Live birth and infant death record linkage: methodological and policy issues.

Using records from Ohio annual vital statistics tapes, we describe a method for linking live birth and infant death certificates and for dealing with late-registered and unregistered births. In our 1985-87 Cleveland and East Cleveland study population, deceased infants with late-registered births were found to be similar to those with timely registered births. Approximately 4.6% of decedents, however, had unregistered births and these tended to be very premature infants from socially disadvantaged backgrounds who died shortly after delivery (including homicides following birth at home). We discuss the policy implications of failing to link infant deaths with unregistered births in studies of birth outcomes.