Ecological and socio-economic predictors of transmission assessment survey failure

Abstract

Since 2011, the Transmission Assessment Survey (TAS) has been recommended to determine whether cessation of mass drug administration (MDA) for lymphatic filariasis (LF) is warranted. Ministries of health typically implement TAS in evaluation units (EUs) that have had more than five rounds of MDA. Under TAS guidelines, sample size calculations determine a decision value: if the number of individuals testing positive exceeds this threshold, the EU fails the TAS and MDA continues. The objective of this study was to determine if geospatial covariates predict failure of TAS. We geo-referenced 746 TAS EUs of which 65 failed and extracted geospatial covariates using R to estimate odds of failure. We implemented stepwise backwards elimination to select covariates for inclusion in a logistic regression to estimate the odds of TAS failure. Covariates included environmental predictors (aridity, distance to fresh water, elevation, and enhanced vegetation index), cumulative rounds of MDA, measures of urbanicity and access, LF species, and baseline prevalence. Presence of Brugia was significantly associated with TAS failure (OR 4.79, 95% CI 2.52 – 9.07), as was population density (OR 2.91, 95% CI 1.06 – 7.98). The presence of nighttime lights was highly protective against failure (OR 0.22, 95% CI 0.10 – 0.50), as was an increase in elevation (OR 0.36, 95% CI 0.18 – 0.732). This work identifies predictors that are associated with TAS failure at the EU areal level given the data currently available, and also identifies the need for more granular data in order to conduct a more robust assessment of these predictors.