Ascertaining dementia-related outcomes for deceased or proxy-dependent participants: an overview of the Women's Health Initiative Memory Study supplemental case ascertainment protocol.

OBJECTIVE: The aim of the study was to compare a two-staged clinic-based standardized protocol with a supplemental proxy-based protocol.
METHODS: The Women's Health Initiative Memory Study enrolled 7479 women, aged 65-79 years and free of dementia, in a clinical trial of postmenopausal hormone therapy who were followed for up to 13 years with annual two-staged clinic-based standardized protocols to identify incidence of probable dementia. A supplemental proxy-based protocol, involving telephone administration of the dementia questionnaire, was designed to assess the cognitive status of women who could no longer attend clinic visits because they died (n = 1058) or became dependent (n = 228). Chi-squared tests were used to compare characteristics of women eligible for proxy-based versus clinic-based assessment. Risk factor relationships were described using proportional hazards regression.
RESULTS: Women who were eligible for proxy-based assessments tended to have worse cognitive impairment risk factor profiles and had higher rates of probable dementia (15.2% vs 3.5%) than clinic-assessed participants. Augmenting the clinic-based cases with those identified from proxy interviews reduced undercounting and materially altered observed relationships that years since menopause, smoking status, diabetes, and prior use of hormone therapy had with incidence of probable dementia.
CONCLUSIONS: Although proxy interviews were successful in reducing biases in estimated incidence rates and risk factor relationships, it is unlikely that they will fully eliminate many biases. Proxy-based assessments are necessary in longer term studies to reduce undercounting of dementia cases and to characterize risk factor relationships.