Prevalence and Antibiotic Resistance in Campylobacter spp. Isolated from Humans and Food-Producing Animals in West Africa: A Systematic Review and Meta-Analysis

Campylobacter species are one of the leading causes of gastroenteritis in humans. This review reports on the prevalence and antibiotic resistance data of Campylobacter spp. isolated from humans and food-producing animals in West Africa. A systematic search was carried out in five databases for original articles published between January 2000 and July 2021. Among 791 studies found, 38 original articles from seven (41%) out of the 17 countries in West Africa met the inclusion criteria. For studies conducted in food-producing animals, the overall pooled prevalence of Campylobacter spp. was 34% (95% CI: 25–45). The MDR prevalence was 59% (95% CI: 29–84) and half (50%, 13/26) of the animal studies had samples collected from the market. The human studies recorded a lower pooled prevalence of Campylobacter spp. (10%, 95% CI: 6–17), but a considerably higher rate of MDR prevalence (91%; 95% CI: 67–98). The majority (85%, 11/13) of the human studies took place in a hospital. Campylobacter jejuni and Campylobacter coli were the most common species isolated from both animals and humans. Our findings suggest that Campylobacter spp. is highly prevalent in West Africa. Therefore, improved farm hygiene and ‘One Health’ surveillance systems are needed to reduce transmission.


Introduction
Animals are natural reservoirs for Campylobacter spp. [1], which are among the leading causes of bacterial gastroenteritis in humans, worldwide [2]. Human Campylobacter infection is mainly acquired by the consumption of undercooked poultry, livestock, or by direct contact with animals [1]. A significant proportion of the population in Africa keeps livestock and/or poultry [3]. However, these animals are often reared and slaughtered under poor hygienic and sanitary conditions [4] and high frequencies of Campylobacter have been

Antimicrobial Resistance Profile of Campylobacter Species
Apart from ciprofloxacin and nalidixic acid, a higher proportion of antibiotic resistance was observed in Campylobacter isolates from humans than from animals ( Figure 7). The majority of Campylobacter isolates recovered from animals were susceptible to gentamicin (94%, n/N = 1183/1265) and those from humans to ciprofloxacin (69%, n/N = 175/255).

Discussion
This review shows that the majority (58.8%) of countries in West Africa has no published studies on Campylobacter prevalence that met our inclusion criteria. This finding is in agreement with other systematic reviews that also report low numbers of Campylobacter research in Africa [8,14]. The cumbersome procedures involved in isolating Campylobacter spp. makes it difficult for most low-income countries to conduct such studies. Nigeria (50%) and Ghana (23%) had the highest number of studies, probably because these countries have a higher socio-economic status in the region, hence they can afford well-equipped health and research facilities needed to conduct such research [51]. The number of studies published within 2011-2021 was far more than during the preceding decade. This could be because knowledge of new and advanced methods of detection (such as PCR) became available in recent years. Additionally, researchers are now becoming more aware of the burden of Campylobacter infections in humans and animals.
Poultry and livestock samples recorded the highest pooled prevalence of Campylobacter spp. The intestinal tract of poultry and livestock are frequently colonized in high numbers by Campylobacter spp., hence constituting a natural reservoir and an important source of transmission [1,52]. Studies conducted on Campylobacter spp. colonization in poultry and livestock are in agreement with the current findings [1,53]. This suggests that poultry and other animals are primary reservoirs responsible for Campylobacter infections in humans [54,55].
Our review shows that the pooled prevalence of Campylobacter infections in humans was 10% (95% CI: [6][7][8][9][10][11][12][13][14][15][16][17]. Summarized findings from Ethiopia [56] and Sub-Saharan African countries [8] have reported a similar pooled prevalence in humans (9-10%). West Africa has just four published articles on Campylobacter infection in children under 5 years of age. Surprisingly, these studies recorded a pooled prevalence of 4%, lower than the 10% reported in a summary of results from Ethiopia [57]. The lower Campylobacter prevalence observed in children might be because most of the included studies were conducted in healthy (non-diarrhea) subjects. The low number of studies conducted within this age group shows that some populations in West Africa have not been investigated; hence, there is the need to conduct more studies in these populations. Our review reported high heterogeneity between studies, which could be due to differences in environmental conditions, socio-demographics, sociocultural factors and disease awareness levels. Additionally, the protocols used and the experience level of staff in isolating the bacteria could account for the differences in prevalence observed between studies.
Approximately 56% of the studies used culture and biochemical tests to identify Campylobacter spp. Other reviews conducted in Africa have also observed that this method is most common for identifying Campylobacter in the region [8,11]. However, the culture method has some limitations; environmental stress during sample transportation and processing can make some Campylobacter spp. viable but not culturable on media [58], this could lead to lower sensitivity [59]. In our review, the culture and biochemical method produced a high pooled prevalence of Campylobacter species. However, these findings must be interpreted with caution because the culture method has lower specificity compared to PCR-based methods. A lot of researchers in West Africa rely on the laborious and time-consuming culture method because their laboratories are not well equipped to use PCR in the diagnoses of Campylobacter. This could be a possible reason why fewer studies have been conducted.
C. jejuni and C. coli were isolated by 28 and 24 studies, respectively, making them the predominantly isolated Campylobacter species from both food-producing animals and humans. Other authors have reported similar observations [8,11]. The high numbers of virulence genes associated with C. jejuni and C. coli [60] possibly make researchers develop research questions focused on discovering more of these species. Additionally, the high prevalence of C. jejuni and C. coli observed in this study could be attributed to the culture and biochemical test method used for speciation, which is incapable of detecting the lesserknown Campylobacter spp. [61]. Campylobacter selective media containing antibiotics and higher incubation temperatures does inhibit the growth of some Campylobacter species such as C. upsaliensis and C. lari [11]. Nonetheless, it is well known that the two Campylobacter species most frequently associated with diarrhea in humans are C. jejuni and C. coli [62]. The 69% AMR recorded by this review shows that Campylobacter's resistance to commonly used antibiotics is widespread in both humans and animals. Among the most tested antibiotics, Campylobacter were found to be highly resistant to tetracycline, nalidixic acid, ciprofloxacin, erythromycin, chloramphenicol, ampicillin and gentamycin. Ciprofloxacin resistance rate in animal isolates was higher than humans, even though it is not approved for use in veterinary medicine. Consistent with our findings, 88% of E. coli isolated from poultry farms in Ibadan, Nigeria were resistant to ciprofloxacin [63], suggesting that its use in poultry and livestock farming may be on the increase. We also observed high erythromycin resistance in human isolates compared to animals. The high erythromycin resistance observed in humans could be explained by the overuse of azithromycin due to its low risk of side effects [64]. The high antimicrobial-resistant rate observed in our study agrees with findings from similar studies conducted in both low and middle-income countries [65] and high-income countries [66] showing an increasing trend of antibiotic resistance in Campylobacter spp. The increasing trend might also be due to the extensive use of antimicrobials in animal farming for growth promotion and prophylaxis [67] and the indiscriminate use in humans [68]. Carbapenem resistance is on the increase in Gram-negative bacteria [69]; however, our review observed that all isolates tested against imipenem were susceptible. The high rate of Campylobacter susceptibility observed against imipenem might be attributed to it not being authorized for use in animal husbandry [70].
To lower the high Campylobacter prevalence and antimicrobial resistance observed in this review, we recommend the appropriate use of antibiotics in human and veterinary medicine, improved hygiene and sanitation practices and the implementation of biosecurity measures in farms [65]. If possible, antimicrobial susceptibility testing should be performed before the administration of antibiotics to humans. Since the virulence and pathogenicity of Campylobacter is affected by the genetic variants, we recommend the use of molecular diagnostic methods in addition or as a replacement to the widely used culture method, in order to accurately diagnose infections and to determine the real Campylobacter burden [60]. Furthermore, strong commitment from policymakers is needed to implement 'One Health' surveillance systems.
Our systematic review and meta-analysis have few potential limitations. The search strategy was limited to only articles published in English, there might be articles published in other languages that were not considered. The analyses were not uniformly spread since data was absent from majority of the countries and most of the studies were conducted between 2011 and 2021. Another limitation of our review is that majority of the studies used culture methods, which is not the preferred method for reporting Campylobacter prevalence. We recorded high heterogeneity because studies conducted in different countries and under different conditions were pooled together. Since we only found data from less than half of the countries in West Africa, our findings may not be generalizable to the entire region.

Study Design and Systematic Review Protocol
The protocol of this review is registered at PROSPERO with registration number: CRD 42021260515. This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [71]. The UN macrogeographical definition of West Africa (https://unstats.un.org/unsd/methodology/m49/ (accessed on 29 September 2021) was used to define West African countries included in this review, namely: Benin, Burkina Faso, Cape Verde, Côte d'Ivoire, Gambia, Ghana, Guinea, Guinea-Bissau, Liberia, Mali, Mauritania, Niger, Nigeria, Saint Helena, Senegal, Sierra Leone and Togo.

Selection Criteria and Literature Search Strategies
A systematic search for original articles covering West African countries and published between January 2000 and July 2021 was conducted using the following databases: Medline (via PubMed), Directory of Open Access Journals (DOAJ), Google Scholar, African Index Medicus and the African Journal Online (AJOL) database. The systematic search of these databases was performed using the search terms listed in the Supplement File S1. EKP screened the titles and abstracts of all recovered articles. Articles were eligible for full-text review when: (i) they contained data from a West African country, (ii) they were published between January 2000 to July 2021, and (iii) they were written in English. During the full-text review, two authors (EKP, SA) independently assessed the articles to determine if each one met the inclusion criteria. An article was included if it contained primary data, was conducted in food-producing animals and/or humans, and Campylobacter prevalence was reported or can be calculated from available information. Articles whose full texts could not be accessed and those with inconsistent results, overlapping or duplicate data were excluded. Additionally, articles that did not report on the age of study participants, type of samples collected and laboratory diagnostic method used were excluded. In case of any disagreement in the review process, a third reviewer (LAO) was available to give a decisive opinion on any unresolved issues.

Data Extraction
For each included original full-length study article, we extracted data on the first author, year of publication, name of the country where the study was conducted, type of food-producing animals sampled, age of human participants, type of samples collected, sample size, study design and study setting. We also collected data on the laboratory diagnostic methods used, Campylobacter prevalence, Campylobacter spp. isolated, antimicrobials tested and antibiotic resistance.

Risk of Bias Assessment
Conventional funnel plots show inaccurate results when assessing publication bias in systematic reviews on prevalence studies [72]. This is because of the unequal and small sample sizes, high prevalence diversity due to study design differences and zero prevalence which may be recorded in studies. Therefore, we decided to include all studies that met the final inclusion criteria without assessing the risk of publication bias. Nonetheless, funnel plots were plotted to indicate the across-study biases and between-study heterogeneity.

Data Analysis
In studies that did not explicitly report Campylobacter prevalence, but reported the number of positives and the total number of samples collected, the prevalence was calculated as the fraction of both terms. The Meta (version 5.0-1, R Core Team, Vienna, Austria) package, in R software (version 4.1.1, R Core Team, Vienna, Austria) [73], was used to calculate pooled prevalence estimates using a random-effects model [74]. The pooled prevalence with a 95% confidence interval (CI) was presented using forest plots and tables. The heterogeneity of study prevalence estimates was evaluated by computing the inverse variance index (I 2 ) statistic. Heterogeneity was considered to be high when I 2 , which describes the percentage of total variation between studies that is attributable to prevalence differences rather than chance, was above 75%.
Subgroup analyses were used to investigate potential associations with the prevalence estimates. The potential sources of heterogeneity were investigated considering the year of publication, the country where sampling occurred, study setting, sample type, age of human participants, laboratory diagnostic method used and types of Campylobacter species isolated. For human studies, the subgroup analysis also included patients with and without diarrhea, HIV and urinary tract infections (UTI). The proportion of Campylobacter spp. that were resistant to commonly tested antibiotics was calculated for both food-producing animal and human studies. The ggplot2 package, in the R (version 4.1.1) statistical environment, was used to plot a bar chart to illustrate the proportion of resistant Campylobacter in humans and animals. QGIS software (version 3.18.3, QGIS Development Team, Zurich, Switzerland) [75] was used to draw a map to show the number of Campylobacter prevalence studies across West Africa.

Conclusions
Research articles on Campylobacter prevalence were not available from 59% of countries in West Africa. Countries in West Africa should be supported to have well-equipped laboratories for Campylobacter research. To curb the high Campylobacter prevalence and resistance observed in this review, routine diagnosis, appropriate use of antibiotics, improved hygienic practices and 'One Health' surveillance systems should be implemented. Furthermore, strong commitment from policymakers and societal actions are needed to improve farm hygiene and antimicrobial usage in food-producing animals and humans.