[Statistical approaches to test for spatial heterogeneity of relative risks of cause-specific mortality in Alto Vicentino Municipalities (years 1991-2000)].

OBJECTIVE To review the statistical approaches to test for spatial heterogeneity of relative risks. Three different statistical tests (Gail, Martuzzi-Hills and Potthoff-Whittinghill) are reviewed and applied--as motivating example--to the analysis of cause-specific mortality records (years: 1991-2000) of the Municipalities belonging to the Local Health Unit Alto Vicentino. METHODS Spatial heterogeneity was found in 17 (Martuzzi-Hills) and 18 (Gail, Potthoff-Whittinghill) among 70 selected causes of death. Cohens Kappa test was chosen to assess the agreement among the tests (k = 0.596; p < or = 0.001). CONCLUSION Spatial heterogeneity must be interpreted with great caution, taking into account the available evidence about risk factors for specific causes of death and carefully evaluating available exposure data.