Search

• District-wide schistosomiasis interventions has resulted in mistreatment in many populations. • There is currently no standardized sub-district survey protocol available for schistosomiasis. • This publication estimates three schools/sub-district provides precise prevalence estimates. • Thirty children sampled per school was also sufficient in reliably detecting schistosomiasis. Schistosomiasis control programs typically launch with district-level, school-based preventive chemotherapy (PC). Recent World Health Organization recommendations are to shift to community-wide treatment where schistosomiasis prevalence is >10%. Simultaneously there is a push to move to sub-district PC to prioritize communities in need of treatment and alleviate the pressure on global praziquantel need, but few countries have sub-district prevalence data and no guidelines on how to collect this information. Data collected from 57,161 school-aged children (SAC) across six countries (Burkina Faso, Ghana, Mali, Senegal, Sierra Leone, and Togo) to generate spatially realistic gold standard datasets that were used to evaluate different numbers of schools per sub-district (1-10) and district (5–30), number of SAC sampled per school (10–50), on accuracy of prevalence estimates. Sampling fewer children in more schools maximized accuracy of prevalence at the sub-district and district level. Surveying three schools per sub-district or 15 schools per district gave precise prevalence estimates. Increasing the number of SAC beyond 30 per school led to negligible improvements in reliably detecting schistosomiasis. Failure to detect schistosomiasis occurred more frequently in low (1–10%) prevalence and larger districts/sub-districts. This study provides guidelines for evaluating sub-district schistosomiasis in a range of transmission settings. Among two-stage cluster surveys for schistosomiasis, our simulations show surveying three schools per sub-district and 20-30 SAC per school optimized cost-efficiency and minimized risk of mistreatment. Population size and endemicity influenced survey estimates, with the probability of misclassification being greater as populations increased or prevalence decreased. [ABSTRACT FROM AUTHOR]