PREreview of Antimicrobial Sensitivity Testing Using the Kirby-Bauer Disk Diffusion Method; Limited Utility in Ugandan Hospitals

This full PREreview is the result of a live-streamed preprint journal club organized and hosted by AfricArXiv, Eider Africa, eLife, PREreview, and Training Centre in Communication Africa as part of the Open Peer Reviewers in Africa workshop training.

The authors have done a good job in conducting such a crucial study at a time when many developing countries are facing challenges in the medical field and with limited access to technologies and capacity to address the challenges. This was a study from hospital records carried out to determine the utility of antimicrobial sensitivity testing (AST) by Kirby-Bauer disk diffusion method in 3 major Ugandan hospitals. The study employed a mixed methodology (retrospective; November-December 2013 and Cross Sectional; January-February 2014). The researchers used the abstraction checklist to collect data on AST requests, AST performed, AST turnaround time, dispensed antibiotics from in patient files, laboratory and pharmacy records. The primary aim of this study was to investigate the possibility of using AST to identify the extent of antimicrobial resistance (AMR) in Uganda. The study had two designs (i) Retrospective design for the period from November to December 2013 and (ii) Cross-sectional design for the period between 27th January to 14th February- 2014. The main findings were that there were very low numbers of documented AST requests (2.1%), a high analytical turnaround time of 5 days (IQR 4-8 days), and discrepancy between AST requests and dispensed antibiotics and that the correlation between antibiotic consumption and AST was weak. The authors therefore concluded an underutilization of AST, is inconsistent with the hospital antibiotic consumption, and may be related to the use of the disk diffusion method. They therefore recommended the use of alternative, faster methods (no suggestions given) and better coordination in the procurement of AST diagnostics in Ugandan hospitals.

List of major issues and feedback

The reviewers believe that addressing the following “major” issues would significantly strengthen the conclusion of the study:

• The study design or methodology appears confusing or even contradictory. To carry out this study, authors mentioned that they used both cross-sectional design and retrospective collection of data. The reviewers are wondering how they used both in the same study and why? If there is a rationale for the use of these two approaches, we encourage authors to explain it in the methods and results sections. Clarity would be added by elaborating further on the specific analytical approach undertaken for each one. For example, authors may consider adding details on inclusion and exclusion criteria, the calculation/estimation of sample size and the rationale behind it.
• Clearly stating the main objective(s) of the study and describing what aspects/scope of utility of AST Facilities were considered in the assessment, would significantly help the reader assess the soundness of the results and conclusions.
• While we understand that data collection in the current pandemic situation may be more challenging than normal, we think the sample size is low (n=607). Therefore, the number of hospitals involved may not geographically represent the entire country or practices in other hospitals and may therefore affect the generalizability of the results. A sample size calculation would have been better to explain the selection of patients’ files to be reviewed. This could be based on a formula adopted by Cochran 1963 to determine the sample size.

n = N/ [1 + N (e) 2] Where; n – sample size N – Population size e – Level of significance

• We recommend the authors discuss the limitations of low sample size in the discussion,including possible over-interpretation and generalizability of the study
• Is there a reason/rationale behind having selected a 1-month time window for data collection in the retrospective study, November-December 2013? If so, it would be helpful to include that in the methods section.
• Three Ugandan hospitals were chosen as representatives of the whole country. .Overall, the number of hospitals involved may not geographically represent the entire country or practices in other hospitals and may therefore affect the generalizability of the results. The reviewers believe this limitation should be discussed alongside the rationale for choosing those three hospitals.
• The study didn’t specify clear inclusion and exclusion criteria. The authors could make this clearer by providing which criteria were used to either make certain hospitals/patients part of this study and some not. They could write sentences like any hospitals/patients that (fulfilled certain conditions) were included, while those that failed to meet the conditions were excluded. Examples of inclusion: complete medical records available, use of AST and no other method.
• Are depersonalized demographic data and clinical history data available?
• Authors didn’t include a tool (a supplementary or in the main text) used for abstraction of data from patients’ records. We suggest the authors include (as an appendix) the data collection tool (s) to facilitate a better understanding and interpretation of the results presented.
• This study rationale is good but these major issues need to be clarified: The study design or methodology should be rewritten for clarity, while also indicating the study design and methods used for the comparative country (Ethiopia) and justify why the researcher used both cross-sectional design and retrospective collection of data.

List of minor issues and feedback

The issues observed in this preprint are:

• Appropriate references weren‘t included. For example, reference 7 by Gutema et al, discusses the use of antibiotics, its prevalence, consumption but does not mention AST.
• Grammatical errors. The reviewers suggest that the authors use free grammar checking and correcting tools like Grammarly https://app.grammarly.com/ or Power Thesaurus https://www.powerthesaurus.org/ to correct the grammatical errors. This will enhance the readability and coherency of the work.
• The study design and methods used for the comparative country (Ethiopia) wasn‘t stated in the preprint. The authors give some little information on how a similar study was carried out in Ethiopia and how it relates to the current study. More information is required in the manuscript so it is easier for the readers of this work to understand the rationale for the comparison.
• The authors could provide additional information in some sections e.g. in the discussion section it says “the study showed very low utilization of AST...” The authors had attempted to directly compare their findings with other similar studies, Ethiopia then USA. The authors could include more information on the reason behind the low utilization and this will strengthen the manuscript. The author also defined Uganda as an under-resourced country, they could provide additional details what being an under-resourced country means.
• In figure 1 the authors should also give a summary of what the figure shows, what does each dot represent? and clearly label all axes e.g. X- antimicrobial drugs dispensed and what is the Y-axis? They should also differentiate between percentages versus proportions.
• In the conclusion section, the results do not align to the study objective, the authors should align the results and the objectives and draw conclusions based solely on the results of their study.
• To further improve the manuscript, the authors could organize the manuscript in a logical format clearly outlining the different sections including the limitations.
• Please highlight the current state of AMR in Uganda in general to give a basis and background for the current study.
• Please include other tests for detecting bacteria and explain the choice of AST given that results are available after 48-72 hours after sampling as bacteria need to be cultured first before AST can be performed.
• Please provide the clarification between disk diffusion and antibiotics disc should be explained.

We thank the authors for sharing their work as a preprint. We hope our feedback above will be helpful as they consider any revisions to the manuscript or future lines of work.