Spatial heterogeneity in mass drug administration from a longitudinal epidemiological study assessing transmission interruption of soil transmitted helminths in the Wolaita zone of southern Ethiopia (Geshiyaro Project)

Objectives Deworming programmes of soil-transmitted helminths are generally monitored and evaluated by aggregating drug coverage and infection levels at a district level. However, heterogeneity in drug coverage at finer spatial scales means indicators may remain above thresholds for elimination as a public health problem or of transmission in some areas. This paper aims to highlight the misleading information that aggregating data at larger spatial scales can have for programme decision making.Methods Drug coverage data from the Geshiyaro project were compared at two spatial scales with reference to the World Health Organisation's targets. District (woreda) and village (kebele) level were compared. The association between infection levels and drug coverage was analysed by fitting a weighted least-squares function to the mean intensity of infection (eggs per gram of faeces) against drug coverage.Results The data show clearly that when the evaluation of coverage is aggregated to the district level, information on heterogeneity at a finer spatial scale is lost. Infection intensity decreases significantly (p = 0.0023) with increasing drug coverage.Conclusion Aggregating data at large spatial scales can result in prematurely ceasing deworming, prompting rapid infection bounce-back. There is a strong need to define context-specific spatial scales for monitoring and evaluating intervention programmes. For soil-transmitted helminths (STH), control is centred around mass drug administration (MDA) of eligible subgroups without prior diagnosis, aiming to achieve 75% drug coverage. MDA programmes are monitored by assessing whether drug coverage and infection prevalence and intensity, aggregated to a spatial scale, typically an administrative unit such as a district, have met programmatic targets. Spatial heterogeneity in drug coverage, prevalence, and intensity is commonplace and well documented. However, little attention has been paid to the spatial scales of this heterogeneity and the implications it has on providing the basis to inform the cessation or continuation of prophylactic mass treatment. We present a clear analysis showing that fine scale spatial heterogeneity in drug coverage is commonplace even in well implemented MDA programmes. We show that aggregating data at the district level, which is routine for deworming programmes, can give misleading information on which to base future treatment decisions, specifically the premature reduction in frequency, or cessation of MDA, because fine scale heterogeneity is not captured by larger aggregations. This work highlights the critical need to define context specific spatial scales for monitoring and evaluating interventions implemented in different transmission settings.