Use of dual responses to increase validity of case-control studies.

It has been demonstrated that randomly distributed inaccuracy of dichotomous reports in case-control studies biases estimates of the effects of exposure toward one. We show that a simple procedure involving the use of dual exposure reports may lessen this bias. The use of this procedure is illustrated with a data set in which the actual relative risk associated with dichotomous exposure is 4.0. We observe that, if exposure is reported with 60% accuracy and the observed relative risk is approximately 1.3, an additional set of equally accurate exposure reports can be used to uncover an assessed relative risk as high as 1.7. The supplementing of exposure reports with 90% accuracy by an additional set of equally accurate reports can be used to uncover an assessed relative risk as as high as 1.7. The supplementing of exposure reports with 90% accuracy by an additional set of equally accurate exposure reports can increase the observed relative risk from 3.0 to as high as 3.9. We conclude that using dual reports could be valuable in strengthening the inference of case-control studies assessing the effects of dietary, occupational, or other exposures.