Mapping and prediction of schistosomiasis in Nigeria using compiled survey data and Bayesian geospatial modelling

Abstract

Schistosomiasis prevalence data for Nigeria were extracted from peer-reviewed journals and reports, geo-referenced and collated in a nationwide geographical information system database for the generation of point prevalence maps. This exercise revealed that the disease is endemic in 35 of the country’s 36 states, including the federal capital territory of Abuja, and found in 462 unique locations out of 833 different survey locations. Schistosoma haematobium, the predominant species in Nigeria, was found in 368 locations (79.8%) covering 31 states, S. mansoni in 78 (16.7%) locations in 22 states and S. intercalatum in 17 (3.7%) locations in two states. S. haematobium and S. mansoni were found to be co-endemic in 22 states, while co-occurrence of all three species was only seen in one state (Rivers). The average prevalence for each species at each survey location varied between 0.5% and 100% for S. haematobium, 0.2% to 87% for S. mansoni and 1% to 10% for S. intercalatum. The estimated prevalence of S. haematobium, based on Bayesian geospatial predictive modelling with a set of bioclimatic variables, ranged from 0.2% to 75% with a mean prevalence of 23% for the country as a whole (95% confidence interval (CI): 22.8-23.1%). The model suggests that the mean temperature, annual precipitation and soil acidity significantly influence the spatial distribution. Prevalence estimates, adjusted for school-aged children in 2010, showed that the prevalence is <10% in most states with a few reaching as high as 50%. It was estimated that 11.3 million children require praziquantel annually (95% CI: 10.3-12.2 million).

Keywords

• schistosomiasis, prevalence, geo-referencing, geographical information system, risk mapping, Bayesian geospatial modelling, control, Nigeria.