Introduction

Antibiotic resistance (ABR) is currently identified as one of the biggest threats to not only global health but also to food security and development1. Resistance occurs when the antibiotics (medicines used to prevent and treat bacterial infections) are no longer effective at inhibiting the growth of the bacteria1. There is a growing increase of resistance by bacteria to antibiotics2,3,4,5,6,7,8,9,10 with the World Health Organization (WHO) through its Global Antimicrobial Surveillance System (GLASS) report revealing that there are high levels of antibiotic resistance in both low- and high-income countries11. In fact, the European Centre for Disease Prevention and Control (ECDC) reported that 25,000 people died of diseases caused by antibiotic-resistant bacteria in 2007, which is over half the number caused by road traffic accidents in the same countries12. In 2015, this number increased to about 33,000 deaths resulting from an estimated 671,689 infections of selected antibiotic-resistant bacteria leading to 874,541 total disability-adjusted life-years (DALYs)13. This indicates that the burden in the European Union and European Economic Area is on the rise. Likewise, the World Health Organization (WHO) predicted that by 2050, the number of people who will die due to antibiotic resistance would increase from 700,000 to about 10 million per year globally14. As a result of antibiotic resistance, more than 2.8 million people are infected, and more than 35,000 die each year in the USA15,16.

The burden caused by antibiotic resistance is greater in low- and middle-income countries (LMICs) whose health care systems are poor and lack tools to perform rapid diagnosis of the numerous neglected infectious diseases17. In the community settings in LMICs, high prevalence of multidrug-, extensive drug-, and pan drug-resistant commensal Escherichia coli isolated from healthy humans has been reported18. Thus, greater efforts should be placed in LMICs to contain the spread of antibiotic-resistant E. coli, especially with its relaxed antibiotics prescription policies. Antibiotic resistance has led to an increase in poverty in LMICs16, and antibiotics misuse is associated with the carriage of resistant commensal E. coli from healthy children in community settings worldwide19,20. In high-income countries, stringent antibiotic prescription policies are in place to reverse the course of antibiotic resistance21.

Commensal Escherichia coli is a gram-negative bacterium located in the gut of humans, animals, birds, and also exists in the environment22. It is a pathogen on the WHO global critical pathogen priority list for research, discovery, and new antibiotics development23. When we ingest antibiotics for the treatment of bacterial infections, the commensal E. coli is exposed to these antibiotics and can develop resistance to these antibiotics through natural selection24. Indeed, commensal E. coli is one of the major reservoirs for the transmission of antibiotic resistance to other pathogenic bacteria through plasmid exchange, for example (Fig. 1)25,26,27,28,29,30,31,32. Humans can be exposed to viable commensal antibiotic-resistant E. coli by contact with livestock or a contaminated natural environment and by inadequately cooked food or cross-contamination33,34,35.

Figure 1
figure 1

Transfer of resistance between bacteria through plasmid exchange. (a) Commensal E. coli, pathogenic E. coli and other pathogenic bacteria come together in the gut, (b) Bacteria attached and exchanged plasmids conferring antibiotic resistance. (c) Bacteria-acquired resistant plasmids.

Full size image

Many studies have shown a high prevalence of resistance to antibiotics by pathogenic and commensal bacteria in healthcare settings8. However, when comparing these studies to those conducted in community settings there is a large discrepancy, especially in LMICs36. This is as a result of the fact that little attention is given at the community level to contain the global antibiotic resistance crises; despite some studies that compare the prevalence of resistance to antibiotics in both communities and hospitals all showed consistently high values with no significant difference8,22,37,38,39,40. Thus, similar attention should be given to contain the cause of resistance to antibiotics by bacteria in communities, as is the case in hospitals. If this situation is not addressed, many of the gains in modern medicine will be lost and the commitment to achieve universal health coverage by world leaders will be in vain7,41. In this paper, we aim to provide an appraisal of the evidence of the high prevalence of antibiotic resistance by commensal E. coli to commonly prescribed antibiotics in community settings (i.e., locations outside the hospital such as homes and schools) in LMICs to bring to light the extent of the problem and inform interventions targeted at controlling and preventing antibiotic resistance. Indeed, a multitude of knowledge, attitudes, and practices (KAP), education, and community engagement interventions exist in community settings in low- and middle-income countries (LMICs), yet data to support and justify their set-up are often lacking. Thus, our data should prove useful to support the course for the fight against antibiotic resistance by researchers, community pharmacists, public health policymakers, advocacy groups, farmers, among others. A collective approach involving every country to fight antibiotic resistance is crucial to reduce the mortality, morbidity, associated health and healthcare costs, and the spread of resistant bacteria10,42.

Results

Literature search

A total of 9363 articles were obtained from the search (PubMed = 3634, EMBASE + MEDLINE = 2103, Web of Science = 3046, CINAHL = 290 and Cochrane Library = 289). Out of the 9363 articles, 2280 duplicates were removed using EndNote X8. We screened 7089 articles to identify article hits that met our inclusion criteria (Fig. 2). We performed a full-text screening of 53 studies and data were extracted from 33 articles8,28,29,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72. A total of 20 articles in which the isolates were pathogenic (14), were collected in a high-income country (3), could not be obtained online (2), or review articles (1) were excluded for data extraction.

Figure 2
figure 2

Flow diagram of the literature search strategy73. Additional records were identified from the reference lists of some of the included studies.

Full size image

Study location, strength of evidence, and study type

Out of the 33 included studies, 10 were from Africa, 13 from Asia, 1 from Europe, 8 from South America, and 1 from multiple locations (Fig. 3a,d). The quality of the evidence was assessed as described previously74,75. Of the 33 studies, 16 were marked as high quality, 16 as a medium, and 1 as low quality (Fig. 3b). These included studies were cross-sectional (25), cohort (6), case–control (1), and double-blind randomized control trials (1) (Fig. 3c).

Figure 3
figure 3

Characteristics of the included studies. (a) Location, (b) quality of the evidence, and (c) type of studies. (d) World map showing the number of studies by continents. The world map was obtained from (https://yourfreetemplates.com/) on 17th December 2020.

Full size image

Age, sample size, and gender

The included studies comprise of about 7755 health individuals in community settings from 0 to 77 years of age from 1989 to 2019. One limitation of this study is that some articles did not record the number of males and females, thus it was challenging to determine the ratio of males to females. However, we calculated the number of males and females using the studies that recorded the numbers. Of these studies, 2443 were males and 2010 were females. Thus, assuming a total population of 7755, we extrapolated the number of males to be 4255, and females 3500.

High prevalence of antibiotic resistance in the community settings will potentially impact health policy

The pooled prevalence of commensal E. coli isolated from healthy individuals in community settings in LMICs for the different antibiotics are summarized in Table 1, Fig. 4, and Supplementary Fig. 1. A high prevalence was seen for some of the commonly prescribed antibiotics in these countries like ampicillin (72%, 95% CI: 65–79), cefotaxime (27%, 95% CI: 12–44), chloramphenicol (45%, 95% CI: 35–53), ciprofloxacin (17%, 95% CI: 11–25), co-trimoxazole (63%, 95% CI: 52–73, nalidixic acid (30%, 95% CI: 21–40), oxytetracycline (78%, 95% CI: 65–88), streptomycin (58%, 95% CI: 44–72), tetracycline (67%, 95% CI: 59–74), and trimethoprim (67%, 95% CI: 59–75). These findings will be very useful for evidence-based health policy design aimed at combating the spread of antibiotic resistance in the community.

Table 1 Prevalence of antibiotic resistance in commensal E. coli isolated from human sources in community settings in low- and middle-income countries.
Full size table
Figure 4
figure 4

Forest plots showing the prevalence of antibiotic resistance in commensal E. coli isolated from human sources in community settings in low- and middle-income countries. (a) Cell wall synthesis inhibitor (ampicillin), (b) protein synthesis inhibitor (tetracycline), (c) folate synthesis inhibitor (co-trimoxazole), and (d) nucleic acid synthesis inhibitor (ciprofloxacin).

Full size image

Prevalence of antibiotic resistance data of a plethora of antibiotics will potentially impact antibiotic stewardship programs

In this study, the prevalence of resistance was collected for a range of diverse antibiotics and mechanisms of actions such as, inhibition of protein, nucleic acid, folic acid, or cell wall synthesis. For some antibiotics, the resistance was very high, others were emerging, and for some it was low (Table 1 and Supplementary Table 4). This evidence should prove useful to inform antibiotic stewardship programs.

Investigating the source of heterogeneity

The results show a high I2 value, which indicates considerable heterogeneity. Funnel plots and Egger’s regression test was used to explore the sources of heterogeneity. The funnel plots for all antibiotics were asymmetrical (Supplementary Fig. 2), thus indicating a possibility of publication bias. We further investigate the existence of publication bias per antibiotic using Egger’s regression test (Supplementary Table 5). Most of the antibiotics (ampicillin, cefotaxime, co-trimoxazole, nalidixic acid, streptomycin, and trimethoprim) did not exhibit publication bias. The antibiotics that exhibited publication bias were chloramphenicol, ciprofloxacin, oxytetracycline, and tetracycline. On further stratification of each antibiotic prevalence, according to continents (Africa, South America, and Asia), there was a reduced likelihood of publication bias based on visual examination of the funnel plot.

Discussion

Antibiotic resistance (ABR) is a serious global health threat that needs to be addressed urgently2,3,4,5,6,7,8. ABR's impact is particularly greater in low- and middle-income countries (LMICs), which bear the highest-burden and subsequently suffer the most from this problem, primarily because their healthcare systems lack the resources needed to contain or to treat challenging infectious diseases caused by drug-resistant bacteria17. Indeed, it has led to an increase in poverty in LMICs16. While there has been a general increase in multidrug-resistant pathogenic bacteria in community settings77 recent evidence suggests that the prevalence of multidrug-resistant commensal Escherichia coli isolated from healthy individuals is particularly high in LMICs18. In this study, we synthesized a total of 33 articles to obtain a pooled prevalence of ABR in the top ten antibiotics commonly prescribed in community settings (i.e., locations outside of a hospital, such as schools and homes) in LMICs.

There are several factors that contribute to ABR78, with a complicated inter-relationship that spans across different sectors outside of healthcare alone, such as agriculture and industry. Among the main factors identified leading to resistance in commensal E. coli in LMICs are overcrowding, poverty, socioecological behaviours, food and supply chain safety issues, highly contaminated waste effluents and inadequate surveillance systems79. Nevertheless, the primary driver of multidrug-resistance in LMICs has been misuse and over-prescription of antibiotics80. Commensal E. coli are typically present in the guts of humans, animals, birds, as well as in the environment, and can develop resistance to antibacterial agents through natural selection when ingested for the treatment of bacterial infections22,24.

This review revealed a high prevalence of antibiotic resistance in commensal E. coli to the most prescribed antibiotics in LMICs. Moreover, this was a consistent finding across several classes of antibiotics with different mechanisms of action. For instance, the pooled prevalence of antibiotic resistance for the β-lactam antibiotic ampicillin was 72%, 95% CI: 65–79, while for trimethoprim, a folic acid synthesis inhibitor, the pooled prevalence was 67%, 95% CI: 59–75. Similar observations were reported in a recent systematic review investigating ABR in E. coli strains isolated from humans, animals, food, and the environment in several middle- and high-income countries. The authors presented high rates of resistance against a range of antibiotics found in E. coli isolates, although the pooled prevalence was generally lower than in the isolates from healthy individuals from LMICs presented here81. For ciprofloxacin, the pooled prevalence from our study was 17%, 95% CI: 11–25, which is in the range of the resistance seen for treating E. coli associated urinary tract infection (8% to 65%)11, and other E. coli isolated from farmed minks in Zhucheng, China76. Likewise, for cotrimoxazole, our data (63%, 95% CI, 52–73) agrees with another study carried out in Zimbabwe, where the prevalence was 68% for Gram-negative bacillli82. The main limitation of our study is the fact that the heterogeneity between studies was very high (Table 1, Fig. 4). We performed an additional statistical analysis stratified by continent. The goal was to solve the heterogeneity issue, however, there was no significant difference between the pooled prevalence of antibiotic resistance values between continents. The high heterogeneity between studies could stem from the different factors associated with the carriage resistant commensal E. coli in LMICs highlighted in the discussion section above. Since I2 statistics test for heterogeneity can be misleading during meta-analysis of observational studies83,84, we performed an alternative assessment of the strength of evidence of the different studies (supplementary table 3)74,75. The included studies utilized disc diffusion or synergy test to investigate the expression of resistant genes in the presence of antibiotics. However, those that employed an additional method, such as PCR, plasmid transfer assay, nucleic acid identification, mass spectrometry, to validate the presence of genes conferring resistance to a particular antibiotic were graded as high.

Our study is in line with the WHO’s Global action plan on antimicrobial resistance, which calls for improved awareness of the problems arising from antibiotic resistance as stated in one of the five strategic objectives85. Indeed, raising awareness can be facilitated, for example, through the making of participatory videos86. As expected, in this study, resistance was commonly detected from stool samples collected from healthy volunteers in community settings. Our findings have important ramifications for public health policies and antibiotic resistance stewardship through a one-health approach for the fight against ABR. In some of the studies screened, factors such as previous antibiotic use19,55,59,65, geographical location8,57, age8,59, socioeconomic status43,55,57, and exposure to animals49,65 were highlighted as being associated with a high prevalence of resistance. The situation is even more pressing with the emergence of global pandemics such as the coronavirus disease 2019 (COVID19) caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2)87. Prior to the availability of approved vaccines, different medicines have been tested randomly in clinical trials to find a cure for this deadly pandemic87,88,89,90,91,92,93,94. Since viral infections are often associated with bacterial infections95,96, the antibiotic azithromycin in combination with the antimalarial drug hydroxychloroquine was proposed as an option for the treatment of COVID19 patients97. Thus, we may see a post-COVID19 global health crisis with a surge in antibiotic resistance leading to many deaths97.

Conclusion

This study provides further evidence of the high prevalence of antibiotic resistant commensal E. coli from healthy human sources in community settings in LMICs. These findings should encourage health researchers, medical professionals, advocacy groups, and health policymakers to work together to develop appropriate interventions to counteract this growing global health threat. We recommend that the strategies that have been implemented in healthcare settings to contain the spread of resistance, such as surveillance, raising awareness, improve sanitation and hygiene, rapid diagnosis of diseases, and stringent prescription policies should also be urgently implemented in the communities to curb antibiotic resistance.

Methods

Design

A systematic approach was used to retrieve and synthesize studies that met our inclusion criteria following PRISMA guidelines73.

Type of studies

The types of studies included in the systematic review are those in which the main outcome was the prevalence of antibiotic resistance in commensal Escherichia coli: cross-sectional, case–control, cohort studies, and randomized control trials.

Type of participants (study population)

This review included studies concerning the general healthy populations in community settings in LMICs.

Outcome of interest

  • The primary outcome was prevalence of resistance to antibiotic by commensal E. coli in community settings in LMICs.

  • The secondary outcomes were: odds ratio, risk ratio, rate, 95% confidence interval, and p-value.

Other inclusion and exclusion criteria

Articles written in English language were included from 1989 to May 2020 in LMICs. Articles containing studies conducted in a country that was classified as a LMIC before transforming into a high-income country according to the World Bank definition were also included. The included studies must investigate the resistance of commensal E. coli on either solid or liquid growth media in the presence of antibiotics.

Essential data necessary for inclusion

Studies were eligible for inclusion if they reported at least one of the primary or secondary outcomes listed in the outcome of interest section above.

Data sources

Published data from PubMed, EMBASE, MEDLINE, Web of Science, CINAHL and Cochrane Library, reference lists of selected studies and unpublished data such as abstracts from Conference proceedings; dissertations and theses were the data sources.

Systematic search strategy

A literature search was performed on the 10th of March 2018 and was updated on the 18th of May 2020. Briefly, we searched PubMed, EMBASE, MEDLINE, Web of Science, CINAHL, and Cochrane Library using MeSH terms for PubMed and the comparable terms for the other databases. The search terms were “E. coli OR Escherichia coli OR Enterobacteriaceae” AND "antibiotic resistance OR antimicrobial resistance OR drug resistance” AND "prevalence OR incidence OR morbidity OR odd ratio OR risk ratio OR confidence interval OR p-value OR rate". For PubMed, EMBASE, and MEDLINE, studies performed in humans were selected using the Species filters. While for Web of Science and Cochrane Library additional search words were added to select for species (human* OR infant* OR child* OR adolescen* OR male* OR female OR age OR adult*) since there was no sorting filter for species. For CINAHL, no selection for species was performed. Search words were designed from the different categories in the PICO (Population, Intervention, Comparison, and Outcome) format. Details of the search terms used are summarized in Supplementary Table 1. The articles obtained from the search were exported to EndNote for duplicate removal. The unique hits were further exported to Rayyan QCRI website for screening and data extraction98. An initial screening was performed by title and abstract, followed by full article text.

Data extraction and quality assessment

A two-step process was followed involving screening of titles and abstracts to identify relevant articles, which was followed by full-text reading of the relevant articles. A total of 53 full-text articles were screened, and the data extracted and recorded on an excel spreadsheet by five researchers. To exclude selection bias, a sixth researcher was available to solve the disagreements that arose during the data extraction. The parameters that were extracted are listed in Supplementary Table 2. The quality of evidence in the included studies was assessed as described before72,73 (Supplementary Table 3). In brief, studies that utilized disc diffusion or synergy test in combination with one of the following biochemical test such as, polymerase chain reaction (PCR), plasmid transfer assay (PTA), pulsed-field gel electrophoresis (PFGE), nucleic acid sequencing, and mass spectrometry to detect the presence of resistant genes was graded as high. Furthermore, studies which use only disc diffusion of synergy test with a sample size less than 15 were classified as medium. Lastly, studies which sample size of below 15 and utilized only disc diffusion or synergy test were classified as low.

Statistical analysis

Statistical analysis was performed in Stata99 and JBI SUMARI100. The prevalence of antibiotic resistance of commensal E. coli was defined as the proportion of the isolates in a specific study that were found to be resistant to a given antibiotics presented as a percentage. The pooled prevalence was calculated using the metaprop command in Stata101. Metaprop pools proportions and presents a weighted sub-group and overall pooled estimates with inverse-variance weights obtained from a random-effects model. In this case, it involved a meta-analysis of the prevalence values of the individual publications weighted on sample size while accounting for potential heterogeneity between studies. For JBI SUMARI, proportional meta-analysis was calculated using the random-effects model of Freeman-Tukey transformation. Besides, a forest plot was constructed for each of the top ten most reported antibiotics in our study.

The source of heterogeneity was explored by stratifying the pooled prevalence by year of sample analysis (study period), geographical location (continents), and type of antibiotic. Funnel plots of overall effect size were run to determine the existence of publication bias by visual inspection. Also, the Egger’s test was used to assess the occurrence of small size effect. The level of significance was maintained at 0.05.