The purpose of the proposed study is to systematically evaluate proxy respondent's ratings of current functional status and changese in functional status in older persons. Major issues to be addressed are whether proxies over- or under-report present status and change in functional status when compared to direct observations and self-reports of older subjects themselves. Proxy-subject agreement will also be evaluated as a function of proxy and subject characteristics. The areas of functioning to be studied include: physical and instrumental activities of daily living; cognitive status; affective status and social interaction. For many areas of functioning, direct observations of subjects will be made; hence, the validity of proxy responses will be evaluated for these areas of questioning. Study issues will be addressed through analysis of data obtained from hip fracture patients enrolled in an ongoing study and a designated proxy for each. Subjects include hip fracture patients aged 65 and older admitted from the conununity to 8 Baltimore area hospitals. For all patients (N=320) who will have a one year evaluation after May 1, 1991, data obtained through the parent study during the 6 and 12 month post-fracture evaluation will be used. In addition, a series of direct observations of physical and instrumental activities of daily living will be made. A proxy respondent will also be identified for each patient and interviewed within a week of the 12 month patient evaluation to obtain his/her assessment of the patient's current functional status and 6 month change in functioning. Linear model analyses will be used to evaluate the degree of agreement and bias in proxy reports compared to direct observations and patient self-reports, and to evaluate the effect of proxy and subject characteristics on the degree of agreement and bias observed. Study results will be useful for determining the extent to which proxy respondents can be used for evaluating frequently asked questions about patient functioning and change in functioning. Additionally, data will be used to estimate the bias which may result from selecting particular proxy respondents based on knowledge about the type of information sought, and characteristics of the subject and proxy.
