Monitoring perinatal mortality depends on the quality of data of information systems. The different definitions for reporting and calculating perinatal mortality may affect the magnitude and analysis of its components. We compared data availability on live born, fetal and early neonatal deaths in civil registry, SIM (Mortality Information System) and SINASC (Birth Registration System) for 8 Brazilian states, with reporting coverage above 90% in 2002. SINASC had a higher coverage of events than the civil registration and excellent completeness of data (above 99%). SIM has a different situation, there is a high absence of information on birth weight (23.4%), gestational age (9.1%), type of pregnancy (13.8%), mother's age (18.5%), and mother's years of education (40.6%) for fetal deaths. Early neonatal deaths showed similar figures: birth weight was missing in 22.6% of records, gestational age (17.8%), type of pregnancy (19.1%), age (27.9%), and mother's years of education (38.5%). It was not possible to verify if fetal deaths were during delivery or pre delivery due to lack of information. However, these data could be easily obtained considering that more than 95% of all events occurred in hospitals. Criteria for reporting fetal deaths and live born in information systems make it difficult to compare the magnitude and the participation of perinatal mortality components internationally. The absence of information affects attainment of specific indicators, making monitoring activities difficult. Some activities are necessary for improving the SIM.
Perinatal mortality; Mortality and live born information systems; Event reporting; Quality of information
