Problems in the use of dead controls in case-control studies. I. General results.

A recently completed case-control study in the Minneapolis-St. Paul area using population-based living controls and dead controls afforded the opportunity to compare these two control groups on their exposure histories. Detailed information was obtained by interview from 697 living controls and the next of kin of 493 dead controls. The dead controls of both sexes were reported to have been significantly heavier cigarette smokers compared with living controls, as well as heavier consumers of hard liquor, beer, and drugs, and to have had more adulthood diseases. There were no consistent differences between the control groups for consumption of nonalcoholic beverages, some aspects of diet, ethnic and religious background, usual occupation, and residential history. It appears that exposures associated with premature death are overrepresented in dead controls compared with living controls, while those variables not associated with premature mortality are distributed more or less similarly between the two groups.