Long-term effects of azithromycin mass administration to reduce childhood mortality on Streptococcus pneumoniae antimicrobial resistance: a population-based, cross-sectional, follow-up carriage survey

Background: Mass drug administration (MDA) programmes with the macrolide antibiotic azithromycin to reduce childhood mortality are expanding in Africa; however, concerns remain about the long-term effects of these programmes on antimicrobial resistance (AMR). We aimed to evaluate the persistence and spread of Streptococcus pneumoniae AMR following a community-randomised MDA trial. Methods: This population-based, cross-sectional, pneumococcal carriage survey was conducted in Mangochi, Malawi, 3·5 years after the MORDOR trial, in which communities received twice-yearly azithromycin or placebo for 2 years. Eligible participants in this carriage survey were children aged 4–9 years who lived in an azithromycin-treated or placebo-treated cluster during the MORDOR trial, and children aged 1–3 years who were resident in a cluster but born after the MORDOR trial ended. Nasopharyngeal swabs were collected from participants and analysed by whole genome sequencing; pneumococcal genomes obtained from a distant site in Malawi, in which MDA had not been conducted, were used as reference genomes. The primary outcome was the prevalence of S pneumoniae macrolide resistance, comparing placebo-treated and azithromycin-treated clusters at baseline, 6 months post-MDA, and 3·5 years post-MDA. Findings: Between April 8 and May 14, 2021, 924 children aged 1–9 years were screened, of whom 19 were excluded and 905 were recruited to the follow-up carriage survey: 452 from azithromycin-treated clusters and 453 from placebo-treated clusters of the MORDOR trial. We assessed 426 isolates from these participants (190 from azithromycin-treated clusters and 236 from placebo-treated clusters), as well as samples from the baseline of the MORDOR trial (164 isolates; 83 from azithromycin-treated clusters and 81 from placebo-treated clusters) and from 6 months post-MDA (223 isolates; 119 from azithromycin-treated clusters and 104 from placebo-treated clusters). In azithromycin-treated clusters, macrolide resistance increased from 21·7% (95% CI 14·2–31·7; 18 of 83 isolates) at baseline to 31·9% (24·2–40·8; 38 of 119 isolates) 6 months post-MDA and to 32·1% (25·9–39·0; 61 of 190 isolates) 3·5 years post-MDA. In placebo-treated clusters, resistance increased from 21·0% (13·5–31·1; 17 of 81 isolates) at baseline to 25·0% (17·7–34·1; 26 of 104 isolates) 6 months post-MDA and to 30·9% (25·4–37·1; 73 of 236 isolates) 3·5 years post-MDA. No significant differences were observed in odds ratios between treatment groups across the survey timepoints: 0·97 (95% CI 0·36–2·55) at baseline, 1·46 (0·67–3·17) at 6 months post-MDA, and 1·12 (0·66–1·91) at 3·5 years post-MDA. Macrolide resistance in the non-MDA site remained stable: 16·9% (95% CI 12·8–21·8; 45 of 267 isolates) at baseline, 16·5% (13·3–20·3; 70 of 424 isolates) at 6 months, and 16·5% (12·5–21·4; 44 of 267 isolates) at 2·5 years. Among children born into azithromycin-treated clusters after MDA, macrolide resistance was 36·0% (27·7–45·1; 41 of 114 children). Multidrug resistance to at least three antibiotic classes was significantly higher in azithromycin-treated (p=0·0015) and placebo-treated (p<0·0001) clusters than in the comparator population at 3·5 years post-MDA and was associated with integrative conjugative elements. Interpretation: Azithromycin MDA is associated with macrolide resistance that persists and potentially spreads to untreated populations. The co-existence of multidrug resistance and transmissible resistance on integrative conjugative elements in these populations is a public health concern. Careful monitoring of AMR is essential in areas where MDA is implemented. Funding: The Gates Foundation, the National Institute for Health and Care Research, and the Wellcome Trust.