Association between antimicrobial usage and resistance on commercial broiler and layer farms in Bangladesh

Antimicrobial resistance has emerged as a significant health problem worldwide, including in Bangladesh, where chickens are an important protein source for human nutrition. One of the factors accelerating the development of antimicrobial resistance is the inappropriate use of antimicrobials on commercial chicken farms. A cross-sectional study was conducted in 2019 on 140 commercial chicken farms in the Chattogram district of Bangladesh to investigate the association between antimicrobial use and resistance in Escherichia coli and Salmonella spp. cultured from cloacal swabs of chickens and from the poultry shed environment. All E. coli and Salmonella spp. isolates were resistant to multiple antimicrobial classes, including those categorized as “Highest Priority Critically Important Antimicrobials” for human medicine. Notably, resistance was observed in E. coli isolates from farms that did not use these antimicrobial classes in the current production cycle. For example, although quinolones were not used on 43.9% of E. coli positive farms, 95.7% of these farms had quinolone-resistant E. coli isolates. The results of the path analysis revealed that there was a “direct effect” of the frequency of antimicrobial usage on “high” resistance, with resistance increasing when antimicrobials were administered more frequently (β = 0.28, p = 0.002). There was a “direct effect” of the purpose of antimicrobial use on “low” resistance, with resistance marginally decreasing when antimicrobials were administered solely for therapeutic use (β = −0.17, p = 0.062), but increasing when they were used prophylactically. Overall, the study results could be used to educate farmers on better practices for antimicrobial administration, and to guide government agencies to update policies on antimicrobial use and resistance surveillance in the poultry sector of Bangladesh.


Introduction
Antimicrobial resistance in humans has been associated with the administration of antimicrobials in livestock (1).Antimicrobial resistant bacteria or their resistance genes can be transmitted to humans through direct contact with animals, consumption of animal products, or through environmental exposure (2).
Under the "One Health" paradigm, the World Organization for Animal Health (WOAH), the Food and Agriculture Organization (FAO) and the World Health Organization (WHO) have provided recommendations to reduce the global use of antimicrobials and limit the risk of global spread of resistant bacteria (15).A database monitoring the type and quantities of antimicrobial agents used in animals was launched in October 2015 (15).The FAO has also launched an Action Plan to minimize the impact of antimicrobial resistance (16).This plan aims to develop the capacity for monitoring the use of antimicrobials in food animals (16), while WHO has developed a Global Action Plan to reduce the non-judicious use of antimicrobials in animals and humans (17).
The Bangladesh government formulated a National Action Plan (2017-2022) to tackle antimicrobial resistance in humans, livestock and fisheries and strengthen infectious disease prevention and control (18).However, for implementation, reliable data on antimicrobial use and resistance in the livestock sector is required.Our previous research assessed antimicrobial usage on commercial chicken farms, identifying that "Medically Important Antimicrobials" were frequently purchased from feed and chick traders, and often used in the absence of clinical signs and without adherence to withholding periods (6).
Escherichia coli is a commensal bacterium in food producing animals and is commonly used to monitor the development of antimicrobial resistance and the risk of antimicrobial resistant bacteria transmission from animals to human (19).Colibacillosis, caused by avian pathogenic E. coli (APEC), is among the most prevalent bacterial infections in chickens and other poultry in Bangladesh (20).Chickens are continually exposed to E. coli through feces, water, dust, and their environment (20).Since E. coli are always present in the gastrointestinal tract of birds, they easily spread via feces (20).In a study conducted on a Bangladeshi chicken farm, 36 of 99 samples, comprising internal organs, feces, and air from inside poultry sheds, tested positive for the APEC-associated virulence genes fimC, iucD, and papC (21).Salmonella is a zoonotic foodborne bacterium and one of the major causes of human food borne disease (22).Both bacteria, which are often cultured from healthy animals at slaughter, are used worldwide as bioindicators for antimicrobial susceptibility.
In this study, we aimed to quantify the association between antimicrobial usage and susceptibility of E. coli and Salmonella spp.cultured from samples collected on commercial chicken farms in Bangladesh.In addition, we aimed to identify farm management factors associated with the increased antimicrobial resistance.

Field data collection
A cross-sectional study was conducted from February to May 2019 on 140 commercial layer and broiler farms from eight upazilas (subdistricts) in the Chattogram district of Bangladesh (Figure 1).
Data collection included sampling chickens and their environment and the administration of a structured questionnaire to obtain data on antimicrobial usage, and farm management practices implemented on these farms, as described previously (6).Briefly, commercial chicken farms were chosen from a sampling frame of 1748 commercial chicken farms using simple random sampling by applying the syntax RANDBETWEEN in Microsoft Excel (Microsoft Corporation, 2010).The number of farms per upazila was maintained using probability proportional to size sampling.
On each farm, the following strategy was followed to select the chicken shed of interest: i If the same number of antimicrobials was used in all sheds, the shed with the oldest chickens was selected.ii If the number of antimicrobials used across sheds differed, then the shed with the highest number of antimicrobials used was selected.
The structured questionnaire was validated by piloting it on five layer and five broiler farms.Additionally, photographs of antimicrobial packages were taken to cross-check the information provided.If available, photographs of the drug registration book kept on the farms were also taken to verify the data further.
Both cloacal and environmental swab samples were collected from the selected shed using the following strategy: i Cloacal swab samples were collected from five randomly selected healthy chickens and pooled into a 15 mL sterile falcon tube.ii Environmental swab samples were collected from the four corners and the middle of the chicken shed and pooled into a falcon tube containing buffered peptone water (BPW) (Neogen Corporation, 620 Lesher Place, Lansing MI 48912, United States).Only farmers actively engaged in the management of poultry farming were interviewed.One farmer from each farm was interviewed.Verbal consent was obtained from the farmers for the interview and collection of cloacal and environmental samples.Interviews were conducted by trained researchers from the Chattogram Veterinary and Animal Sciences University (CVASU).
All positive isolates were stored at −80°C in brain heart infusion broth with 50% glycerol (Neogen Corporation) (24).Identification was confirmed by Vitek (VITEK IVP, Inc., United States) at the Bangladesh Livestock Research Institute (BLRI).Details of the bacterial culture are summarized in the Supplementary material.

Antimicrobial susceptibility testing of
Escherichia coli and Salmonella spp.
All E. coli and Salmonella spp.underwent Kirby-Bauer disk diffusion antimicrobial susceptibility testing against 12 antimicrobials following the Clinical and Laboratory Standard Institute (CLSI) guidelines (25).Details of antimicrobial susceptibility testing are summarized in the Supplementary material.
These antimicrobials were selected because they were identified as being commonly used on chicken farms in Bangladesh (12,26).As resistance for one antimicrobial in a class often selects for resistance for other antimicrobials in that class (27), we categorized the antimicrobials into classes for data analysis.If any antimicrobial in the class was resistant, the bacteria was classified as resistant to that antimicrobial class.

Data analysis
A farm was classified as positive for E. coli or Salmonella spp.if either a cloacal or environmental sample tested positive for the bacteria.
E. coli and Salmonella spp.isolates identified as "resistant" or "intermediate" in the susceptibility testing were categorized as "resistant" for further data analysis.When a farm tested positive for E. coli or Salmonella spp., the cloacal or environmental sample with resistance to the largest number of antimicrobial classes determined the farms' level of resistance.The association between usage and non-usage of antimicrobial classes and, resistance and susceptibility to these antimicrobial classes was explored using scatter plots and the Fisher's exact test (28).
Multiclass resistance was defined as "resistance to antimicrobials of three or more different classes" (29).
Path analysis was used to examine potential relationships either in a "direct" or "indirect" way among a set of observed variables (30).A "direct" relationship exists when an exogenous variable (in regression, known as an independent variable) directly affects an endogenous variable (known as an outcome variable), while an "indirect" relationship occurs when an exogenous variable affects an endogenous variable through a mediator variable (30).The "indirect" relationship can involve one or more mediator variables between the exogenous variable and the endogenous variable (31).Path coefficients (β) are then estimated to quantify the relationships between variables along a hypothesized pathway (30).
We utilized path analysis to explore the relationship between published management factors related to antimicrobial usage that could directly or indirectly increase antimicrobial resistance in E. coli.Details on the management factors that could potentially increase antimicrobial usage in E. coli on commercial chicken farms are shown in Table 1.Path analysis was not conducted for Salmonella spp.because this bacterium was cultured from only a limited number of farms.A diagram showing the hypothesized direct and indirect farm management pathways that could impact antimicrobial resistance on these farms is shown in Figure 2.For the path analysis, we classified resistance into two categories: "high" resistance to more than 5 antimicrobial classes (coded as "1") and "low" resistance to up to 5 antimicrobial classes (coded as "0").
Variables with no significant "direct" and/or "indirect" effect at p < 0.05 on increasing the level of antimicrobial resistance were not included in the final path model.The fit of the path model was assessed using the Chi-square (χ) 2 statistic, with a p-value >0.05 indicating a good fit (35), the Root Mean Square Error of Approximations (RMSEA), with a value <0.05 indicating a good fit and a value up to 0.08 indicating an acceptable fit (36), the Comparative Fit Index (CFI), with a value >0.95 indicating very good fit and ≥ 0.90 an acceptable fit (35), and the Standard Root Mean Square Residuals (SRMR) with a value ≤0.05 indicating a close-fitting model and a value between 0.05 up to 0.10 an acceptable fit (36).

Association between antimicrobial usage and antimicrobial resistance
Supplementary Table S2 shows the frequencies of farms using antimicrobial classes on E. coli-positive and Salmonella spp.-positive commercial poultry farms, while the frequency of E. coli-positive and resistant farms where antimicrobials classes were not used in the current production cycle are presented in Supplementary Table S3.The relationship of usage and resistance pattern of quinolones, macrolides, polymyxins, aminoglycosides, beta lactams, sulphonamides, and tetracyclines is presented in Figure 3.
Our results highlight that E. coli-positive farms were frequently resistant to antimicrobial classes even when these classes were not reported to have been used during the studied production cycle.Indeed, quinolones, classified among the "Highest Priority Critically Important Antimicrobials" (HPCIAs) (37) were not reported being used on 43.9% (47/107) of the farms where E. coli was cultured, but 95.7% (45/47) of these non-user farms had quinolone-resistant

Predictors Categories Results
Farm type Layer Layer farms have been shown to have a higher prevalence of antimicrobial resistance (11).Broiler Flock size ≤2,500 birds Larger flock sizes have been associated with a higher prevalence of antimicrobial resistance (11).>2,500 Farmers' decision to follow advice provided by feed and chick traders on the administration of antimicrobials Not or rarely followed Farmers who purchased antimicrobials from feed and chick traders were more likely to misuse antimicrobials (8) and it is hypothesized that these farms may have a higher prevalence of antimicrobial resistance.
Frequently or always followed Purpose of using antimicrobials Prophylactic with or without therapeutic Prophylactic administration of antimicrobials contributes to a higher prevalence of antimicrobial resistance (32).Only therapeutic

Once
Frequent administration of antimicrobials contributes to a higher prevalence of antimicrobial resistance (33).Twice Three times Adherence to the withholding period No Non-adherence to withdrawal periods might be associated with higher prevalence of antimicrobial resistance (34).Yes

Association between farm management factors and antimicrobial resistance
Descriptive statistics of management procedures used on commercial chicken farms that were considered as predictors in the path analysis are summarized in Table 3.
The results of the path analysis (Figure 4) indicated that there was a "direct effect" of frequency of antimicrobial usage on higher resistance, with resistance increasing with more frequent administration of antimicrobials (β = 0.28, p = 0.002).Conversely, there was a "direct effect" of purpose of antimicrobial usage on lower resistance, with resistance marginally declining when antimicrobials were only administered therapeutically (β = −0.17,p = 0.062; or increasing when antimicrobials were administered prophylactically).There was an "indirect effect" of flock size on resistance, which decreased in larger flocks (β = −0.12,p = 0.005).Having larger flock sizes was associated with antimicrobial usage for therapeutic purposes only (β = 0.23, p = 0.009), while in smaller flocks antimicrobial for prophylactic purposes was more common (i.e. less antimicrobial usage for therapeutic purposes only β = −0.32,p < 0.001).
The other management factors were not significant at p < 0.05 in the final model.The model fitted the data well, as indicated by χ 2 = 2.347 (p = 0.309), RMSEA = 0.040, CFI = 0.987 and SRMR = 0.042.

Discussion
The present study provides unique insights into the relationship between antimicrobial usage and resistance in E. coli and Salmonella spp.isolates cultured from commercial chicken farms in Bangladesh.It should be noted that isolates were cultured from cloacal swabs of apparently healthy birds and from their environment.
In our study, Salmonella was detected in 27.3% of cloacal swabs.In other studies in Bangladesh, Salmonella spp.have been detected in 27-48% of cloacal swabs in poultry.Prevalence may differ due to differences in isolation and identification procedures, but also may be influenced by type of poultry, flock size and factors such biosecurity, and hygiene (37,40).Some antimicrobials that were administered to chickens are classified as HPCIAs for human health, such as quinolones and polymyxins (33).Resistance to these antimicrobials was found on the studied farms.These antimicrobials are not recommended for use in animals without culture and antimicrobial susceptibility testing (34).Establishing a direct link between antimicrobial usage and resistance is challenging as multiple pathways exist and resistance might develop over many years.Nevertheless, this study identified that antimicrobial resistance patterns do not only depend on antimicrobial usage, as high levels of resistance were found on farms not reporting the use of these antimicrobials in the current production cycle.Furthermore, repeated administration of antimicrobials within a single day might have contributed to increased resistance.The impact of antimicrobial overuse on the development of antimicrobial resistance has been previously described (44) and selection pressure, cross-resistance and/ or co-resistance might have contributed to the occurrence of antimicrobial resistance (32).
The presence of resistant bacteria in environmental samples is concerning as it represents a risk of contamination for areas surrounding poultry sheds.This contamination can occur through run off from poultry sheds, which may, in turn, contaminate water sources and the environment (45).Resistant bacteria can transfer resistance genes between different strains and across species (46).Such resistant commensal or pathogenic bacteria transfer can then result in resistant infections in humans, leading to treatment failure, higher medical costs, prolonged hospital stays, and increased mortality (47).
Antimicrobial agents may remain in the environment for years (48).Poor poultry husbandry practices, including inadequate disinfection, waste disposal into the environment, and poor infection control practices may contribute to the development of an environmental reservoir of resistant bacteria over time (49).The presence of resistant bacteria cultured from cloacal samples is also concerning because they may contaminate food products, resulting in foodborne illness (50).Final path analysis results for farm management factors that had an effect on increasing resistance on commercial chicken farms where E. coli was cultured. 10.3389/fvets.2024.1435111 Frontiers in Veterinary 08 frontiersin.org Resistance was higher on small scale commercial chicken farms.This could be attributed to lower biosecurity standards and management practices, including inadequate housing facilities, or inadequate disinfection of water sources as reported by FAO (51).Our previous investigations into the biosecurity practices and antimicrobial usage on these farms revealed that inadequate biosecurity was more prevalent on farms with smaller flock sizes (<2,500 birds) compared to those with larger flocks (≥2,500 birds); and usage of antimicrobials was more common on small scale farms (6,9).Small-scale farmers often lack formal training in biosecurity (13) and may be advised by representatives of pharmaceutical companies (8), and unqualified veterinary care givers who offer more affordable animal care (52).Although it is commonly reported that farmers are influenced by the advice of feed and chick traders (8), our study did not find evidence supporting this.Farmers rearing smaller flocks may more frequently administer antimicrobials prophylactically compared to farmers with larger flocks, to reduce the potential disease risk associated with poorer biosecurity.
Farms where antimicrobials were used prophylactically had higher levels of resistance in E. coli.Mass prophylactic administration of antimicrobials might result in lower adaptive immunity in chicken flocks (53), and their effectiveness in preventing poultry diseases may diminish over time (54), Thus, the misuse of antimicrobials in broiler (12) and layer (13) farms in Bangladesh may select for antimicrobial resistance in E. coli and potentially increase resistance.
We have used path analysis to investigate the factors contributing to high antimicrobial resistance.This approach enabled us to clarify complex interrelationships among variables, highlighting the most important pathways that were associated with the outcome (55).It allowed to display the direction and magnitude of both "direct" and "indirect" relationships between farm management factors and their impact on antimicrobial resistance in E. coli.However, although path analysis was used to analyze the hypothesized pathways, it is primarily based on correlations and cannot be used to prove causality (55).
The data collection in this study had several limitations.Firstly, while we could identify if an antimicrobial class was used during the current production cycle, we could not determine the frequency, dose or the duration of administration.Therefore, actual antimicrobial usage may have been underestimated.Secondly, some farmers might not have accurately remembered which antimicrobials they had administered.However, by focusing on antimicrobial use in the current production cycle, we aimed to minimize potential recall bias.Thirdly, antimicrobial susceptibility testing was performed using the disk diffusion method, which might not be optimal for detecting polymyxins (i.e.colistin in this study) resistance (25,56).Colistin is a cationic, multicomponent, lipopeptide antimicrobial agent that diffuses slowly in antimicrobial susceptibility testing media, therefore the resulting zone of inhibitions tends to be inaccurate; and may bias the interpretation (57).Minimum inhibitory concentration (MIC) methods are preferred for colistin (25, 56), however, were not feasible to conduct in the laboratory in Bangladesh.In future studies, the use of MIC methods would be recommended for all antimicrobials, especially for colistin.However, MIC methods for colistin may also be affected by its cationic properties (58).Finally, the lack of veterinary breakpoints for most of the antimicrobials tested (25, 56) is a general limitation in veterinary studies, as breakpoints need to be extrapolated from humans or other species.
Overall, this study provides important baseline data that could be used to develop recommendations aimed at reducing antimicrobial usage in chicken production.Such recommendations may include initiatives for training farmers and raising awareness regarding the appropriate use of antimicrobials.Basic training in poultry pathology and necropsy would allow farmers to make informed decisions when selecting antimicrobials based on gross pathology.Routine availability of culture and antimicrobial susceptibility testing, and molecular techniques to identify and characterize commensal and pathogenic organisms, along with residue testing performed by governmental authorities, could also form integral components of an effective antimicrobial stewardship program for Bangladesh.The author(s) declare that financial support was received for the research, authorship, and/or publication of this article.This field data collection was supported by the BALZAC research program 'Behavioral adaptations in live poultry trading and farming systems and zoonoses control in Bangladesh' (BB/L018993/1), a joint research initiative between the Biotechnology and Biological Sciences Research Council, the Defense Science and Technology Laboratory, the Department for International Development, the Economic and Social Sciences Research Council, the Medical Research Council and the Samples were kept in an insulated box containing ice packs and transported to the Poultry Research and Training Center (PRTC) laboratory in Chattogram, Bangladesh within 4-6 h.The samples were stored at −20°C until analysis.Bacterial culture and antimicrobial susceptibility testing were conducted between June and September 2019.

FIGURE 1
FIGURE 1 Commercial layer and broiler farms in the Chattogram district of Bangladesh.The eight upazilas sampled are highlighted in purple.Pie charts display the number of farms by farm type (broiler farm, layer farm), with the size of the pie charts representing the number of farms sampled per upazila.
identified with resistant and intermediate susceptibility.

FIGURE 3
FIGURE 3 Usage and resistance pattern of antimicrobial classes on commercial chicken farms in the Chattogram district of Bangladesh where E. coli was cultured.Each data point represents a E. coli positive farm.The data points are jittered to avoid overlapping.

TABLE 1
Management factors related to antimicrobial usage that potentially could have an effect on increasing resistance in E. coli on commercial chicken farms.

TABLE 2
Results of antimicrobial susceptibility testing of E. coli and Salmonella spp.isolates cultured from cloacal and environmental samples collected on commercial chicken farms in the Chattogram district of Bangladesh.

TABLE 3
Descriptive statistics of the management factors related to antimicrobial usage which potentially have an effect on increasing resistance in E. coli on commercial chicken farms.
Natural Environment Research Council.corresponding author of this publication was supported by the Australia Awards Scholarship.