Shrinking the Lymphatic Filariasis Map of Ethiopia: Reassessing the Population at Risk through Nationwide Mapping

Background: Mapping of lymphatic filariasis (LF) is essential for the delineation of endemic implementation units and determining the population at risk that will be targeted for mass drug administration (MDA). Prior to the current study, only 116 of the 832 woredas (districts) in Ethiopia had been mapped for LF. The aim of this study was to perform a nationwide mapping exercise to determine the number of people that should be targeted for MDA in 2016 when national coverage was anticipated. Methodology/Principal Finding: A two-stage cluster purposive sampling was used to conduct a community-based cross-sectional survey for an integrated mapping of LF and podoconiosis, in seven regional states and two city administrations. Two communities in each woreda were purposely selected using the World Health Organization (WHO) mapping strategy for LF based on sampling 100 individuals per community and two purposely selected communities per woreda. Overall, 130 166 people were examined in 1315 communities in 658 woredas. In total, 140 people were found to be positive for circulating LF antigen by immunochromatographic card test (ICT) in 89 communities. Based on WHO guidelines, 75 of the 658 woredas surveyed in the nine regions were found to be endemic for LF with a 2016 projected population of 9 267 410 residing in areas of active disease transmission. Combining these results with other data it is estimated that 11 580 010 people in 112 woredas will be exposed to infection in 2016. Conclusions: We have conducted nationwide mapping of LF in Ethiopia and demonstrated that the number of people living in LF endemic areas is 60% lower than current estimates. We also showed that integrated mapping of multiple NTDs is feasible and cost effective and if properly planned, can be quickly achieved at national scale. Author Summary: About 1.4 billion people are believed to be living in areas where Lymphatic filariasis (LF) is actively transmitted. However, the distribution of this disfiguring mosquito-borne parasitic disease and the true population at risk that can be targeted for treatment have not been defined for all endemic countries. By 2013, Ethiopia had not delineated the majority of the endemic implementation units that can be targeted for MDA. Here, we present the results of a nationwide mapping exercise conducted in 2013 to determine the number of people that should be targeted for treatment in 2016 when nationwide treatment coverage is expected. We adopted a two-stage cluster purposive sampling method for the integrated mapping of LF and podoconiosis in seven regional states and two city administrations. Using a WHO mapping strategy for LF, based on sampling 100 individuals per community ICT positive individuals (ICT+) and two purposely selected communities per district, we examined 130 166 people in 1315 communities in 658 districts. Only 140 people were found to be positive for LF antigen in 89 different communities. According to WHO guidelines, 75 of the 658 districts surveyed in the 9 regions were found to be LF endemic. Including the 37 endemic Woredas identified enprior to this study, 112 woredas across the country are known to be endemic for the disease with 11 580 010 people exposed to infection. However 6 190 482 of those resided in woredas where our survey results were borderline with only one ICT positive individual identified. We have demonstrated that the number of people living in areas of active LF transmission is at least 60% lower than current WHO estimates of 30 million. We also showed that integrated mapping of multiple NTDs is feasible and cost effective. However, the sensitivity of the diagnostic test used for LF is less than 100% and the identification of a single ICT positive adult may not provide evidence of disease transmission. Based on these limitations, and in addition to the restricted geographical representation of just two sites within a woreda, we recommend conducting research in the 45 woredas with borderline results (one ICT+) to shrink the denominator even further.