RESUMO
BACKGROUND: Maternity histories provide a means of estimating fertility and mortality from surveys. METHODS: The present analysis compares two types of maternity histories-birth histories and pregnancy histories-in three respects: (1) completeness of live birth and infant death reporting; (2) accuracy of the time placement of live births and infant deaths; and (3) the degree to which reported versus actual total fertility measures differ. The analysis covers a 15-year time span and is based on two data sources from Matlab, Bangladesh: the 1994 Matlab Demographic and Health Survey and, as gold standard, the vital events data from Matlab's Demographic Surveillance System. RESULTS: Both histories are near perfect in live-birth completeness; however, pregnancy histories do better in the completeness and time accuracy of deaths during the first year of life. CONCLUSIONS: Birth or pregnancy histories can be used for fertility estimation, but pregnancy histories are advised for estimating infant mortality.