Genetic Determinants of Resistance among ESBL-Producing Enterobacteriaceae in Community and Hospital Settings in East, Central, and Southern Africa: A Systematic Review and Meta-Analysis of Prevalence

Background The world prevalence of community and hospital-acquired extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae is increasing tremendously. Bacteria harboring ESBLs are currently the number one critical pathogens posing a major threat to human health. Objective To provide a summary of molecular evidence on the prevalence of ESBL-producing Enterobacteriaceae (ESBL-E) and associated genes at community and hospital settings in East, Central, and Southern African countries. Methods We conducted a systematic literature search on PubMed and Google Scholar databases for the available molecular studies on ESBL-E in hospitals and community settings in East, Central, and Sothern Africa (ECSA). Published studies in English language involving gene characterization of ESBLs from human samples in hospital and community settings were included in the review, inception to November 2019. A random effect meta-analysis was performed to estimate the prevalence of ESBL-E. Results A total of 27 studies involving molecular characterization of resistance genes from 20,225 ESBL-E isolates were included in the analysis. Seventy-four percent of all studies were hospital based, 15% were based in community settings, and others were done in both hospital and community settings. Of all the studies, 63% reported E. coli as the dominant isolate among ESBL-E recovered from clinical samples and Klebsiella pneumoniae was reported dominant isolates in 33% of all studies. A random pooled prevalence of ESBL-E was 38% (95% CI = 24–53%), highest in Congo, 92% (95% CI = 90–94%), and lowest in Zimbabwe, 14% (95% CI = 9–20%). Prevalence was higher in hospital settings 41% (95% CI = 23–58%) compared to community settings 34% (95% CI = 8–60%). ESBL genes detected from clinical isolates with ESBL-E phenotypes in ECSA were those of Ambler molecular class A [1] that belongs to both functional groups 2be and 2d of Bush and Jacob classification of 2010 [2]. Majority of studies (n = 22, 81.5%) reported predominance of blaCTX-M gene among isolates, particularly CTX-M-15. Predictors of ESBL-E included increased age, hospital admissions, previous use of antibiotics, and paramedical use of herbs. Conclusion Few studies have been conducted at a molecular level to understand the genetic basis of increased resistance among members of ESBL-E in ECSA. Limited molecular studies in the ECSA region leave a gap in estimating the burden and risk posed by the carriage of ESBL genes in these countries. We found a high prevalence of ESBL-E most carrying CTX-M enzyme in ECSA with a greater variation between countries. This could be an important call for combined (regional or global) efforts to combat the problem of antimicrobial resistance (AMR) in the region. Antibiotic use and hospital admission increased the carriage of ESBL-E, while poor people contributed little to the increase of AMR due to lack of access and failure to meet the cost of healthcare compared to high income individuals.


Introduction
Pathogenic bacteria evolve to resist the actions of antimicrobials through acquired and intrinsic mechanisms including production of β-lactamase enzymes, which inactivates antibiotic and decreases its therapeutic value [1,2]. Extended-spectrum β-lactamases produced by many gram-negative bacteria, mostly Enterobacteriaceae, are able to hydrolyze penicillins, cephalosporin, and monobactams . ey are mostly effective against a range of β-lactam drugs including ceftazidime, ceftriaxone, cefotaxime, and aztreonam [3][4][5]. In many cases, resistance to these antibiotics is transferred among bacteria through gene transfer systems of mobile genetic elements carried in bacteria plasmids or transposons by bacterial recombination process that involves conjugation, transformation, and transduction. e world prevalence of community and hospital-acquired ESBL-E is increasing tremendously. Bacteria harboring ESBL enzymes are currently the number one critical pathogens posing a major threat to human health [6]. e spread and dissemination of infections caused by ESBL-E are associated with increased morbidity and mortality, health care costs, the need for development of new wide-spectrum antimicrobials and lengthy hospital stay of infected patients.
is is because of a major decrease in therapeutic value of mostly used drugs as a result of resistance [7][8][9][10][11].
ere has been a significant advancement in the understanding of ESBL-producing bacteria epidemic which was previously related to hospital-acquired infections [12,13]. Recent increased recovery of ESBL-E from community and environmental samples [14][15][16], especially E. coli commonly causing community acquired urinary tract infections (UTIs) [17], indicates a probability of the shift of importation of ESBL-producing bacteria to hospitals rather than vice versa. e spread of community acquired ESBL carrying pathogens is accelerated by between-persons transmission of ESBL bacteria in the communities. Some studies suggest there are significantly higher transmission rates of ESBL-producing bacteria among community households as compared to hospital transmissions [18]. However, detailed studies describing the ESBL-E reservoirs and transmission routes in diverse settings are still limited.
In most poor resource countries of East, Central, and Southern Africa, there is lack of routine surveillance systems that could estimate the magnitude and risk factors as well as clinical outcomes associated with ESBL Enterobacteriaceae [19]. e overuse and misuse of antimicrobial agents in the environmental sector, agriculture, and human and veterinary medicine propel the spread of antimicrobial resistance among infectious bacteria. Other factors such as easy access to antibiotics, weak health systems, environmental contaminations, poor hygiene and sanitation services or practices, incomplete decontamination of medical devices, and lack of laboratory capacities for pathogens detection and surveillance have been described as important drives of the increasing resistance among members of ESBL Enterobacteriaceae in the region [20][21][22].

Objective.
To better understand genetic determinants of resistance among ESBL-producing Enterobacteriaceae in East, Central, and Sothern Africa, we sought to summarize molecular evidence on the prevalence of ESBL-E and associated genes at community and hospital settings.

Material and Methods
is review and meta-analysis of prevalence has been conducted in compliance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and checklist [23].

e Study Area.
e review provides information on published articles from East, Central, and Southern Africa (ECSA) countries in accordance with Africa Union Countries profile [24]. We included six countries from the eastern Africa which are Tanzania

Study Inclusion and Exclusion Criteria.
All available published molecular studies (involving genotypic characterization of ESBL) in English language on human subject reporting ESBL-E in hospitals and communities in ECSA were considered eligible. Unpublished studies, editorials, letters, studies on nonhuman subjects, studies published in other languages than English, and studies that did not utilize molecular tools were excluded. All studies available on the selected databases at the time of data extraction were examined (inception to November 2019).

Data Extraction.
Data extraction checklist was developed to guide the acquisition of the information which included name of the author(s), year of publication, study setting (community or hospital), study design, subjects/ target population, source of isolates/specimen, clinical samples, sample size, bacteriological methods for estimating ESBL's, molecular methods used, isolate species, number of isolates analysed, ESBL positive isolates, genes encoding for ESBL identified and risk factors associated with ESBL infection if studied.

Data
Analysis. Acquired data were entered into Excel spreadsheet and statistical analysis was done using Stata version 12 (STATA Corporation, College Station, TX, USA). We performed a random effect meta-analysis to determine heterogeneity of ESBL-E prevalence in ECSA. Decision to perform random effect meta-analysis over fixed effect metaanalysis was made due to an assumption that the difference in ESBL-E prevalence among studies in ECSA is attributed to different factors such as study settings (hospital and community) and different laboratory methods used for detection of ESBL-E. A new program in STATA (Metanprop) specific for pooling binomial data including methods of computation of the 95% confidence intervals (CI), continuity correct, and the Freeman-Tukey transformation was used [25]. e risk of publication bias was assessed using funnel plot and Begg's rank correlation test for funnel plot asymmetry [26].

Factors Associated with ESBL-E Infections in ECSA.
Five of 27 articles reviewed stressed out factors associated with ESBL-E infections to include increased age, hospital admissions, and previous use of antibiotic. A study in Gabon found that hospitalization and at least 5 years of age were the risks for the carriage of ESBL Enterobacteriaceae in children [43]. Higher ESBL-E carriage was strongly associated with treatment of HIV infection in Zimbabwean children [34]. In an area where antibiotics could not be afforded due to poverty and civil war, ESBL-E carriage in children was associated with high income families [52]. Increased number of children in the household, high median age, previous use of antibiotics, and use of local herbs for paramedical purposes were also predictors of ESBL-E carriage [36][37][38][39][40].

Discussion
In the present review, we present the summary of molecular determinants and the pooled prevalence of ESBL-producing Enterobacteriaceae in Eastern, Central, and Southern Africa. Few studies have been conducted to determine ESBL genes in these countries. Tanzania did most of the genotypic characterization of ESBLs while in other ECSA countries there was no any molecular work on AMR due to ESBL production by members of Enterobacteriaceae. e overall pooled prevalence was high (38%), close to previous findings by Sonda et al., [54] but higher than that found by Tansarli et al. [19] and Bulabula et al. [55].
According to the data generated from the present review, there was low pooled prevalence of ESBL-E in Zimbabwe and higher in DR Congo, Tanzania, and Malawi. Comparing our study with previous findings, it is surprising to find the prevalence of ESBL-producing pathogens in Malawi has increased by 89% within 12 years since the introduction of third-generation cephalosporins. e rate of resistance due to ESBL production at the time of cephalosporins introduction in Malawi in 2005 was very low (0.07%); the reason for low prevalence at that time was tight restriction on the use of antibiotics [45]. However, the rate rose rapidly to 92% in 2017 indicating either misuse of third-cephalosporin drugs or loosened restrictions on the antibiotic use [51]. e prevalence of ESBL-E was higher in the hospital settings than the community settings; this was similar to finding of others [19,56]. We noted that majority of isolates were obtained from cultures of patients with nosocomial infections leading to the higher rate of ESBL genes detection in hospital settings. Similarly, it has been described elsewhere that most nosocomial infections are associated with ESBL-producing bacteria [9,22,37,43,49,57]. Our findings complement previous review by Tansarli et al. [19]. ese findings can suggest possible influx of ESBL pathogens into the communities from the hospitals. However, bringing together various disciplines to collectively investigate the burden of ESBL pathogens in the environment, human, and animals interface could provide information on the contribution of these different reservoirs to the ESBL-E epidemic.
Higher prevalence of ESBL-E in ECSA according to this review could be attributed to some factors including the focus of the studies reviewed and resistant genes targeted. Many studies across these countries reporting ESBL-E from human either were focused on certain ESBL bacteria, mainly E. coli and K. pneumoniae [22,42,49], or targeted the detection of specific ESBL genes [32,35,38,40,41]. erefore, there were high chances of escalating the prevalence of ESBL producers without inclusion of other β-lactamases in the studies. However, there could also be inadequate infection control in healthcare systems contributing to the increased spread of ESBL-E in these poor resource countries as previously suggested [58]. Selective pressure due to heavy use of β-lactam antibiotics as a first-  line treatment of infections caused by Enterobacteriaceae has been described to increase the spread of resistance and hence could lead to high prevalence in ECSA [32,42]. Increased carriage of ESBL-E has also been associated with the use of local herbs such as Aloe vera and other herbal extracts used for paramedical purposes [35]. In most instances, poverty has been linked with increased incidence of ESBL-E. However, our study suggests that people who are poor contribute little to the increased spread of resistance among member of ESBL-E compared to those who can afford to buy and use antibiotics. Since majority of the studies were from hospital settings, it could be that the poor are not well represented due to either lack of access or failure to meet the cost of healthcare. e fact that few molecular studies have been conducted in the ECSA region undermines the understanding of ESBL   [35,38,40,41,[60][61][62][63]. Previous studies have suggested that commensal isolates originating from infection of the gut and intestines carried different ESBL genes depending on bacteria species, plasmids, and location of the chromosome bearing the resistant gene [33]. However, in this study we observed invariability of ESBL gene types between community and hospital isolates across the studies. e importance of genotypic detection of ESBL genes in bacteria over phenotypic screening cannot be ignored; this is because these genes are carried on plasmids permitting higher chances of horizontal transfer which cannot be detected phenotypically. Phenotypic screening for ESBL pathogens is useful for surveillance purposes but they can underestimate the burden of resistance posed by ESBL pathogens. Previous reports detected more resistant ESBL producers when molecular methods were used compared to phenotypic methods [27]. For definitive therapy selection, relying on phenotypic determination of ESBL enzymes alone may not be sufficient. erefore, molecular detection is inevitable.

Strengths and Limitations.
One of the strengths of this review is that we included studies published in English which is the official language of many countries in the study area. Searching Google Scholar database provided an advantage to retrieve articles that could have not been published in highly indexed journals. Institutional subscription also allowed retrieval of articles that were not available for 6 Canadian Journal of Infectious Diseases and Medical Microbiology free access. However, the study was limited by the challenge encountered in determining true prevalence because some studies were laboratory surveillance based and did not provide the actual sample size or number of participants from which the clinical samples were drawn.

Conclusion and Recommendations.
Few studies have been conducted at a molecular level to understand the genetic basis of increased resistance among members of ESBL-E in ECSA. Limited molecular studies in the ECSA region leave a gap in estimating the burden and risk posed by the carriage of ESBL genes in these countries. We found a high prevalence of ESBL-E most carrying CTX-M enzyme in ECSA with a greater variation between countries. is could be an important call for combined (regional or global) efforts to combat the problem of AMR in the region. Antibiotic use and hospital admission increased the carriage of ESBL-E, while poor people contributed little to the increase of AMR due to lack of access and failure to meet the cost of healthcare compared to high income individuals.

Conflicts of Interest
e authors declare no conflicts of interest.