Selection biases during recruitment of patients and relatives for a family study in the elderly

The aim of the present study was to examine selection effects during recruitment of patients; controls and relatives for a family study in the elderly. The primary sample consisted of 368 in-patients (aged above 60 years) admitted in the years 1992 and 1993. One-hundred and eighty-four subjects (50%) suffering from dementia of Alzheimer type or major depression fulfilled the diagnostic inclusion criteria. Finally, 100 subjects participated in the family study. Demographic data of participants, ineligible subjects, uncooperative candidates, and control subjects from the general population were examined. Demographic parameters, reasons for refusal of personal interviews, and family history information were compared in first-degree relatives of participants and of 40 control subjects. According to demographic data, participants were representative for the whole sample of demented or depressed patients, and were comparable with the control sample. Demographic parameters of relatives were also equivalent in both groups. Rates of psychiatric disorders were equal in interviewed and unavailable relatives of patients (18.0% and 18.8%, respectively), However, interviewed relatives of controls had significantly fewer psychiatric disorders than unavailable relatives (7.8% vs 20%). This selection effect indicates the need for family history information on unavailable relatives in family studies on geriatric patients. Equivalence of demographic data alone was not a sufficient indicator of sample comparability. A second hospitalized comparison group might serve to increase the validity of conclusions resulting from comparative family studies.