Background: Modern epidemiology increasingly uses data on families. The authors constructed an extended fertility database for women born in Denmark from 1930 onwards by supplementing the existing Fertility Database with household data from the 1970 census.
Methods: A fertility history was constructed for all women participating in the 1970 census, but aiming for complete data only for women aged 20-39. The fertility history of these women prior to the 1970 census was constructed from the census data including 1,648,813 persons coded as children. An algorithm was used transforming household information into fertility history data by matching women and children according to family position. Children for whom the algorithm gave no match were searched for in the Fertility Database; children not found in the Fertility Database either were searched for manually. The fertility history after the 1970 census was retrieved from the Fertility Database.
Results: Using data from the census 1970, 98.5% of the children were linked to a mother, and 99.6% of these links were estimated to be correct, corresponding to 98.1% of the children being linked correctly. In total, 964,720 children of women aged 20-39 in 1970 were identified, which was equivalent to 96.6% of the expected live-born children, and to 99.1% of the expected surviving children.
Conclusion: Census household data proved to be an excellent data source for construction of fertility histories.