LAKANA randomized villages in Mali to placebo, twice-yearly azithromycin, or quarterly azithromycin; a nested AMR sub-study sampled 59 nonurban villages for repeated cross-sectional surveillance. Sampling targeted children aged 4–14 months (to capture those recently exposed to MDA) and untreated contacts aged 49–59 months. Nasopharyngeal swabs were used to isolate S. pneumoniae and rectal swabs to recover E. coli. Samples were taken at baseline and at 12, 24, and 36 months; all samples at the prespecified primary endpoints (36 months for younger children, 24 months for older contacts) were analyzed without selection.
Laboratory methods followed standardized procedures. Nasopharyngeal swabs were cultured in enriched broth and plated for pneumococci identification using optochin and bile solubility criteria; erythromycin discs provided a macrolide resistance screen with E-test MIC confirmation for erythromycin-resistant isolates. Rectal swabs were cultured onto chromogenic agar to pick presumptive E. coli, confirmed by indole testing; azithromycin MICs used E-tests and, because EUCAST lacks E. coli azithromycin breakpoints, investigators applied MIC thresholds from prior publications (primary threshold ≥32 mg/L, with ≥16 mg/L used for sensitivity analyses). Disk diffusion followed EUCAST v12.0 for most agents, with stated exceptions. Quality control strains were used for both organisms.
The prespecified primary outcome compared prevalence of phenotypic macrolide resistance in isolates from children aged 4–14 months at 36 months across the three arms. A +10 percentage-point noninferiority margin had been defined a priori. At baseline, macrolide resistance prevalences varied across arms for both organisms. By 36 months, reported prevalences for S. pneumoniae were 13.4% in the placebo arm, 14.5% in the twice-yearly azithromycin arm, and 17.2% in the quarterly azithromycin arm. For E. coli, corresponding prevalences at 36 months were 5.1% (placebo), 6.4% (twice-yearly), and 9.1% (quarterly).
Investigators calculated differences between placebo and twice-yearly arms that were small and compatible with the pre-specified +10 percentage-point noninferiority margin for both organisms. The quarterly azithromycin arm showed higher point estimates of macrolide resistance relative to placebo, and investigators reported that a modest, dose-dependent increase could not be excluded for S. pneumoniae and E. coli. The report explicitly states the trial may not have had sufficient size to rule out a small increase in macrolide resistance in the quarterly-dose villages.
The sub-study also evaluated untreated contacts aged 49–59 months to assess potential spillover of resistance beyond treated infants. Samples in this older group were obtained at baseline and at 12 and 24 months. Data indicated no marked intergroup differences in macrolide resistance at baseline or at the 24-month endpoint. Investigators therefore did not detect evidence of clear resistance spillover from the infant-targeted MDA to untreated older children within the sampled villages and time frame.
Beyond macrolides, the study assessed resistance to other antibiotic classes using disk diffusion for agents including ampicillin, ciprofloxacin, co-trimoxazole, ceftriaxone, meropenem, gentamicin, and others depending on organism. The authors reported no consistent increase in resistance to nonmacrolide antibiotics across the monitoring period. Annual monitoring over the three years therefore did not reveal a reproducible signal of broad-spectrum cross-class resistance linked to the infant-targeted azithromycin regimens in this setting.
The data indicate that twice-yearly azithromycin administered to infants aged 1–11 months over two years met the predefined noninferiority margin for macrolide resistance in the sampled isolates and did not show detectable resistance spillover to untreated older children in this cohort. Quarterly dosing produced higher point estimates of macrolide resistance in both S. pneumoniae and E. coli, and the investigators could not exclude a modest dose-dependent increase.
Limitations were clearly stated. The sub-study was nested in a subset of LAKANA villages selected for accessibility; sample sizes—particularly for detecting small differences in the quarterly arm—may have been insufficient. The authors cautioned against extrapolating these findings to programs treating broader age ranges (for example children aged 1–59 months) or to higher-coverage campaigns, where greater antibiotic exposure could change resistance dynamics.
The study team recommended that any MDA program, especially those extending treatment to larger population shares or using different dosing schedules, include continuous AMR surveillance to monitor emergence and spread of resistance. The report also noted that the findings are context-specific and depend on baseline resistance levels, the proportion of the population treated, and the regimen used—factors that should guide policy and further data collection.