Background: Childhood mortality is a major public health problem in sub-Saharan Africa. For the implementation of efficient public health systems, knowledge of the spatial distribution of mortality is required.
Methods: Data from a demographic surveillance research project were analysed which comprised information obtained for about 30 000 individuals from 39 villages in northwest Burkina Faso (West Africa) in the period 1993--1998. Total childhood mortality rates were calculated and the geographical distribution of total childhood mortality was investigated. In addition, data from a cohort of 686 children sampled from 16/39 of the villages followed up during a randomized controlled trial in 1999 were also used to validate the results from the surveillance data. A spatial scan statistic was used to test for clusters of total childhood mortality in both space and time.
Results: Several statistically significant clusters of higher childhood mortality rates comprising different sets of villages were identified; one specific village was consistently identified in both study populations indicating non-random distribution of childhood mortality. Potential risk factors which were available in the database (ethnicity, religion, distance to nearest health centre) did not explain the spatial pattern.
Conclusion: The findings indicate non-random clustering of total childhood mortality in the study area. The study may be regarded as a first step in prioritizing areas for follow-up public health efforts.