OBJECTIVE: To evaluate the quality of reported adult deaths by unknown causes and to explore their relationships with some indicators. METHOD: There were evaluated adult deaths by unknown causes for all Brazilian federal units from 1990 to 2000. Criteria were established to categorize the overall proportion of deaths by quality into four groups. The age patterns of ill-defined causes were represented by typical states and correlation coefficients between these causes and other indicators were estimated, such as external causes and urbanization level. RESULTS: "Good" quality data was found in the South-Southeast regions and "fair" quality in the North-Northeast regions. Improvement on reported deaths was seen in half of the states, especially for women. The proportion of ill-defined deaths increased with age and a strong association with death coverage, urbanization level and external causes was verified. CONCLUSIONS: Although lowered quality of reported deaths by basic causes was seen in several Brazilian states, the overall quality can be considered at least as satisfactory. This means the explanatory power of death statistics by defined causes can be recovered.
Mortality; Vital statistics; Cause of death; Under-registration; Coverage of deaths; Ill-defined causes
