Background:
Western Australia (WA) has extensive record-linked population health data collections. Midwives' records of WA births during 1980–1992 inclusive were linked and validated by the author to form sibships, which enable longitudinal family studies. Binary indicators of maternal morbidity were created using this database of prospectively collected birth data. A validated system, using case note reviews, for scoring obstetric complications as risk factors for psychiatric morbidity has been adapted for coding population data. This will be extended to 2001 births and include the mother's clinical history.
Methods:
Women who have had only one birth during this period, who had multiple births or who had missing births in their sibship were excluded from analysis. Codes created reflect morbidity in the current pregnancy only, the one immediately prior, any prior pregnancy or, the current or any prior pregnancy.
Results:
The database includes records of 472 733 births to 249 119 women with paternal data available for the majority of children. Of these, approximately 90 000 (36%) women were eligible for inclusion. Of the 67 242 eligible women, who delivered during 1980–1992, severe underestimation of maternal conditions, affecting pregnancy, such as thyroid dysfunction, essential hypertension or epilepsy occurred when only index birth records were used.
Conclusions:
Using data from the index birth only could underestimate the effects of maternal morbidity, as risk factors for outcomes such as schizophrenia, when compared with analysis of models that use longitudinal binary indicators of current and prior maternal illness.
