Selection bias during recruitment of elderly subjects from the general population for psychiatric interviews.

The aim of the present study was to determine and assess a possible selection bias in an epidemiologic investigation in the elderly. A stratified sample of 1305 probands aged 60-99 years was initially contacted by mail and then by telephone to obtain their consent to participate in a psychiatric interview. A liberal recruitment procedure led to interview participation of only 291 subjects. The proportion of younger, male, and married subjects participating in the study was greater than that of elderly, female, and single or widowed subjects. Subjects without a psychiatric lifetime diagnosis were more cooperative than those with a psychiatric disorder. The latter finding demonstrates the need to determine and assess the selection bias in psychiatric epidemiologic studies in elderly subjects.