Criteria for selection of decedent versus living controls in a mortality case-control study.

The mortality odds ratio computed with a decedent control series (ORD) and the mortality odds ratio obtained with a living control series (ORL) estimate the mortality rate ratio (RRM) under very different assumptions. The bias of the ORD as an estimate of RRM is a function of the mortality rates from other causes (those causes for which controls were selected) in the underlying exposed (E) and nonexposed (E) populations, while the bias of ORL as an estimate of RRM is a function of the prevalence of the study disease in the underlying E and E populations. The ORL closely estimates the RRM if the prevalence of the disease under study is small. If the study disease mortality rate is small relative to the other-causes mortality rate, the ORD closely estimates the cause-specific proportional mortality ratio.