Background: Urinary Schistosomiasis infection, a common cause of morbidity especially among children in less developed countries, is measured by the number of eggs per urine. Typically a large proportion of individuals are non-egg excretors, leading to a large number of zeros. Control strategies require better understanding of its epidemiology, hence appropriate methods to model infection prevalence and intensity are crucial, particularly if such methods add value to targeted implementation of interventions.
Methods: We consider data that were collected in a cluster randomized study in 2004 in Chikhwawa district, Malawi, where eighteen (18) villages were selected and randomised to intervention and control arms. We developed a two-part model, with one part for analysis of infection prevalence and the other to model infection intensity. In both parts of the model we adjusted for age, sex, education level, treatment arm, occupation, and poly-parasitism. We also assessed for spatial correlation in the model residual using variogram analysis and mapped the spatial variation in risk. The model was fitted using maximum likelihood estimation.
Results and discussion: The study had a total of 1642 participants with mean age of 32.4 (Standard deviation: 22.8), of which 55.4 % were female. Schistosomiasis prevalence was 14.2 %, with a large proportion of individuals (85.8 %) being non-egg excretors, hence zero-inflated data. Our findings showed that S. haematobium was highly localized even after adjusting for risk factors. Prevalence of infection was low in males as compared to females across all the age ranges. S. haematobium infection increased with presence of co-infection with other parasite infection. Infection intensity was highly associated with age; with highest intensity in school-aged children (6 to 15 years). Fishing and working in gardens along the Shire River were potential risk factors for S. haematobium infection intensity. Intervention reduced both infection intensity and prevalence in the intervention arm as compared to control arm. Farmers had high infection intensity as compared to non farmers, despite the fact that being a farmer did not show any significant association with probability of infection. These results evidently indicate that infection prevalence and intensity are associated with risk factors differently, suggesting a non-singular epidemiological setting. The dominance of agricultural, socio-economic and demographic factors in determining S. haematobium infection and intensity suggest that disease transmission and control strategies should continue centring on improving socio-economic status, environmental modifications to control S. haematobium intermediate host snails and mass drug administration, which may be more promising approaches to disease control in high intensity and prevalence settings.