Mortality by road traffic injuries in Brazil (2000-2016): capital cities versus non-capital cities

To compare the magnitude and trend of mortality by road traffic injuries (RTI) in the capitals and other municipalities of each Brazilian state between 2000 and 2016.

A time series analysis of mortality rates by RTI standardized by age was performed, comparing the capitals and the cluster of non-capital municipalities in each state. Data on deaths were obtained from the Sistema de Informações sobre Mortalidade (SIM - Mortality Information System). RTI deaths were considered to be those, whose root cause was designated by ICD-10 codes V01 to V89, with redistribution of garbage codes. To estimate mortality rates, we used the population projections of the Brazilian Institute of Geography and Statistics (IBGE) from 2000 to 2015 and the population estimated by polynomial interpolation for 2016. The trend analysis was performed using the Prais-Winsten method, using the Stata 14.0 program.

There were 601,760 deaths due to RTI in the period (114,483 of residents in capital cities). Mortality by RTI did not present an increasing trend in any of the Capitals in the period under study. Among non-capital municipalities, the trend was growing in 14 states. The greatest increase was observed in Piaui (AIR = 7.50%; 95%CI 5.50 - 9.60). There was a decreasing trend in RTI mortality in 14 capitals, among which Curitiba showed the greatest decrease (AIR = -4.82%; 95%CI -6.61 - -2.92). Only São Paulo and Rio Grande do Sul showed a decreasing trend in mortality by RTI in non-capital cities (AIR = 2.32%; 95%CI -3.32 - -1.3 and AIR = 1.2%, 95%CI -2.41 - 0.00, respectively).

We conclude that RTI mortality rates in non-capital cities in Brazil showed alarming trends when compared with those observed in capital cities. The development of effective traffic safety actions is almost always limited to Brazilian capitals and large cities. Municipalities with higher risk should be prioritized to strengthen public policies for prevention and control.