Small-sample antimicrobial resistance surveillance using lot quality assurance sampling in Uganda, Malawi, and Zambia.

OBJECTIVES

We evaluated lot quality assurance sampling (LQAS) adapted for antimicrobial resistance (AMR) surveillance (AMR-LQAS), a small-sample approach designed to classify district AMR prevalence against a clinically meaningful resistance threshold in Uganda, Malawi, and Zambia.

METHODS

The study was conducted across nine districts/locations, with each country including one urban, one peri-urban, and one rural district/location. Upper and lower resistance thresholds were 20% and 5%, respectively, with predefined misclassification probabilities (α = 0.05, β = 0.10), yielding a minimum sample size of 44 per district. In Uganda, classifications were compared with larger probability samples (n ≥ 240 in each of three districts). Organisms assessed included extended-spectrum beta-lactamase (ESBL)-producing Escherichia coli and Klebsiella pneumoniae, and methicillin-resistant Staphylococcus aureus. Antibiotics assessed were those routinely used in the study countries.

RESULTS

In Uganda, concordance between AMR-LQAS and probability-sample classifications was nine of 12 organism-district comparisons. Across antibiotic-district comparisons, concordance was 46/48, resulting in overall reliability of 92% (55/60 comparisons). Resistance to ciprofloxacin, gentamicin, and trimethoprim-sulfamethoxazole in ESBL-producing E. coli exceeded the 20% threshold in all districts, whereas meropenem resistance exceeded the threshold in only one district. Zambia had fewer antibiotics below resistance thresholds than Uganda and Malawi.

CONCLUSION

AMR-LQAS reliably classified district-level resistance using substantially smaller sample sizes than conventional prevalence surveys, generating actionable subnational information to support antibiotic stewardship and surveillance in low-resource settings.