[Data quality analysis of deaths committed by the police in the city of São Paulo, Brazil, 2014-2015].

Deadly police force is a public health problem. Although the Mortality Information System (SIM) is the most reliable record of deaths from violence, the same is not true for cases of deadly police force, which displays a high degree of underreporting when compared to data from the São Paulo Department of Law Enforcement (SSP-SP). The current study aimed to estimate underreporting in the two official data sources (SIM and SSP-SP), identifying the ICD-10 categories used in cases of incorrectly classified deadly police force and mortality rates in the years 2014 and 2015 in the city of São Paulo, Brazil. Using linkage of data from the SIM and SSP-SP databases, we describe the use of underlying causes of death in cases of deadly police force, estimating underreporting in the SIM and the SSP-SP with the capture-recapture methodology and mortality rates in the city. Based on the database linkage, most of the deaths from deadly police force were classified incorrectly (53%) as other underlying causes of death in the SIM. Both the SIM and SSP-SP underreported the deaths committed by police officers, with different magnitudes (53.2% in the SIM and 7.9% in the SSP-SP). Reclassification of the deaths via linkage added a gain in the SIM, which now had the same mean mortality rate as the SSP-SP (3.44/100,000), thereby decreasing the underreporting in comparison to the initial scenario. Correct recording of death is the first step to the ensuring the right to justice and truth. Recording with quality means to guarantee the right to information, which is not an end per se, but the start in the task of prevention. Data-sharing and inter-sector work are urgently needed.