Impact of macrolide resistance on azithromycin for prevention of rehospitalization or death among children discharged from hospitals in Western Kenya.

BACKGROUND The Toto Bora trial tested whether a 5-day course of azithromycin reduced the risk of re-hospitalization or death in the 6 months following hospitalization among Kenyan children and found no overall benefit. We hypothesized that macrolide resistance in gut microbes could modify azithromycin's effect. METHODS From June 2016 to November 2019, Kenyan children aged 1-59 months were enrolled at hospital discharge and randomized to azithromycin or placebo. DNA from fecal samples and E. coli isolates was analyzed for common macrolide resistance genes. Cox proportional hazards regression models, including interaction terms between randomization arm and individual macrolide-resistance genes, were used to analyze time to rehospitalization or death, with Bonferroni correction applied to account for multiple comparisons. RESULTS Among 1,393 children tested, 94.7% had at least one macrolide-resistance gene in their fecal DNA at hospital discharge; most commonly mph(A) (68.6%; 955/1393), followed by msr(D) (67.3%; (937/1393), and erm(B) (60.7%; 846/1393). Mef(A) (23.7%; 330/1393) was the only macrolide-resistance gene that modified azithromycin's effect on re-hospitalization or death (interaction p-value=0.008). In children without the mef(A) gene, azithromycin reduced the hazard of rehospitalization or death by a third (HR=0.66, 95%CI: 0.45-0.99) whereas among children with the mef(A) gene, there was a higher risk in those randomized to azithromycin (HR=2.72, 95%CI: 1.21-6.09). The effect size of azithromycin's impact on mortality and rehospitalization as separate outcomes in children with and without mef(A) were consistent but underpowered. INTERPRETATION Macrolide resistance in the gut microbiome may influence the efficacy of azithromycin in children discharged from the hospital.