Objective: To identify predictors of nonresponse in a community survey of cognitive status in the elderly.
Design: Cross-sectional community survey with two stages of recruitment: an initial, less-intensive method, followed by a more aggressive approach that included face-to-face contact. Characteristics of initial nonresponders and responders were compared.
Setting: A close-knit rural community with higher than usual proportions of elderly, especially the very old. Subjects were interviewed in their homes. Collateral informants were subsequently interviewed by telephone.
Participants: Utah heads of household aged 75 and older who resided in a noninstitutionalized setting.
Measurements: Mini-Mental State Examination (MMSE), Dementia Questionnaire, and an autobiographical risk factor and family history questionnaire provided measures for all independent variables. The dependent variable was status as initial responders or initial nonresponders.
Results: An initial participation rate of 63% was achieved, but a final rate of 93% was achieved when initial nonresponders were contacted later face-to-face. MMSE score was significantly related to responder status when analyzed alone (beta = -.19, P = 0.02) and remained a significant predictor after adjusting for education and whether born in Cache County (beta = -.16, P = 0.041) or current drinking, diabetes, or "other" health problems (beta = -.18, P = 0.028). After controlling for the informant report of subject's problems with activities of daily living, MMSE score fell just below statistical significance (beta = -.16, P = 0.079).
Conclusions: Nonresponders in community surveys of the elderly appear to be disproportionately cognitively impaired. The increase in participation rates achieved after more persistent recruitment suggests that many initial nonresponders can still be recruited if intensive methods are used.