[Quality of data on myocardial infarction deaths, Brazil].

OBJECTIVE To evaluate the quality of data on acute myocardial infarction deaths from mortality information systems. METHODS Data on in-hospital acute myocardial infarction mortality collected from database of the Mortality Information System (SIM) and Hospital Information System (SIH), in 2000, were analyzed. Then data collected from medical records from two hospitals affiliated to the Unified Health System (SUS) in Rio de Janeiro, Brazil, were also analyzed. Medical records, death certificates, and hospital admission forms (AIH) were compared using the World Health Organization criteria of acute myocardial infarction diagnosis. Agreement among different sources was analyzed using Cohen's Kappa statistics and intraclass correlation coefficient. RESULTS In-hospital death registries in SIM are much larger than in SIH/SUS. There were identified three mechanisms that could explain most of the observed discrepancy: missing hospital admission forms (32.9%), different main diagnosis registered in SIH/SUS (19.2%), and under reporting of deaths in hospital admission forms (3.3%). The medical records review could confirm the diagnoses of acute myocardial infarction in 67.1% of all deaths reported in death certificates. The sensitivity of data on acute myocardial infarction deaths in death certificates was about 90% for both health information systems analyzed. CONCLUSIONS There is a need for actions to improve the quality of data registered in SIH/SUS such as standardization of criteria for issuing hospital admission forms during hospital emergencies and training local staff on registration systems.