Using spatial analysis to identify areas vulnerable to infant mortality.

OBJECTIVE To analyze the spatial distribution of infant mortality and identify clusters with high risk of death in the first year of life. METHODS The Thiessen (Voronoi) polygon method was used to analyze spatial distribution of the infant mortality rate, calculated by municipality. The triennium 2006 - 2008 was used as a reference to estimate the average infant mortality rate, and the first analysis of the spatial distribution of the rate was performed to test for first-order spatial stationarity. The spatial pattern was then analyzed using Moran's index and G-statistic (α = 5%). RESULTS The surface projections on trends showed that infant mortality is not constant in space. The Moran index (0.34, P < 0.01) and G-statistic (0.03, P < 0.01) confirmed a spatial autocorrelation between infant mortality and clusters when the Thiessen polygon method was used. CONCLUSIONS The Voronoi polygons proved accurate for spatial analysis of infant mortality and were predictive of clusters with high risk of death in the first year of life.