Sample size considerations

The sample size for the morbidity communities was based off the primary morbidity outcome, which was macrolide resistance in Streptococcus pneumoniae. We assumed 12% baseline resistance (based on previous studies) and an ICC of approximately 0.051 (based on the Trachoma Elimination Follow-up study [15]). We estimated that inclusion of 30 villages (15 per arm) and 10 samples per community would yield approximately 80% power to detect a difference in prevalence of resistance of 18% (e.g., 12% versus 30%) assuming 80% carriage of S. pneumoniae. For the infant adverse event survey, the sample size was limited by the number of 1 to 5 month old children residing in the 30 communities during the study period.