Atlas of Mortality in Spanish Municipalities and Census Tracts (1984-2004)

To create maps which indicate high-risk areas and low-risk areas of each cause of death, we used the following procedure. First, for each area, we tested a null hypothesis of relative risk equalling 1. If the departure from the null relative risk value was statistically significant at 5% level, then the area was considered as a high-risk or low-risk area depending on whether its empirical Bayes estimate of relative risk was above or below 1 respectively. Within the set of all high-risk areas, we indicated those within the top quartile of the relative risk estimate as the "highest" areas, tose bek the median as "high" areas. Similarly, the bottom quartile and below median areas were specially marked for the set of all low-risk areas.