, and we assumed that practices of the same size might face the same challenges and might also have the same stewardship capacities.
We provided some more information about the definition of clinics and practices in chapter 4.1.3. We also discussed this point in the discussion (paragraph from line 223). These 2 sections should now summarize all the arguments to explain our methodology.

Table 1: Green columns: Do you really mean animals or consultations? What is the percentage of treatments? Is it the percentage of consultations with treatments?

Many thanks for your comment. The green columns present indeed the number of treatment of animals with antibiotics, that we called “animal treatments with antibiotics”. We added a footnote in the table to increase readability und understanding from the readers.
The percentage of treatment is indeed the percentage of consultations that resulted in an antibiotic treatment/prescription. This value is presented in the methodology (L395-399) and newly as a footnote from table 1.

ll. 129/130: sometimes twice as high

This mistake was corrected directly in the document. (now L157)

ll. 132-134: This sentence would be a good sentence at the beginning of the results part.

Many thanks for your comment. We move this sentence to the beginning of the results in chapter 2.1. (Lines 110-113)

ll. 145-146: Please check the numbers. If 17/35 were very high users, only 15/35 can be high users.

Many thanks for spotting this error. We proofed and corrected the values. The number of practices per categories does not exceed the total number of practices anymore. (lines 172-174)

l. 149: None of the 32(?) practices?

Indeed, all practices identified as high and very high users in 2024 for dog prescriptions prescribed at least one critical AM in the same year. The 32 practices are detailed in Table A.1, 10 mixed practices. 19 practices <4500 consultations/year and 3 practices <4500 consultations/year (Appendix A).

l. 149-150: Am I wrong, or is there again a mistake in the numbers? If 21 out of 32 were very high users, how can 14 of 32 then be high users?

That was indeed the same mistake as above. We corrected the values. (lines 177-178)

In Table 1 results per animal species and type of animal for the year 2024 for the 152 variables presented in the BM report. --> This sentence seems to be incomplete for me. Please check.

Many thanks for your comment. We however deleted the sentence, following the comment from another reviewer.

l. 155 etc: Please report p-values for statistical tests.

We provided p-values directly into the text instead of the statement “statistically significant”. More details are still available in Table B.1 in Appendix B as referred in the text. (example line 154)

The subtitle and the paragraph ll. 151-158 don't fit together in my understanding. A general comparison of animal species does not belong to "Individual summary of AMU at practice level"

Many thanks for your comment. Indeed, the subtitle was not matching the content of the paragraph. We changed the title with “Differences of AB treatment percentages between species and practice types” (line 179)

l. 162: require? You don't know that. I would say: "receive"

Many thanks for identifying this poor choice of wording. We replaced it with the verb “involve”. (L190)

l. 289: witnessed

Many thanks for your comment. We completely updated the sentence and changed it into: “However, such an impact can be better assessed after several years” (L258-259)

ll. 385 ff: Please provide more information on the quantile regression: package, regression parameters, level of significance etc.

Many thanks for your comment. We updated the section and provided all the requested details. I hope this now clear the programming details. (chapter 4.2.3, lines 375-386)

Supplementary Material A: Maybe you should remove the equine part and translate it into English.

Many thanks for your comment. The official translation of the document is only available in German, French and Italian. Our resources won’t allow us to provide an official translation in English, but we translated the document with the help of DeepL.This document is now available as Supplementary material.
We provided a new version of the Supplementary material without the equine section.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

GENERAL COMMENTS

The authors provide a detailed summary of the Information System for AMU in Veterinary Medicine or IS ABV. Veterinary practitioners are required by Swiss law, as set out by the authors, to provide high quality information on prescribing practices each year. The objective of the Swiss approach to antimicrobial stewardship (AMS) is to collect data on AMU as a benchmarking tool, particularly focused on the antimicrobial treatment indicator at practice level, abbreviated pATI. Certain thresholds of AMU are set, and individual veterinary practices receive a national report summarising all received data with the results of the recipient practice highlighted. This is the essence of benchmarking, allowing each practice to see where they are situated amongst their peers.

The manuscript is very informative and provides an excellent template for other countries to review and consider when planning their own approach to AMS.

The inclusion of an example of the report that is generated as a Supplement potentially adds a lot of value. The version included is in German, while Italian and French versions are also generated. An English version would be a great asset as it would allow a greater number of readers to see the value of the report.

I have made a number of edits and added a number of comments in a separately attached document.

My main recommendation to the authors is to consider the missing dimension of AMS.

Quantity of use (as calculated by pATI) is an important initial assessment of AMS. However, the ultimate target of AMS is to ensure that each prescription and use of antimicrobials is appropriate. Low use can nevertheless be inappropriate use. The manuscript refers tin various places to ‘responsible use’, ‘improved compliance with guidelines; ‘increased prudent use’, and ‘quality of AM prescriptions’.

It would be a valuable addition to the discussion to include current thinking on adding some assessment of ‘quality to use’ or appropriate use. For example, the indication for each AMU is amongst the data collected. As indication and selected antimicrobial agent are known, this would enable assessment against guidelines to calculate one important indicator of appropriateness (ie that selected antimicrobial is consistent with current guidelines). There are of course other important indicators of appropriateness. However, the Swiss system collects key data and could be first country in the world to assessment not just AMU on a national scale, but also quality of use

SPECIFIC COMMENTS

See attachment

Comments for author File: Comments.pdf

Comments on the Quality of English Language

Suggested corrections to English expression are included in the attachment.

Generally the English is fluent, just a few areas for reconsideration.

Author Response

The inclusion of an example of the report that is generated as a Supplement potentially adds a lot of value. The version included is in German, while Italian and French versions are also generated. An English version would be a great asset as it would allow a greater number of readers to see the value of the report.
- Many thanks for your comment. The official translation of the document is only available in German, French and Italian. Our resources won’t allow us to provide an official translation in English, but we translated the document with the help of DeepL.This document is now available as Supplementary material.
I have made a number of edits and added a number of comments in a separately attached document.
- Many thanks for all your corrections and comments in the manuscript. Almost all your grammatical suggestions were directly integrated into the document. Comments are answered separately below.
My main recommendation to the authors is to consider the missing dimension of AMS. Quantity of use (as calculated by pATI) is an important initial assessment of AMS. However, the ultimate target of AMS is to ensure that each prescription and use of antimicrobials is appropriate. Low use can nevertheless be inappropriate use. The manuscript refers tin various places to ‘responsible use’, ‘improved compliance with guidelines; ‘increased prudent use’, and ‘quality of AM prescriptions’. It would be a valuable addition to the discussion to include current thinking on adding some assessment of ‘quality to use’ or appropriate use. For example, the indication for each AMU is amongst the data collected. As indication and selected antimicrobial agent are known, this would enable assessment against guidelines to calculate one important indicator of appropriateness (ie that selected antimicrobial is consistent with current guidelines). There are of course other important indicators of appropriateness. However, the Swiss system collects key data and could be first country in the world to assessment not just AMU on a national scale, but also quality of use
- We completely agree with your point of view and regret the lack of details and clarity in our first manuscript. We added several sentences in the discussion (L265-270) to highlight how we intent to move forward. Indeed, we already started further analyses regarding compliance to therapy guidelines and its evolution over time. These analyses however are more time consuming, and we expect to publish results in the next years.

SPECIFIC COMMENTS

See attachment

L13 Abstract: ”Antimicrobial” is usually defined as including agents effective against viruses, fungi, protozoa and bacteria - ie all microbial types. As the current paper is restricted to ANTIBACTERIAL USE, I recommend that as early as possible in the MS that antimicrobial be defined as equivalent to antibacterial in this paper.
- Many thanks for this comment. We adapted the whole manuscript and only use antibiotics.
L43: There are many publications dated well before those referenced (3-7). I therefore recommend sentence be reworded, for example: “has become a topic OF PARTIUCLAR IMPORTANCE in AMR studies recently [3-7].”
- Many thanks for your comment. We agree with your statement and changed the text accordingly. (line 43)
L47: Recommend adding: “in either direction” - ie there are many examples of AMR gene flow from humans to animals as well as from animals to humans.
- Many thanks for your comment. We agree with your statement and changed the text accordingly. (line 48)
L51: support OR guide?
- We modified the verb used. (line 52)
Figures 2a and 2b. It is not surprising but seems that the distribution of practices and treatments by Swiss canon is identical in figures 2a and 2b, just the units are different. 2 suggestions: 1. Could both figures been combined with both subjects (practices and treatments retained; 2. Could the top 4 or 5 cantons be identified on the map
- Many thanks for your comment. We agree that the graphs might look redundant; especially because we underexploit the information contained in both. We deleted figure 2b with the number of animal treatments and focused on the number of practices.
Figure3: Violin plots (if that is what these are) are an information rich way of presenting data and very appropriate in this MS. As such plots are not widely appreciated in the veterinary and other readership categories, I recommend that the text include a brief description of the various data elements included in each plot - this could be in the text or as a separate diagram as another appendix.
- Thank you very much for your comment. We took the opportunity to include an explanation of violin plots in the text itself and provide a few additional explanations (lines 157-161). We hope this helps readers to understand all the information contained in Figure 3.
L129: This should be TWICE AS HIGH
- Many thanks. We corrected the sentence. (now L157)
L175: Not clear what ‘punctually’ means or implies. Alternatives to consider: ‘over limited periods’, ‘for single BM studies’
- Thanks for identify this unclarity. The sentence was however deleted after a comment from another reviewer.
L286: “Seems” is very subjective. On what basis was meaningfulness assessed?
- Many thanks for your comment. We tried to improve the clarity of the sentence by providing more insight into our thoughts. (lines 255-256)
L310: Indication + treatment used together allow some early assessment of appropriateness or quality of use. Of cause the indication also implies that the diagnosis is correct, another subject for investigation.
- Many thanks for your comment. We added several sentences in the discussion (L265-270) to highlight how we intent to move forward. Indeed, we already started further analyses regarding compliance to therapy guidelines and its evolution over time.
L323, ref 55: WHO now have published the MIA list (the 7th revision). Perhaps some mention of this could also be included - especially as the classification of the CIA is still relevant.
- Many thanks for your comment. We updated reference 55 accordingly (now reference 53, line 300).
L336: Is there a difference between a practice and a clinic? This is no universal definition.
- Terminology might change between countries but there is a differentiation between general (primary) practice, emergency point-of-care (secondary) and referral and specialised (tertiary) clinic. This is no official definition, but the veterinary sector still differentiates them. We added a sentence in the chapter 4.1.3 (lines 332-335)
L376: Can the description ‘acceptable’ be applied based only on the quantity of use? Acceptable generally requires the use to be appropriate as well and this aspect of AMU is not required for this term to be applied.
- Many thanks for your comment. We discussed vividly at our office how these categories should be named and settled with the ones described in the paper. We cannot change what has already been sent to veterinary practices in the previous years so we would advocate to use the same terminology in the paper. However, you raise a very valid point, and we will discuss it internally to try to agree to a less complicated term for the next years in the reports for veterinarians.
Ref 46: If redacted, should this document be a reference?
- This manuscript has been submitted to Preventive Veterinary Medicine recently. We deleted the reference as it is not yet published or under final revision.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for preparing this well-written and informative article. The main purpose of benchmarking may be considered its impact on encouraging altered prescribing behaviour among those practices reported to have very high pATI. It will be great to be able to gauge the impact of benchmarking on AMU in the future.

Could you provide some of the additional assumptions used to develop the pATI term? Did you assign a typical duration for each antibiotic (i.e. cefovecin counts as 14 days or 1 day duration?)

Some minor comments:

The axis labels for Figure 1 are very small. Could these be increased to improve legibility? It may be worth offering an interpretation of one of the sample figures so that the reader can appreciate what the different forms mean? This has taken me some time to understand.

Line 129: consider twice as high rather than twice higher

Line 145-146: I do not understand how 53.1% could be both very high and high users? Should the second figure be 46.9%?

Throughout: is it better to put the associated p value rather than (statistically significant)?

Lines 198-212: could you provide a little more detail as to the calculations for generating the pATI? Does a single dose of a fluoroquinolone count the same as one, two or three doses of amoxiclav? What if a double dose (10mg/kg enrofloxacin) is used? Is there any allowance for differences in dose or frequency of administration?

Line 216: DDD = defined daily dose not dosis

Lines 231-234: could it also be that mixed veterinarians have poorer stewardship?

Line 277: suggest change to: we received emailed feedback from 13 practitioners.

Author Response

Could you provide some of the additional assumptions used to develop the pATI term? Did you assign a typical duration for each antibiotic (i.e. cefovecin counts as 14 days or 1 day duration?)
- Many thanks for your comment. We provided in Appendix D explanations about the calculation of therapy days. In table D.1. we detailed all the values of active substance effective time that we used for the calculation per active substance. For example, the value for cefovecin counts as 14 days for cats and 10 days for dogs; it is also mentioned in the text L232-233.

Some minor comments

The axis labels for Figure 1 are very small. Could these be increased to improve legibility? It may be worth offering an interpretation of one of the sample figures so that the reader can appreciate what the different forms mean? This has taken me some time to understand.
- We increased the labels of the figure to increase legibility. We also worked on the legibility of other Figures and provided updated versions.
- An interpretation of Figure 1 was provided directly into the legend of the graph to ease the understanding of the reader. (lines 121-128)
Line 129: consider twice as high rather than twice higher
- Many thanks, we corrected the text accordingly. (now L157)
Line 145-146: I do not understand how 53.1% could be both very high and high users? Should the second figure be 46.9%?
- Many thanks for your attentive reading. The figures were not correct. We proofed the values and corrected them accordingly. (lines 172-174 and 177-178)
Throughout: is it better to put the associated p value rather than (statistically significant)?

Many thanks for your comment. We included the p-values directly into the manuscript instead of the statement “statistically significant” (example line 154). More details are still available in Table B.1 in Appendix B as referred in the text.

Lines 198-212: could you provide a little more detail as to the calculations for generating the pATI? Does a single dose of a fluoroquinolone count the same as one, two or three doses of amoxiclav? What if a double dose (10mg/kg enrofloxacin) is used? Is there any allowance for differences in dose or frequency of administration?
- Many thanks for your comment. We provided in a new Appendix D details about the calculation of the therapy days. We hope that would answer any interest from readers about this value.
- Our indicator, the pATI, counts only the number of treatment days and the duration of the active substance (AST), i.e. therapy days. In your example a single treatment of 1 day of fluoroquinolone would count for 1 therapy day (1 day of treatment -1 + 1 day of active substance efficiency time = 1); when a single treatment of 1 day of amoxicillin would count for 2 therapy days (1 TtmtD -1 + 2 AST = 2). If a veterinary product with 2 active substances (example of amoxiclav) is prescribed, this calculation is applied to both active substance and the total of therapy days for the prescription sum up both individual therapy days.
- Regarding allowance for differences in doses or frequency of administration, they are not considered in the indicator. If a dosage is higher as the recommendations, it cannot be assessed via our indicator. In our next research we aim to highlight the compliance of veterinarians to the therapy guidelines in terms of antibiotic choice and dosage but this research is only starting. We also plan to compare different indicators (e.g. DDD and our pATI) to identify high users; but this will be done in a second phase.
Line 216: DDD = defined daily dose not dosis
- Many thanks for your comment. We corrected all wrong mentions.
Lines 231-234: could it also be that mixed veterinarians have poorer stewardship?
- Many thanks for your comment. This point was mentioned in lines 225, and we improved the clarity of the sentence by mentioning clearly the differences of resources for stewardship.
Line 277: suggest change to: we received emailed feedback from 13 practitioners.
- Many thanks for your comment. We adapted the text accordingly. (lines 246-247)

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

Line 133: is it 70 or 72 practices vs table 1

Line 145: proportion of high users and very high users is confusing. Requires clarification

Line 152: what is the relevance of the opening statement?

Line 170-176: move to the conclusion

Line 176: recast the initial statement of "for example........"

Line 176-194: should form part of the introduction.

Line 199-226: this is not relevant to the discussion. Kindly move it to the methodology

Line 256-271: this is all methodology. Please move

Why exclude pATI for consultation fewer than a 100?

Author Response

Line 133: is it 70 or 72 practices vs table 1
- Many thanks for your comment and attentive reading. Some practices are classified differently according to the species: above and below 4500 consultations per year for cats and dogs. To provide a global overview (not per species), we had to classify them in only one category. That is the reason of the difference.
Line 145: proportion of high users and very high users is confusing. Requires clarification
- Many thanks for your attentive reading. The figures were not correct. We proofed the values and corrected them accordingly (lines 172-174 and 177-178).
Line 152: what is the relevance of the opening statement?
- Many thanks for your comment. We deleted this opening statement of the paragraph.
Line 170-176: move to the conclusion
- Many thanks for your comment. We added a section “conclusion” and move this section. (lines 415-421)
Line 176: recast the initial statement of "for example........" + Line 176-194: should form part of the introduction.
- Many thanks for your comment. We move this section in the introduction and could rephrase the sentence at the same time. (from line 63)
Line 199-226: this is not relevant to the discussion. Kindly move it to the methodology
- Many thanks for your comment. We indeed removed this section from the discussion as, with a new eye, it was more an argumentation. We however did not move the paragraphs in the methodology as we explained already in details our choices and calculations. Instead, we summarized the main point of discussion and provided new references. The new section is now available lines 200-207.
Line 256-271: this is all methodology. Please move
- Many thanks for your comment. This paragraph was distributed between sections 4.1.1 IS ABV and 4.1.2 Number of consultations in the methodology.
Why exclude pATI for consultation fewer than a 100?
- We excluded very small practices (less than 2 consultations per week) because we assumed that this activity (consultation of companion animals) might be conducted only in peculiar situations (e.g. vaccination of a dog when visiting the farm, treatment of emergency cases only). We did not intent to include practices that have no interest or habits in treating companion animals and prefer excluding their influence in the BM system (lines 317-319).