Risk factors and micro-geographical heterogeneity of Schistosoma haematobium in Ndumo area, uMkhanyakude district, KwaZulu-Natal, South Africa.

Schistosomiasis is a snail-transmitted parasitic disease endemic in most rural areas of sub-Saharan Africa. However, the currently used prediction models fail to capture the focal nature of its transmission due to the macro-geographical levels considered and paucity of data at local levels. This study determined the spatial distribution of Schistosoma haematobium and related risk factors in Ndumo area, uMkhanyakude District, KwaZulu-Natal province in South Africa. A sample of 435 schoolchildren between 10 to 15 years old from 10 primary schools was screened for S. haematobium using the filtration method. Getis-Ord Gi* and Bernoulli model were used to determine the hotspots of S. haematobium infection intensity based on their spatial distribution. Semiparametric-Geographically Weighted Regression (s-GWR) model was used to predict and analyse the spatial distribution of S. haematobium in relation to environmental and socio-economic factors. We confirmed that schistosomiasis transmission is focal in nature as indicated by significant S. haematobium cases and infection intensity clusters (p<0.05) in the study area. The s-GWR model performance was low (R(2)=0.45) and its residuals did not show autocorrelation (Moran's I=-0.001; z-score=0.003 and p-value=0.997) indicating that the model was correctly spelled. The s-GWR model also indicated that the coefficients for some of the socio-economic variables such as distances of households from operational piped water collection points, distance from open water sources, religion, toilet use, household head and places of bath and laundry significantly (t-values+/-1.96) varied across the landscape thereby determining the variation of S. haematobium infection intensity. This evidence may be used for control and management of the disease at micro scale. However, there is need for further research into more factors that may improve the performance of the s-GWR models in determining the local variation of S. haematobium infection intensity.