Contextual determinants of mass drug administration performance: Modelling fourteen years of lymphatic filariasis treatments in West Africa

by Brian B. Fuller, Vance Harris, Caleb Parker, Andres Martinez, Emily Toubali, Blandine Clarisse Ebene, Kofi Asemanyi-Mensah, Massitan Dembele, Adamou Bacthiri Salissou, Cathérine Kabré, Aboulaye Meite, Ndeye Mbacke Kane, Ibrahim Kargbo-Labour, Wilfrid Batcho, Aissatou Diaby, Violetta Yevstigneyeva, Diana Maria Stukel

Background

Effective mass drug administration (MDA) is the cornerstone in the elimination of lymphatic filariasis (LF) and a critical component in combatting all neglected tropical diseases for which preventative chemotherapy is recommended (PC-NTDs). Despite its importance, MDA coverage, however defined, is rarely investigated systematically across time and geography.Most commonly, investigations into coverage react to unsatisfactory outcomes and tend to focus on a single year and health district. Such investigations omit more macro-level influences including sociological, environmental, and programmatic factors.The USAID NTD database contains measures of performance from thousands of district-level LF MDA campaigns across 14 years and 10 West African countries. Specifically, performance was measured as an MDA’s epidemiological coverage, calculated as persons treated divided by persons at risk.This analysis aims to explain MDA coverage across time and geography in West Africa using sociological, environmental, and programmatic factors.

Methodology

The analysis links epidemiological coverage data from 3,880 LF MDAs with contextual, non-NTD data via location (each MDA was specific to a health district) and time (MDA month, year). Contextual data included rainfall, temperature, violence or social unrest, COVID-19, the 2014 Ebola outbreak, road access/isolation, population density, observance of Ramadan, and the number of previously completed MDAs.

Principal findings

We fit a hierarchical linear regression model with coverage as the dependent variable and performed sensitivity analyses to confirm the selection of the explanatory factors.Above average rainfall, COVID-19, Ebola, violence and social unrest were all significantly associated with lower coverage. Years of prior experience in a district and above average temperature were significantly associated with higher coverage.

Conclusions/Significance

These generalized and context-focused findings supplement current literature on coverage dynamics and MDA performance. Findings may be used to quantify typically anecdotal considerations in MDA planning. The model and methodology are offered as a tool for further investigation.