INTRODUCTION: Mortality data are important for monitoring violence, making it possible to assess the trends and the impact of interventions towards its reduction. The objective of the study is to assess the quality of the filling out and codification of the death certificates for unspecified accidents and events of undetermined intent in the city of S. Paulo in 1996. METHODS: Death certificates on which the underlying cause of death (UCD) given was an unspecified accident (ICD-10 X59) or an event of undetermined intent (ICD-10 Y10-Y34) were selected for investigation at the Legal Medicine Institute (IML). After consulting the police reports which accompany the corpses to the IML, the autopsy reports and other legal forms, these were analysed and the UCD was recoded. RESULTS: For unspecified accidents, 53.2% were changed to a specified cause: 15.1% due to pedestrians injured in traffic accidents, 17.5% due to other traffic accidents and 14.5% due to falls. Homicides and suicides constituted 9.8%. In 20.9% no additional information was found. For events of undetermined intent, 2/3 had no clarification; in 1/3 of the cases, the underlying cause changed to falls (10.6%), homicides (7.5%) and pedestrians injured in transport accidents (6.7%). CONCLUSIONS: The quality of mortality information by external causes in the City of S. Paulo is not satisfactory. The IML has not used all the available information to fill out the death certificates. The findings reveal that the instruction of the World Health Organization and the Brazilian Center for the Classification of Diseases to codify as accidents those events for which there is no information on the death certificate about the external cause, does not seem to be appropriate. In that category 66.0% of the deaths were found to have been inferred incorrectly as accidental. The improvement of the quality of mortality data due to external causes may contribute to the monitoring of violence and may give support to decisions leading to its reduction whatever the form that violence may take.
Violence; Mortality registries; Quality
