The impact of alternative model assumptions on both the required sample size and the total survey costs to reliably stop preventive chemotherapy.

This table presents the sample size (nschools×nchildren), the decision cut-off c and the associated total survey costs (Ctot) for different scenarios of parameterizing the aggregation of infections within children (kk), the clinical specificity of Kato-Katz thick smear and the maximum allowed risk of incorrect decision-making for the different soil-transmitted helminths species separately. As a reference we assumed that kk varies as a function of school mean eggs per gram of stool (EPG) and that Kato-Katz thick smears has a perfect specificity. We allowed for a risk of undertreatment (Eundertreat) equal to 0.05 and risk of overtreatment (Eovertreat) equal to 0.25. Also, the fixed aggregation parameter (Fixed kk), which captures the variation between individuals within schools, was 0.326 for Ascaris, 0.257 for hookworm, and 0.532 for Trichuris [6]. A lower value of kk (Lower kk) means higher inter-individual variation in EPG values, assumed to be 2/3 of kk in the main analysis.