EpiFAMU links farm management with poultry antimicrobial use in Zambia

A newly developed epidemiological tool called EpiFAMU can systematically record management factors associated with antimicrobial use on small- and medium-sized broiler farms, according to an exploratory study in Lusaka, Zambia. Researchers assessed the tool across 45 farms and reduced an initial set of 175 potential variables to 101 indicators covering nutrition, veterinary practices, biosecurity, vaccination, brooding, housing, equipment, personnel, markets, extension services, antimicrobial resistance, antimicrobial use, mortality and public health. Six management indicators showed statistically significant associations with treatment frequency, while two farmer characteristics also differed significantly in the antimicrobial-use patterns recorded. The findings demonstrate the feasibility of combining farm-management surveillance with antimicrobial-use monitoring, but they do not establish that any management practice causes higher treatment frequency or changes resistance outcomes.

The research team developed EpiFAMU around a nine-point framework used by the Food and Agriculture Organization of the United Nations, modifying its emphasis from checking questionnaire length to tracking respondent burden. The questionnaire was designed for in-person completion with poultry owners, managers or supervisors and contained closed-ended questions, open-ended questions and options for open responses. A pilot involving five respondents revealed that a full interview took two to three hours, that some participants struggled to recall exact vaccination dates and antibiotic purchases, and that certain biosecurity questions were interpreted inconsistently. Investigators therefore split compound biosecurity items into separate questions and converted some open-ended variables into categorical choices before using Epi-Collect and spreadsheets for data entry.

The study enrolled 45 broiler farms in Lusaka district, where farms had between 100 and 5,000 birds. Recruitment began with 10 farms identified through veterinary and market contacts, and each initial participant referred one to four additional farms, producing a snowball sample. Most farms were intensive, non-commercial or semi-commercial operations serving local markets, and 75.6% were in urban areas. The sample represented men and women farmers in similar proportions, and the median farming experience was five years. The median flock size was 400 birds, and the median cropping age was 42 days. Because Zambia lacked a population registry of broiler farms that would permit probability-based random sampling, the investigators could not rule out selection bias or ensure that the farms reflected all production systems in the country.

Researchers used several statistical checks to construct the final instrument. Variables that produced the same answer at more than 42 farms were removed because they provided little discriminatory information. Cramér’s V was then used to identify pairs of variables with associations above 0.3, and one member of each strongly associated pair was removed according to its relevance to antimicrobial use or infection prevention and control, its biological plausibility and its potential for action. This process excluded 32 variables with little or no variation and a further 42 as redundant proxies, leaving 101 variables. A binary score records whether a farm follows the recommended practice in each applicable item. EpiFAMU is available in paper and electronic formats and can be used alone or alongside the Veterinary Consumption of Antibiotics–International Documentation system, which supplies quantitative antimicrobial-use data.

Antimicrobial exposure was represented by treatment frequency calculated for each production run and day and then averaged across runs for each farm. Two-sample Wilcoxon tests compared treatment-frequency distributions between groups defined by farmer characteristics or by whether farms followed specific management practices. Only six of the 101 epidemiological variables reached the study’s 5% significance threshold. Farms using vitamin supplementation had a mean treatment frequency of 0.18 per day, compared with 0.06 among farms not doing so; administering vitamins before vaccination was associated with a mean frequency of 0.21 versus 0.11; and using vitamins, prophylactic products or sugar to improve chick survival was associated with 0.18 versus 0.05. These comparisons were observational, however, and the investigators noted that commercially sold vitamin products might sometimes contain antibiotics, while respondents may also have confused supplements with antimicrobial products.

Three further significant associations emerged from comparisons of the number of water-dispenser points, litter thickness and attitudes toward antibiotics. The study defined an adequate supply as at least one bell drinker per 100 older chickens. Farms that did not meet this threshold had a mean treatment frequency of 0.33 per day, versus 0.16 among farms that did. Farms judged to have adequate litter thickness recorded a mean frequency of 0.32, compared with 0.14 among the other farms. Farmers who agreed that maximum production could not be achieved without antibiotics had a mean frequency of 0.20, while those who did not agree had a mean of 0.11. The researchers said the drinker result might be consistent with stress or reduced water access during hot periods, which could encourage treatment, but they stressed that direct evidence connecting drinker numbers to antimicrobial use remains limited.

Farmer characteristics also revealed differences. Unemployed farmers had a mean treatment frequency of 0.19 per day, compared with 0.10 among employed farmers, while farms selling both live and dressed chickens had a mean frequency of 0.11, compared with 0.24 among farms selling live chickens only. The two farms selling dressed chickens only had a mean frequency of 0.31. The authors proposed that poultry may be the main income source for unemployed participants, potentially limiting their ability to invest in hygiene or veterinary services, and suggested that market contact might create biosecurity risks. Both explanations remain hypotheses: the observational design did not measure the mechanisms behind the statistical differences, and the dressed-only group contained only two farms.

Several results ran counter to expectations based on prior research. Vitamin use, pre-vaccination vitamins, survival measures and adequate litter had previously been associated in other settings with better health and lower antimicrobial use, yet in this sample each was linked to higher treatment frequency. The researchers suggested that farms facing greater disease pressure might be more likely to adopt recommended practices while still treating more animals, whereas farms with fewer problems might invest less in formal practices and consequently use fewer antimicrobials. They also noted that thicker litter can preserve pathogens in the environment. These interpretations are plausible explanations rather than demonstrated mechanisms, and the researchers emphasized that direct studies of such counterintuitive patterns in poultry production remain needed.

The study has important constraints. In addition to the non-random sample and the very small size of some comparison groups, questionnaire responses were vulnerable to recall bias and social-desirability bias because respondents were asked to report whether they followed recommended practices. More than 100 variables were tested without adjustment for multiple comparisons, making some of the eight reported statistically significant findings potentially attributable to chance. EpiFAMU’s AMR domain assessed awareness only: no direct farm-level resistance outcomes were measured, because most participating small and medium-sized farms did not routinely conduct on-farm AMR testing. The results are also restricted to small- and medium-sized broiler farms in Lusaka and may not represent other livestock sectors, production systems or regions of Zambia.

Within those limits, the study supports EpiFAMU as a practical, low-cost method for collecting comparable information about farm management alongside antimicrobial-use surveillance. Its domain scores could support benchmarking at farm, district, country or regional levels, although that scoring approach still requires evaluation against antimicrobial-use outcomes. The authors propose external validation in larger, stratified samples, direct resistance surveillance through periodic faecal or litter sampling and culture or molecular identification, and longitudinal analysis of whether changes in EpiFAMU scores coincide with reductions in antimicrobial use or resistance. The current study therefore identifies useful signals for further testing rather than a validated predictive model or evidence that changing any single practice would reduce antimicrobial use.

Subject of Research: Development and exploratory application of a farm-level monitoring tool for antimicrobial use in Zambian broiler production

Article Title: EpiFAMU: an epidemiological tool for monitoring farm-level factors essential to antimicrobial use surveillance and infection prevention and control programs in poultry production systems in Zambia—development and application

Article References: Silwamba, I., Kirse, A., Mainda, G., Mudenda, N., Mtui-Malamsha, N. J., Hartmann, M., Brogden, S., Muma, J. B., & Kreienbrock, L. (2026). EpiFAMU: an epidemiological tool for monitoring farm-level factors essential to antimicrobial use surveillance and infection prevention and control programs in poultry production systems in Zambia—development and application. Frontiers in Veterinary Science, 13, Article 1963471. https://doi.org/10.3389/fvets.2026.1963471

Image Credits: AI Generated

DOI: 10.3389/fvets.2026.1963471

Keywords: EpiFAMU, antimicrobial use, broiler farms, poultry production, Zambia, farm management, infection prevention, antimicrobial resistance, treatment frequency, epidemiological surveillance, epidemiological, tool