Population density and spatial differentials in child mortality in Zimbabwe.

Large regional variations in under-five mortality exist within many sub-Saharan countries. Population density as a potential explanatory factor for these regional variations has seldom been considered despite it being implicated as a determinant of mortality at other spatial scales. In Zimbabwe, the "Ndebele provinces"-Matabeleland North and South-have significantly lower levels of under-five mortality than the other ("Shona") provinces. This regional differential is explored using the Zimbabwe Demographic and Health Survey and census data. Factors other than population density that may contribute to the differential are examined. After controlling for the effects of potentially confounding socio-economic, demographic and environmental variables using Cox regression models children aged 1-4 yr living in the Ndebele provinces continued to have a level of mortality 45% lower than their counterparts in the Shona provinces. The possibility that regional variations in health care provision and/or cultural factors contribute to the mortality differential is also examined and rejected. Population densities in the Ndebele provinces are of a far lower order than in the Shona provinces. The main causes of child mortality in Zimbabwe in the time period under consideration were diarrhoea, ALRI, measles and malaria. How population density may affect the transmission of these infections and, hence, mortality is discussed. It is suggested that population density may provide an explanation for the spatial variation in child mortality in Zimbabwe. The implications of changing population densities for child health in urban and rural sub-Saharan Africa are briefly considered.