Background
High pathogenicity avian influenza continues to cause large epizootics in poultry and other animal populations, with ongoing concern about zoonotic spillover. We aimed to estimate the association between specific hygiene, personal protective equipment, and biosecurity practices and farm-level avian influenza infection in commercial poultry farms and backyard poultry settings.
Methods
We conducted a systematic review and meta-analysis. MEDLINE, Embase, Scopus, CINAHL, CENTRAL, CABI, Web of Science, and ClinicalTrials.gov were searched from inception to Dec 11, 2025. Eligible studies were empirical studies that compared use versus non-use, or different adherence levels, of hygiene, personal protective equipment, or biosecurity practices and reported farm-level avian influenza infection status. Screening, data extraction and risk of bias assessment was completed by 2 independent reviewers. The main outcome was farm-level avian influenza infection. Odds ratios were pooled using random-effects models. The protocol was registered with PROSPERO, CRD420251174630.
Findings
The search identified 39 observational studies, including 28 case-control, nine cross-sectional, and two prospective cohort studies. Reduced odds of infection were observed for disinfection of equipment and premises overall (OR 0.37, 95% CI 0.14–0.99), human entry/exit decontamination stations (0.61, 0.40–0.94), personal hygiene practices (0.52, 0.35–0.76), contact-pathway PPE (0.67, 0.51–0.87), quarantine/cross-farm contamination control (0.49, 0.34–0.71), and biosecurity information resources (0.26, 0.07–0.91). Overall risk of bias was high in eight studies, some concerns in 24, and low in seven. Between-study heterogeneity varied across analyses. Certainty of evidence ranged from low to moderate.
Interpretation
Practical, contact-focused biosecurity practices were most consistently associated with lower farm-level avian influenza infection odds. Although the observational evidence does not establish causal effectiveness and may be affected by residual confounding, the findings support prioritising feasible measures that interrupt direct transmission pathways while accounting for implementation quality, local context, and behavioural feasibility.