Exploring drug coverage variability within districts: A CES approach to investigate treatment gaps in Mozambique's schistosomiasis program.

BACKGROUND: Coverage Evaluation Surveys (CES) are tools used by NTD control programs to assess drug coverage after mass drug administration (MDA), typically conducted at the district level. However, little research has explored how this district-level approach may mask uneven drug coverage within districts. Such uniform reporting may obscure gaps at smaller administrative levels, potentially limiting the effectiveness of MDA efforts. This study examines CES findings from a 2021 survey in Mozambique investigating coverage at the sub-district level to better understand heterogeneity in drug coverage, while also exploring other actionable survey data, like reasons for not taking drugs, to better understand why coverage gaps and discrepancies with reported coverage may have occurred. METHODS: A CES was conducted in May of 2021 to assess district level drug coverage following schistosomiasis MDA in Nov-Dec 2020 in 10 districts in northern Mozambique. Following data collection, administrative post-level classifications were added retroactively by linking GPS data from the survey to spatial shapefiles, enhancing the analysis resolution. Programmatic coverage estimates were then calculated as the proportion of eligible individuals who reported taking praziquantel in each administrative unit. Finally, an ANOVA model was applied to examine the variance in drug coverage across province, district, and administrative post level, with cross tabulation exploring sub-district level variation. RESULTS: Effective programmatic coverage (≥75%) was achieved with 95% confidence in five districts (Macossa, Majune, Mandiba, Maua, and Nipepe). Coverage discrepancies between surveyed and reported data were observed in nine out of ten districts, highlighting potential inconsistencies in routine reporting. ANOVA modeling showed that 49% of the observed variation in drug coverage could be explained by provincial, district, and administrative post-level differences. The ANOVA results, along with cross tabulation of coverage among administrative posts together highlight the likelihood of drug coverage variability within districts. Analysis of non-participation revealed that absence from the community at the time of MDA and lack of awareness of the campaign were the leading reasons for not taking the drug, particularly in urban districts. CONCLUSIONS/SIGNIFICANCE: This study highlights how CES can be used to detect heterogeneity of coverage within districts while simultaneously identifying behavioral and operational barriers to participation. Integrating sub-district level spatial analysis with contextual data on reasons for non-uptake can support in the identification of local treatment gaps, which may be integral to tailor approaches to improve MDA participation in future campaigns as national schistosomiasis programs are increasingly urged to conduct MDA at a sub-district level. These findings suggest that more granular data collection within CES could better inform program adjustments and resource allocation.