[The use of the "linkage" technique of information systems in cohort studies on neonatal mortality].

The utilization of record linkage of the mortality and birth information systems in studies of neonatal mortality is presented. The record linkage was used to obtain a cohort of live births and neonatal deaths in Santo André county, located within greater S. Paulo, in 1992. The procedures applied in order to avoid selection and effect biases, are discussed. The use of linked data allows the probabilities of neonatal deaths according to the exposure status of the variables which are registered on the birth certificate, and the identification of the live born at risk, to be calculated. Another advantage of the record linkage is the low financial cost of this type of study, because it uses information already registered.