Non-random distribution of the proportion of ill-defined causes of death in Brazil

Background: The proportion of ill-defined causes of death (PIDD) is a readily available health information quality indicator that can be used as a proxy measure of quality and access to healthcare.

Objectives: To analyze the distribution of PIDD within Brazilian municipalities in 2004.

Methods: A descriptive exploratory ecological study was carried out using secondary data from the Brazilian Mortality Surveillance Database. The PIDD, defined using WHO – ICD10 criteria, was calculated for each municipality, and stratified by age and gender. The roles of region and population size of the municipality were investigated.

Results: In 2004, the 5,503 Brazilian municipalities reported a total of 938,648 deaths. 15.1% of deaths were coded as ill-defined causes. 1,010 (20.2%) municipalities reported PIDD higher than 50%. The smallest PIDD median was observed in the South Region (6.6%), followed by the Center-West (10.0%) and Southeast (14.9%). The North (29.5%) and Northeast (50.0%) regions presented statistically significant (p<0.001) increased PIDD medians compared to the South. A trend of increased PIDD medians with increased age of population dying was also observed. There is a negative non-linear association between population size and PIDD, within all regions (p < 0,001).

Conclusion: The PIDD is a non-random sub-set of the total reported deaths, concentrated in the North and Northeast regions, in small-sized municipalities, and among the elderly population. This non-random distribution pattern indicates that PIDD is likely to hide an unknown pattern of underlying causes of death, which does not necessarily resembles the reported group of identified causes. The magnitude of the PIDD in a large number of Brazilian municipalities is per se an important problem. Despite recent achievements, much still have to be done to improve the Brazilian health databases aiming their use as a valid and reliable epidemiological tool for public health evaluation and monitoring.