Are census data accurate for estimating coverage of a lymphatic filariasis MDA campaign? Results of a survey in Sierra Leone

Background: Preventive chemotherapy was administered to 3.2 million Sierra Leoneans in 13 health districts for lymphatic filariasis, onchocerciasis, and soil transmitted helminthes from October 2008 to February 2009. This paper aims to report the findings of a coverage survey conducted in 2009, compare the coverage survey findings with two reported rates for lymphatic filariasis coverage obtained using pre-mass drug administration (MDA) registration and national census projections, and use the comparison to understand the best source of population estimates in calculating coverage for NTD programming in Sierra Leone. Methodology/Principal findings: Community drug distributors (CDDs) conducted a pre- MDA registration of the population. Two coverage rates for MDA for lymphatic filariasis were subsequently calculated using the reported number treated divided by the total population from: 1) the pre-MDA register and 2) national census projections. A survey was conducted to validate reported coverage data. 11,602 persons participated (response rate of 76.8%). Overall, reported coverage data aggregated to the national level were not significantly different from surveyed coverage (z-test >0.05). However, estimates based on pre-MDA registration have higher agreement with surveyed coverage (mean Kendall’s W = 0.68) than coverage calculated with census data (mean Kendall’s = 0.59), especially in districts with known large-scale migration, except in a highly urban district where it was more challenging to conduct a pre-MDA registration appropriately. There was no significant difference between coverage among males versus females when the analyses were performed excluding those women who were pregnant at the time of MDA. The surveyed coverage estimate was near or below the minimum 65% epidemiological coverage target for lymphatic filariasis MDA in all districts. Conclusion/Significance: These results from Sierra Leone illustrate the importance of choosing the right denominator for calculating treatment coverage for NTD programs. While routinely reported coverage results using national census data are often good enough for programmatic decision making, census projections can quickly become outdated where there is substantial migration, e.g. due to the impact of civil war, with changing economic opportunities, in urban settings, and where there are large migratory populations. In districts where this is known to be the case, well implemented pre-MDA registration can provide better population estimates. Pre-MDA registration should, however, be implemented correctly to reduce the risk of missing pockets of the population, especially in urban settings.