Re-thinking antimicrobial resistance transmission dynamics: a meta-analysis of cross-sectional studies at referral hospitals in Uganda [version 1; peer review: 2 approved with reservations]

Background: Antimicrobial resistance threatens the achievements of modern medicine as well as the sustainability of effective global public health responses to the threat posed by infectious diseases. Extended-spectrum β-lactamase production in bacteria provides the main mechanism of resistance in gram-negative bacteria, particularly those belonging to the Enterobacteriaceae family as well as grampositive bacteria. This study hence aimed at providing insights into the potential role of in-patients, their immediate hospital environments, out-patients, and their communities in the transmission of antimicrobial resistance via identifying gram-negative and grampositive bacteria commonly isolated in samples collected from each of these patients/sites as well as their antimicrobial susceptibility profiles using extended-spectrum β-lactamase production in the same as the basis. Methods: Our study reviewed four cross-sectional studies conducted at national and regional referral hospitals in Uganda. Data on bacterial aetiology and antimicrobial susceptibility testing retrieved from the studies was imported into Microsoft Excel, cleaned, sand then exported to IBM SPSS statistics (version 16) for statistical analysis. The databases used were PubMed and Embase. Results: We report that; Escherichia coli and Klebsiella pneumoniae are the most prevalent Enterobacteriaceae species in the samples that were collected in the studies reviewed; these species account for the highest proportions of extended-spectrum β-lactamase producers; Staphylococcus aureus is the most prevalent of the gram-positive Open Peer Review


Introduction
Antimicrobial resistance not only threatens the achievements of modern medicine but also the sustainability of effective global public health responses to the threat posed by infectious diseases 1,2 . Extended-spectrum β-lactamase production in bacteria confers antimicrobial resistance and provides the main mechanism of resistance in gram-negative bacteria, particularly those belonging to the Enterobacteriaceae family as well as grampositive bacteria 3,4 . Extended-spectrum β-lactamase producing bacteria exhibit resistance to all penicillin-like antibiotics except cephamycins and carbapenems; resistance in these bacteria frequently spans to other classes of antibiotics, including the quinolones and aminoglycosides 5-7 . Resistance to β-lactam antibiotics is mostly plasmid-borne and can be disseminated among bacterial species via horizontal gene transfer; this has consequently led to clonal distribution within and between clinical settings, communities as well as across local and international boarders via patient mobility 7,8 . Combating antimicrobial resistance requires continuous and timely monitoring, surveillance, and assessment systems as well as mechanisms to detect extended-spectrum β-lactamase producing bacteria. In Uganda, the acquisition and transmission dynamics of extended-spectrum β-lactamase producing bacteria at referral hospitals and community settings have not been elucidated. The aim of this study was therefore to provide insights into the potential role of in-patients, their immediate hospital environments, out-patients, and their communities in the transmission of extended-spectrum β-lactamase producing bacteria via identifying gram-negative and gram-positive bacteria commonly isolated in samples collected from each of these patients/sites as well as their antimicrobial susceptibility profiles using extended-spectrum β-lactamase production in the same as the basis.

Methods
We conducted a systematic review of cross-studies on bacterial aetiology and antimicrobial susceptibility profiles conducted at national and regional referral hospitals in Uganda.

Eligibility criteria
The application of the inclusion and exclusion criteria was undertaken independently by the reviewers (GM, IS, and DA) and any difference in opinion was resolved by discussion among the reviewers. Also, the inclusion and exclusion criteria were the same for both the systematic review as well as the meta-analysis.
Included studies were as follows: • Studies whose designs were cross-sectional.
• Studies that were conducted in Ugandan national and regional referral hospitals.
• Studies that reported bacterial aetiology and antimicrobial susceptibility profiles.
• Studies with the requirements above that had been published within the last ten years.
Excluded studies were as follows: • Studies that had been published as reviews • Studies in which data had been reported only as abstracts.
• Studies whose designs were not cross-sectional (i.e. retrospective and prospective) Search strategy. The Pubmed and Embase electronic databases were searched for eligible studies. We used the search string ["Antimicrobial resistance"] AND ["Referral Hospitals"] AND ["Uganda"]. We also sought additional studies through screening the references of the selected studies.

Data extraction
Data extraction was done by GM, IS and DA. Data were entered in prespecified spreadsheet in Microsoft Excel (2019 version). The extracted data elements consisted of (1) bacterial aetiology and (2) antimicrobial susceptibility profiles.

Risk of bias.
For the bias risk analysis, an appraisal tool for cross-sectional studies, AXIS was used 9 . In addition, quality assessment of the included studies referred to a modified method composed of the design of the studies, participant, sites and settings selection methodology and laboratory tests that included; determination of bacterial aetiology, antimicrobial susceptibility testing and screening for the production of Extended-spectrum β-lactamases (extended data 10 ).

Statistical analysis.
The main outcomes of interest in the data analysis were: 1) the proportion of gram-positive and negative bacteria isolated from the samples collected in the studies; and 2) the proportion of the gram-positive and negative bacteria isolated that were producers of Extended-spectrum β-lactamases.
For all the selected studies, data on bacterial aetiology and antimicrobial susceptibility testing was extracted using Microsoft Excel (2019 version). Following data extraction, several tables containing dichotomous data were generated for the relative and cumulative calculation of the frequencies of the outcomes of interest. For all of the meta-analysis procedures, IBM SPSS Statistics (version 16) was used. In SPSS, forest and funnel plots and charts were generated. The proportions of either the gram-positive and negative bacteria was determined by the number of the same divided by the total number bacteria isolated in the studies. A similar approach was used to determine the proportions of gram-positive and negative bacteria isolated that were producers of Extended-spectrum β-lactamases. The I 2 test was used to access the existence of heterogenicity of between the studies (I 2 = 75-100%, p<0.05) 11,12 . Due to the nature of the studies, heterogenicity was expected; we therefore chose to use the random effects model for the meta-analysis as proposed by DerSimonian-Laird 13 . We also performed a sensitivity analysis to test for the effect of the individual influence of each study on the overall estimate, and a subgroup analysis was also performed to reduce the existence of heterogenicity. In addition, we evaluated the existence of publication bias by visual inspection of Begg's funnel plots as well as by Egger's test calculation 14,15 .

Results
Primarily, during the search for studies in the digital databases as well as additional records from other sources, 52 reference studies were found. After the application of the inclusion and exclusion criteria, we refined the results, and 4 eligible studies constituted the present meta-analysis [16][17][18][19] . The flowchart of this selection is shown in the PRISMA diagram is shown in Figure 1.
Of the 4 studies included in the meta-analysis, 2/4 (i.  (i.e. tables, sink taps and light switches); the samples had been collected by purposive sampling from the different sites, consideration was given to sites with constant hand contact, and the samples were collected from June to December 2015 19 .
Bacterial aetiology and antimicrobial susceptibility profiles of gram-negative isolates Studies 1, 2, 3, and 4 reported Enterobacteriaceae making up 47.7%, 49%, 46.9%, and 25.6% out of the total isolates obtained from these respective studies (Table 1). Escherichia coli was found to be the most predominant of the Enterobacteriaceae isolates obtained from studies 1, 2, and 3 accounting for 49.7%, 43.1%, and 53.9% respectively; second to Escherichia coli was Klebsiella pneumoniae, this accounted for 12.8%, 29.4%, and 28.7% of the Enterobacteriaceae isolates in studies 1, 2, and 3 respectively. In a similar regard, Klebsiella pneumoniae was found to be the most predominant of the Enterobacteriaceae isolates in study 4 representing 46.7%; second to Klebsiella pneumoniae was Proteus vulgaris, this represented 33.3% of the Enterobacteriaceae isolates (Table 1).
Bacterial aetiology and antimicrobial susceptibility profiles of gram-positive isolates Staphylococcus aureus was the most predominant grampositive isolate obtained from studies 1, 2 and 4 accounting for 67.4%, 93.2% and 75.4% respectively ( Table 2). Study 3 did not isolate any gram-positive bacteria as it only considered gramnegative bacteria ( Table 2). Proportions of extended-spectrum β-lactamase producing Staphylococcus aureus were 67.4%, 75% and 70.5% for studies 1, 2 and 4 respectively. Study 3 did not isolate any gram-positive bacteria and therefore no extendedspectrum β-lactamase production was reported ( Table 2).

Discussion
This study reviewed four cross-sectional studies on bacterial aetiology and antimicrobial susceptibility profiles conducted at referral hospitals in Uganda with the aim of providing insights into the potential role of in-patients, their immediate hospital environments, out-patients, and their communities in the transmission of antimicrobial resistance via identifying gram-negative and gram-positive bacteria commonly isolated in samples collected from each of these patients/sites as well as their antimicrobial susceptibility profiles using extended-spectrum βlactamase production in the same as the basis.
This study achieved the objectives for which it was conducted; the major findings were that; (i) Escherichia coli and Klebsiella pneumoniae are the most prevalent Enterobacteriaceae species in the samples that had been collected in the studies reviewed; (ii) these species (i.e. Escherichia coli and Klebsiella pneumoniae) account for the highest proportions of extended-spectrum β-lactamase producers; (iii) Staphylococcus aureus is the most prevalent of the gram-positive bacteria isolated from the same samples, and accounts for the highest proportions of extended-spectrum β-lactamase producers in the gram-positive bacteria isolated, and (iv) similar Enterobacteriaceae species and gram-positive bacteria are predominant in samples from inpatients, their immediate hospital environments, and out-patients.
The findings of this study with regards to Escherichia coli and Klebsiella pneumoniae being the most prevalent Enterobacteriaceae species in the samples are consistent with those of similar studies; these studies have not only highlighted these bacterial species to be the most prevalent in samples from in-patients 20-24 , but have also indicated them to be the most prevalent in samples from out-patients as well in those from the hospital environment 7,25-27 . The findings of this study with regards to the isolation of similar Enterobacteriaceae species from in-patient samples, those from their immediate hospital environments, and out-patients could be suggestive of the possibility of complex transmission dynamics of the same.
In a similar light, despite the fact that these remain the most prevalent Enterobacteriaceae species, this study as well the studies reviewed did not ascertain the source of these; however, speculate that these could be from both nosocomial and community sources.
In addition, the findings of this study with regards to these Enterobacteriaceae species being the main extended-spectrum β-lactamase producers of the gram-negative isolates are consistent with those of similar studies 20,22 .
The prevalence of these species in samples collected from both the community and hospital settings continues to increase. The source of ESBL-PE in samples from the community could be as a consequence of the introduction by individuals that had visited health-care settings or had interacted with health-care workers in the same settings, this has also been suggested in other studies 28,29 . Some studies have however suggested that population exposures to antibiotics could offer an explanation to the observation 25,30 , while some have demonstrated carriage of the same bacteria in individuals with no history of having taken antibiotics 31 and have highlighted food products from animals as potential sources of the bacteria 32 .
According to the 2017 World Health Organization list of priority pathogens, these Enterobacteriaceae species have been identified to belong to the priority one or critical category, and particularly when these are producers of extended-spectrum β-lactamases. Extended-spectrum β-lactamases confer resistance to all penicillin derivative antibiotics as well as cephalosporins, except cephamycins and carbapenems 7,33 ; despite the fact that this study and the studies reviewed did not ascertain which extended-spectrum β-lactamase encoding genes were mediating the observed β-lactam resistance and how these are potentially disseminated in the isolates, our study speculates that the extended-spectrum β-lactamase encoding genes could have been either or a combination of the bla CTX-M , bla SHV and bla TEM , and that horizontal gene transfer could be a plausible mechanism by which the dissemination of these genes could be occurring. This is because, not only have most extended-spectrum β-lactamase encoding genes described so far been derivatives of the blaCTX-M, blaSHV and blaTEM, but these have also been reported to be the most prevalent in isolates just like those in the studies reviewed; these genes have also been documented to be mostly encoded by conjugative plasmids; these are mobile genetic elements (MGEs) 7,8,33,34 .
Besides extended-spectrum β-lactamase encoding genes, these MGEs have also been documented to carry genes that confer resistance to other antibiotic classes that include fluoroquinolones, aminoglycosides, and trimethoprim/sulfamethoxazole; these have also been documented to contribute to the selection and persistence of multidrug-resistant extended-spectrum β-lactamase producing strains in samples from in-patients, their immediate hospital environments, and out-patients; this could explain the resistance to other antibiotics observed in the studies reviewed 5,7,8,35,36 .
The findings of our study with regards to Staphylococcus aureus being the most prevalent of the gram-positive bacteria isolated from the samples collected in the studies, and accounting for the highest proportions of extended-spectrum β-lactamase producers in the gram-positive bacteria species from in-patient samples, those from their immediate hospital environments, and out-patients are consistent with those of similar studies; these studies have not only highlighted these bacterial species to be the most prevalent in samples from in-patients, but have also indicated them to be the most prevalent in samples from out-patients, as well in those from the hospital environments 37-39 . In a similar light, despite the fact that these remain the most prevalent gram-positive bacteria species, our study as well as the studies reviewed did not ascertain the source of these, however, similar to the case of the Enterobacteriaceae species, this study speculates that these could be from both nosocomial and community sources.
According to the 2017 World Health Organization list of priority pathogens, Staphylococcus aureus, has been identified to belong to the priority two or high category, and particularly when these are producers of extended-spectrum β-lactamases or which exhibit methicillin-resistance (MRSA). Despite the fact that MRSA has long been a pathogen of nosocomial origin, in recent times, it has emerged as a problematic pathogen in community settings 40,41 . These same studies have also indicated the possibility of community-associated MRSA being replaced by nosocomial-associated MRSA.

Conclusion
The results of our study indicate that similar extendedspectrum β-lactamase producing Enterobacteriaceae species and gram-positive bacteria are predominant in samples from in-patients, their immediate hospital environments, and out-patients. This study offers insights into the potential role of in-patients, their immediate hospital environments, out-patients, and their communities in the transmission of antimicrobial resistance. In addition, the results of our study, underscores the need for continuous monitoring and surveillance of antimicrobial resistance in clinical and community settings, together with efforts to promote the rational and judicious use of antibiotics. Lastly, with the insights provided here, antimicrobial resistance transmission dynamics should be re-thought and more comprehensive studies aimed at investigating the same should be done to ascertain the source and transmission routes of antimicrobial resistant bacteria in community and clinical settings.

Data availability
Underlying data All data underlying the results are available as part of the article and no additional source data are required.

and non-producers of extended spectrum beta-lactamases in urinary isolates
implication for the results.

Minors
Discussion: Please add the references for the sentences: (4.1) "According to the 2017 World Health Organization list of priority pathogens, these Enterobacteriaceae species have been identified to belong to the priority one or critical category, and particularly when these are producers of extended-spectrum βlactamases".
Reviewer Report 20 October 2020 https://doi.org/10.5256/f1000research.27177.r71664 © 2020 Haraka F. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

This manuscript has been reviewed in consideration of the PRISMA guideline and checklist
Mboowa G et al., aimed at investigating anti-microbial resistance transmission dynamics in a hospital setting by pulling existing evidence through systematic review and meta-analysis. Mboowa G et al., searched two databases and restricted their search to Uganda. They found that Escherichia coli and Klebsiella pneumoniae were the most prevalent Enterobacteriaceae species and these species account for the highest proportions of extended-spectrum β-lactamase producers among gram-negative bacteria.
Staphylococcus aureus was the most prevalent. The authors' topic is relevant for both patient and policy decisions, however, the authors will need to revise the manuscript before it can be accepted. Specifically;

Major comments:
The authors will need to redo and expand their search of relevant articles, indicate the time bound of the search and all relevant databases. The description of the search strategy must be detailed to include all components of the PICO question they intend to address.