The design and interpretation of case-control studies of perinatal mortality.

The difficulty of designing prospective studies of perinatal death makes the case-control study a method of choice. The particular problems associated with the identification of risk factors, the definition and enumeration of cases, and the selection of live birth controls in such a study, undertaken on all perinatal deaths occurring in a population of 850,000 people during 1976-1978 are described. The method of control selection, chosen for reasons of feasibility, produced a nonrepresentative sample of controls. This was because controls were selected as the next live birth in the place of delivery where the perinatal death delivery occurred, which resulted in a sample stratified by place of delivery. Knowledge of the place of delivery of all births allowed a correction to be undertaken which was derived from the relative weights for the strata within which matching had occurred.