Epidemiology and preventive interventions: parental death in childhood as a case example.

Illustrates how standard epidemiologic principles form the knowledge base to justify a preventive intervention for an at-risk population. These principles were applied to a sample of 92 from the population of children aged 8 to 15 at alleged risk for mental health disorders because a parent died. Prior work on this alleged risk population is sparse and flawed. Validly determining the population effect of an alleged risk factor requires assessing the influence of sampling bias. The bias found, underrepresentation of deaths of a mother, did not influence the relations among death of a parent and children's depression and conduct disorder, and the modifiable mediators of risk to be changed by the preventive intervention. The epidemiologic measure of effect indicated that death of a parent is a risk factor for major depression but not for conduct disorder among youth. Families recruited for the preventive intervention by epidemiologic methods (ES families) did not differ significantly from the earlier families on whom the knowledge base was formed. Families referred to the intervention by self or others significantly differed from the ES families in two ways that constituted serious biases. The implications of these biases for prevention were discussed.