Introduction

Non-typhoidal Salmonella enterica (NTS) is an important cause of invasive infections in people with significant public health and economic impacts in Africa1,2. NTS bacteremia is particularly problematic for people living with HIV and is associated with malaria, particularly in children and elderly individuals2,3. Annually, NTS causes an estimated 31.8–211.2 million illnesses and 36.3–89.1 million deaths worldwide4. Poultry is an important reservoir for NTS with zoonotic transmission occurring through food products, water, and direct contact5.

NTS infections usually resolve without treatment, but the disease can progress to severe bacteremia or invasive infection6 with cephalosporins being the treatment of choice in countries such as Nigeria7. Frequent use of antibiotics likely ensures that NTS from poultry farms will harbor resistant strains that become more numerous in the presence of antibiotic selection pressure8. Such practices may include in ovo administration to limit chick bacteremia, and oral metaphylactic administration intended to insure against widescale bacterial infections9. And while the cost of cephalosporins probably limits their use for poultry production within low- and middle-income countries, use of antibiotics such as ampicillin will selectively favor extended-spectrum cephalosporin-resistant Enterobacteriaceae10 that are resistant to multiple antibiotics11,12. Consequently, widescale antibiotic use on poultry farms will likely increase the prevalence of antimicrobial resistant NTS on farms, with a commensurate increase in the likelihood that NTS strains will be transmitted to farm workers and poultry products. Like other enteric bacteria, Salmonella enterica acquires of resistance genes by plasmid conjugation, usually with other Enterobacteriaceae. Bacteria that develop resistance via extended-spectrum β-lactamase genes could become a reservoir of resistance genes, that further proliferate these genes in food animal microbiomes.

Like many low- and middle-income countries, Nigeria does not have a surveillance network to track the prevalence of antimicrobial resistance in human or animal settings. Further, Nigeria has no regulatory control of antibiotic use in food-animal production13,14, which leaves opportunities to employ low-cost antibiotics as a primary disease prevention measure. To better understand the prevalence of NTS and antibiotic use practices in the Nigerian poultry industry, we assessed fecal samples from birds and farm workers and assessed the relationship between the probability of detecting NTS on farms relative to basic biosecurity controls. The farms from this study are in Lagos State, which has one of the highest concentrations of poultry operations in Nigeria and is a significant source of egg production for the country15.

Results

A total of 199 randomly selected farms were included in this survey (Table 1). One farm worker completed a questionnaire from each farm, most of whom were male (78.4%), most had a tertiary education (70.9%), and poultry farming was the only source of income for most workers (70.9%) (Table S3). Farms included small-holder semi-commercial operations (25.1%), small-scale commercial operations (23.1%), medium-scale operations (41.2%), and large-scale operations (10.6%). All farms (100%) employed battery cages for animal housing. Presumptive NTS Salmonella was detected by agar plating and PCR confirmation (invA detected) for 28 out of 199 farms (14%) and for 6 out of 60 of human stool samples (10%) (Table 1).

Table 1 Prevalence of non-typhoidal Salmonella (NTS) from pooled fecal samples collected from layer chickens, and individual fecal samples collected from farm workers (one worker per farm).
Full size table

Most poultry farms (84.4%, 168/199) were fenced with most farms (58.8%, 117/199) using foot dips, and most used disinfectants in the dips (Table 2). Only 24.6% (49/199) of respondents washed hands with soap after handling poultry, while the majority (71.9%, 143/199) had farm-dedicated work clothing. Biosecurity measures (foot dips, fencing, dedicated clothing, hand washing) were disproportionately represented by farms that were NTS negative (p < 0.03, Table 2). Use of tire dips and awareness of Salmonella as a hazard were not related to NTS status. Manure sales represent another potential route of transmitting NTS to the public, and selling such products appeared to be equally likely for NTS positive and negative farms. A logistic regression model emphasized the importance of using foot dips (odds ratio = 0.23, 95% confidence interval 0.07–0.58), p = 0.0036) (Table 3).

Table 2 Cross tabulation of biosecurity measures and presence of non-typhoidal Salmonella (NTS) on 199 chicken layer farms in located in southwestern Nigeria (Badagry, Epe, Ikeja, Ikorodu, and Lagos divisions).
Full size table
Table 3 Multivariate logistic regression analysis of factors associated with non-Typhoidal Salmonella occurrence in 29 of 199 poultry farms located in southwestern Nigeria (Badagry, Epe, Ikeja, Ikorodu, and Lagos divisions).
Full size table

Most farms (94.5%, 188/199) reported using antibiotics with a combination of prophylactic and therapeutic applications, and the frequency of use varied from weekly to once every one to two months (Table 4). A minority of participants were aware of recommended withdrawal periods (22.6%, 45/199) or were aware that residuals can be found in food products (< 20%). None of the sampled farms complied with antibiotic withdrawal recommendations. Most farms reported using commercially available antibiotics including cocktails with more than one antibiotic (Table 4). In several cases the products contained medically important antibiotics (e.g., colistin), antibiotics that should not be used in food animals (e.g., chloramphenicol), and human rather than the veterinary analogues (e.g., ciprofloxacin instead of enrofloxacin) (Table 5). Products containing sulfa and tetracycline class antibiotics were used most frequently.

Table 4 Antibiotic use practices at 199 poultry farms located in southwestern Nigeria (Badagry, Epe, Ikeja, Ikorodu, and Lagos divisions).
Full size table
Table 5 Antibiotic use practices at 199 poultry farms located in southwest Nigeria (Badagry, Epe, Ikeja, Ikorodu, and Lagos divisions).
Full size table

Discussion

Poultry is a reservoir for NTS, and regions that produce large quantities of birds and eggs are potential foci for disseminating contaminated products over large geographic areas. Our study region is important because it is the hub of poultry farming and open market trading of veterinary drugs (including antibiotics) in Nigeria. To better understand the magnitude of antibiotic use and prevalence of NTS on farms, it is important to generate unbiased prevalence estimates from a representative sample of farms in our study region. In our case, we approached 217 randomly selected farms of which 199 agreed to provide information and fecal samples for which we found the prevalence of NTS to be 14% (28/199) in poultry farms and 10% (10/60) for otherwise presumptively healthy farm workers. This is likely to be an underestimate given a relatively low-intensity sampling effort, although the prevalence obtained in the our study is similar to the overall prevalence of NTS reported by Fagbamila et al., for laying chickens in Nigeria (14.3%), but lower than what Fagbamila et al. reported for Lagos State (20%)18. Jibril et al.23, however, detected a higher prevalence (47.9%) in northwestern Nigeria. This difference could be due to different sampling methods, collection seasons, culture methods, laying period, housing system and frequency of use of antibiotics administered to poultry.

Estimates of colonization with NTS varied across other regions as well. A similar prevalence (13.3%) was detected in Argentina24 while higher NTS prevalences were reported for Vietnam (45.8%)25 and Morocco (73.3%)26 laying hens. Salmonella found in otherwise healthy laying chickens represents a significant risk factor for transmission to people27, which could be reflected by the detection of NTS in 10% of poultry workers, which was higher than NTS-positive workers reported in Ethiopia (2.8%)28 and in the Mekong Delta of Vietnam (4.4%)25.

Salmonella can be a health challenge for flock health, and thus producers should be incentivized to employ biosecurity measures to reduce the spread of this pathogen and others within and between farms. Bivariate analysis showed that employing low-cost biosecurity measures (Tables 2 and 3) reduced the odds of detecting Salmonella on farms, and a multivariate logistic regression analysis suggested that having foot baths with disinfectant is a particularly effective tool, which benefits both producers and consumers. This finding is consistent with other work showing that relatively simple biosecurity measures on a farm (e.g., foot baths, fenced farms, using farm-dedicated clothing) will limit the risk of entry of a pathogen on a farm18,29.

Antibiotics are widely used around the world to manage food-animal production (growth promotion, disease treatment and prevention), but the current laissez faire market model in southwest Nigeria appears to encourage particularly egregious antibiotic use practices. For example, only one out of eleven antibiotic products (oxytetracycline hydrochloride) observed on farms appeared to be used appropriately, and most farms (86.4%, 172/199) reported using antibiotic cocktails that included medically important antibiotics such as colistin and ciprofloxacin, and antibiotics that are prohibited in food animals raised in the European Union and in the United States. For example, chloramphenicol was banned from food-animal production in the US (1984) because residues can cause dose-independent hemolytic anemia in people30,31. This study corroborates similar findings about use of prohibited drugs in poultry in Oyo, Ogun and Borno states, Nigeria14,32,33.

More generally, no farms followed recommended withdrawal times as corroborated by studies conducted in Oyo, Ogun and Borno States, Nigeria14,32,33. Withdrawal times are intended to protect consumers by allowing time for residues to clear meat products before they are sold to consumers. When withdrawal periods are not followed, consumers are exposed to residues from antibiotics that are can be dangerous, such as chloramphenicol and beta-lactam antibiotics with the latter potentially causing dangerous hypersensitive anaphylaxis in people34.

Conclusions

This study likely underestimated the prevalence of NTS on farms because of limited on-farm sampling, but the large number of farms included in the random sampling frame provided sufficient power to detect the benefits of improved biosecurity, and we assume that the prevalence estimates for antibiotic use are unbiased. Given these findings, we recommend that public health and agricultural measures include increased efforts to educate producers about effective and low-cost biosecurity measures to limit NTS on farms, about use of appropriate antibiotics for poultry, and most importantly, about the importance of adhering to recommended drug withdrawal periods to limit harm to consumers. It is also advisable for the Federal Ministry of Agriculture to develop intervention strategies and relevant policies for microbial food safety monitoring programs, and control of antibiotic use in poultry farms in Nigeria.

Methods

Sampling frame

A cross-sectional study was conducted across the five government divisions of Lagos State (Table 1) between June 2018 and December 2019. A stratified random sampling method was used to select commercial chicken layer farms from a list 785 registered commercial poultry farms obtained from the Lagos State Agricultural Development Authority. Selection was proportional to the number of registered commercial chicken layer farms in each district. If a selected farm refused to participate, it was replaced with a randomly selected farm for a total of 199 participating farms.

Questionnaires

Five research assistants from the University of Ibadan received training and to ensure that there was a common understanding of the questions among the team and to ensure that questions were administered in a consistent manner (each assistant worked in a different district). Questionnaires were administered to consenting farm workers (one per farm) to collect information about the demographic characteristics of farmers, farm characteristics, and antibiotic use practices (Table S3). A subset of these same individuals provided stool samples for bacteriological testing (see below). Study protocols were approved by the University of Ibadan, and University College Hospital Ethics committee (#UI/EC/18/0104), and by the Animal Care and Use Ethics Committee (#UI-ACUREC/17/0108). As such, all research was performed in accordance with relevant guidelines/regulations, and verbal informed consent was obtained from all participants (Table S2).

Sample collection

Only one flock per farm was sampled to estimate the prevalence of Salmonella at the level of farm. Low-intensity sampling of this nature could lead to false negative findings, but this was compensated to some extent by sampling from the longest continuously occupied layer pens that are more likely to be contaminated with Salmonella16. Within flocks, pooled samples were collected to increase the likelihood of detecting Salmonella [shedding is generally intermittent17] and the number of pooled samples depended on flock size (Table 1). Each pool consisted of five freshly voided fecal samples (200–300 g each)18. Poultry handlers were given swabs with Cary-Blair transport medium for stool sample collection, and an appointment was fixed for sample collection. NTS remains viable for at least 48 h in Cary-Blair medium16,18. Human and poultry samples were stored with ice in separate well-labelled containers and transported to the Food and Meat Hygiene Laboratory, Department of Veterinary Public Health and Preventive Medicine, University of Ibadan, within 24 h of collection for processing.

Sample processing and assays

Fecal samples were suspended in buffered peptone water (BPW; 1:10 w/v) and incubated at 37 °C for 24 h. Following ISO 6579 (ISO, 2002), aliquots of cultured BPW were transferred to Rappaport Vassiliadis Soya broth (Oxoid) and Muller-Kauffman tetrathionate novobiocin (Oxoid) broth and incubated for 24 h at 42 °C and 37 °C, respectively. Afterwards, cultures were streaked onto xylose lysine deoxycholate agar (Oxoid) and MacConkey agar (Oxoid) and were incubated at 37 °C for 24 h. Up to three colonies that appeared typical of NTS were chosen and sub-cultured on nutrient agar plates for 24 h at 37 °C19. Putative NTS were isolated and identified according to WHO Global Foodborne Infections Network (2010)20, including a triple sugar iron test (TSI), methylene red (MR) test, Voges-Proskauer test, and indole test (Kovac's reagent)19. Isolates were further confirmed by PCR detection of the invA locus following the methods of Kadry et al.21. S. enterica ATCC S26501 was used as a positive control. Primer sequences, PCR conditions and interpretation are described in Table S1.

DNA for PCR was prepared as boiled extracts following Dias et al.22. Each PCR reaction (25 µl total volume) included 12.5 µl Taq master mix, forward and reverse primers (1.5 μl. conc. 15 pmol. stock solutions), 6.5 µl sterile distilled water, and 5 µl boiled extract. Thermalcycler conditions included 5 min initial denaturation (95 °C), followed by 40 cycles of 45 s at 95 °C, and 45 s at 52 °C. After cycling, reactions were held at 72 °C for 10 min. Each assay run included a no-template negative control, and S. enterica ATCC S26501 was used as a positive control. PCR products were separated by electrophoresis on 1.5% agarose gels, stained with GelRed (Biotium Inc., Hayward, CA, USA) and visualized with a UV transilluminator.

Data from questionnaire and laboratory results were entered and managed using spread sheet software, and data analysis was completed using Epiinfo 7® software (version 7.2.2) and STATA (verion 14.0, STATA Corp). A farm was considered NTS positive if at least one of the pooled poultry samples from that farm tested positive for PCR-confirmed Salmonella. The data was summarized using mean, frequencies, and proportions using bivariate analyses. Variables that were statistically significant at 95% confidence interval (α = 0.05) were subjected to backwards stepwise multivariate logistic regression model using STATA® to estimate the magnitude of association between biosecurity measures and the presence or absence of NTS on the participating farms.