|Modelling the effect of malaria endemicity on spatial variations in childhood fever, diarrhoea and pneumonia in Malawi|
||Lawrence N Kazembe, Adamson S Muula, Chirstopher C Appleton and Immo Kleinschmidt
||International Journal of Health Geographics, 2007, 6:33
Co-morbidity with conditions such as fever, diarrhoea and pneumonia is a common phenomenon in tropical Africa. However, little is known about geographical overlaps in these illnesses. Spatial modelling may improve our understanding of the epidemiology of the diseases for efficient and cost-effective control.
This study assessed subdistrict-specific spatial associations of the three conditions (fever, diarrhoea and pneumonia) in relation to malaria endemicity. We used data from the 2000 Malawi demographic and health survey which captured the history of childhood morbidities 2 weeks prior to the survey date. The disease status of each child in each area was the outcome of interest and was modelled using a trivariate logistic regression model, and incorporated random effects to measure spatial correlation.
The risk of fever was positively associated with high and medium malaria endemicity levels relative to low endemicity level, while for diarrhoea and pneumonia we observed marginal positive association at high endemicity level relative to low endemicity level, controlling for confounding covariates and heterogeneity. A positive spatial correlation was found between fever and diarrhoea (r=0.29); while weak associations were estimated between fever and pneumonia (r=0.01); and between diarrhoea and pneumonia (r=0.05). The proportion of structured spatial variation compared to unstructured variation was 0.67 (95% credible interval (CI): 0.31--0.91) for fever, 0.67 (95 % CI: 0.27--0.93) for diarrhoea, and 0.87 (95% CI: 0.62--0.96) for pneumonia.
The analysis suggests some similarities in subdistrict-specific spatial variation of childhood morbidities of fever, diarrhoea and pneumonia, and might be a result of shared and overlapping risk factors, one of which is malaria endemicity.