Twice-yearly azithromycin held macrolide resistance within the prespecified margin
In rural Mali, a population-level experiment with azithromycin mass drug administration did not produce a broad antimicrobial resistance surge when the drug was given twice yearly to infants aged 1 to 11 months. In the LAKANA sub-study, the comparison against placebo met the trial’s noninferiority threshold for macrolide resistance in both Streptococcus pneumoniae and Escherichia coli, using a prespecified margin of plus 10 percentage points. The signal was more cautious in the quarterly arm, where the confidence intervals could not exclude a modest increase in resistance, especially in S. pneumoniae and E. coli among younger children.
The trial design mattered clinically because the question was never only whether azithromycin changes resistance in the treated child. A mass antibiotic strategy can reshape carriage and selection pressure across a village, and that ecological effect is often what determines whether a public health programme remains sustainable. Here, investigators sampled children repeatedly across 59 villages, pairing nasopharyngeal swabs with rectal swabs at baseline and at 12, 24, and 36 months after village enrollment. The approach captured both respiratory and gut reservoirs of resistance, two compartments that can behave differently under macrolide exposure.
The baseline resistance burden was already uneven across groups, which makes the later analysis more important than any single starting point. Among children aged 4 to 14 months, macrolide resistance in S. pneumoniae measured 16.6% in the placebo group, 15.5% in the twice-yearly group, and 11.1% in the quarterly group at baseline. For E. coli, the corresponding baseline prevalences were 5.0%, 5.0%, and 0.0%. By 36 months, resistance in S. pneumoniae stood at 13.4% in placebo villages, 14.5% in twice-yearly villages, and 17.2% in quarterly villages. In E. coli, the corresponding values were 5.1%, 6.4%, and 9.1%.
The absolute differences against placebo remained small in the twice-yearly group. For S. pneumoniae, the difference was 1.0 percentage point with a 95% confidence interval from minus 4.7 to 6.8. For E. coli, the difference was 1.3 percentage points with a 95% confidence interval from minus 2.4 to 4.9. Those estimates stay inside the trial’s noninferiority boundary, which supports the conclusion that twice-yearly exposure did not measurably worsen macrolide resistance in the tested isolates over the 2-year intervention period.
Quarterly dosing left a residual resistance concern in young children
Quarterly treatment never crossed into a definitive harm signal, but it did not fully clear the bar either. The difference versus placebo reached 3.8 percentage points for S. pneumoniae and 4.0 percentage points for E. coli, with 95% confidence intervals extending up to 10.9 and 8.7 percentage points, respectively. In plain clinical terms, the data leave room for a small dose-dependent rise in resistance when azithromycin is delivered four times per year. The study was not large enough to rule out that possibility, especially for pneumococcal resistance in the youngest children.
Among untreated older children aged 49 to 59 months, no marked between-group differences appeared at baseline or at 24 months. Those children never received azithromycin, so they served as a practical readout of spillover in the surrounding community. The absence of obvious resistance spillover is important because antibiotic pressure in a village can theoretically select for resistant organisms in contacts who never swallow the drug themselves. In this setting, the untreated cohort did not show a clear ecological penalty from the twice-yearly strategy.
The findings fit the biology of macrolide selection pressure. Azithromycin concentrates in tissues and can persist long enough to suppress susceptible organisms while allowing resistant strains to expand. More frequent dosing increases the number of selection windows and can widen the opportunity for resistant clones to establish themselves in the nasopharynx or gut. The current analysis suggests that two rounds per year may be enough to avoid a detectable resistance penalty in this age-restricted programme, while four rounds per year may push closer to the threshold where small shifts begin to emerge.
The trial also looked beyond macrolides to nonmacrolide antibiotic classes, and no consistent increase was seen there. That matters because resistance can cluster through linked genetic mechanisms, not just within a single drug class. Even so, the absence of a broader signal does not remove the need for surveillance. Phenotypic susceptibility testing is only as informative as the sampling frame, and the study’s cross-sectional design cannot prove that low-level shifts would remain absent under longer exposure or in settings with different baseline resistance patterns.
What the LAKANA sub-study can and cannot support
The strongest conclusion is narrow and clinically useful: in this Malian setting, twice-yearly azithromycin administration to infants aged 1 to 11 months for 2 years did not increase macrolide resistance in the tested isolates and met the study’s noninferiority criterion. Quarterly administration deserves more caution. The data do not prove harm, but they do not exclude a modest increase in S. pneumoniae and E. coli resistance either.
Generalization beyond this programme should stay restrained. The intervention targeted infants, not the broader 1- to 59-month age band used in some other mass administration strategies. More frequent treatment, larger coverage, and different background resistance levels could all shift the balance. A programme that treats a larger share of the population creates more antibiotic exposure and more opportunity for selection, so these results cannot simply be lifted into settings with broader age coverage.
The most actionable message is operational rather than theoretical. Any future mass azithromycin programme, especially one aiming for mortality reduction, needs continuous antimicrobial resistance monitoring built into implementation from the start. Without serial surveillance, small changes in macrolide susceptibility can pass unnoticed until they become entrenched in both treated children and the community around them. The LAKANA data provide reassurance for a twice-yearly infant strategy in this setting, but they also leave an open question for more intensive dosing schedules and for places where resistant pathogens are already more common.