[Underreporting of maternal mortality in Rio de Janeiro, Brazil: comparison of 2 information systems]

Two notification systems were compared to estimate maternal death underreporting for 1988 in eight public maternity clinics of Rio de Janeiro. The death rates obtained were compared with the corrected maternal death rate (13.9 per 10,000 live births), which represented all deaths validated by either system. The system based on hospital discharge records yielded a maternal death-rate of 11.6 per 10,000 live births, or 16% underreporting when compared with the corrected death rate. Error was due to system failure in processing the discharge forms, and also to the transfer of patients to nonmonitored hospitals. The system based on death certificates, including "presumed" maternal deaths, presented a rate of 6.6 per 10,000 live births, with 52% underreporting in relation to the corrected rate. The official death rate, based on maternal deaths as declared in death certificates, underreported by 60% compared to the corrected death rate. When the two systems were compared with each other, the one based on death certificates showed 62% underreporting in relation to that based on hospital discharge forms. In order to minimize underreporting of maternal deaths and to broaden the knowledge on circumstances relating to those deaths, it is suggested (a) that a perinatal information system be introduced in all obstetric service units, and (b) to monitor the final outcome of patients transferred to tertiary units.