In older persons, disability is common, and greatly affects health and well-being. This disability is often precipitated by acute illness leading to hospitalization, and substantial evidence indicates that it may be preventable or reversible. The broad objective of this proposal is to develop multidimensional prognostic indices that will stratify hospitalized older medical patients according to risk for new-onset disability. We will analyze data on 1656 patients,age >70 who were independent in activities of daily living (ADL) and who were hospitalized for acute illness and followed prospectivelyfor 1 year. The data include unique information about multiple conceptual domains of risk (biologic, neuropsychiatric, and socioeconomic) clearly predictive of functional decline. The proposal has three specific aims: (1) To develop and validate prognostic indices of patients' risk for disability at discharge, or death; (2) In patients with disability at discharge, To understand trajectories of functioning over time, and to develop and validate prognostic indices of their risk for continued disability or death one year later; and (3) In patients without disability at discharge, To understand trajectories of functioning over time, and to develop and validate prognostic indices of their risk for new-onset disability or death one year later. The proposed study will use a prospective cohort design. Three separate nested logit models will be constructed, permitting prediction of independence, disability, and death. Specific strategies will be used to ensure the robustness of the models, including model reduction using a backward stepwise approach analyzing variables in specific risk domains. Overall probabilities of disability and death at one year will be calculated by multiplying the probabilities from the model of Specific Aim 1 and the conditional models from Aims 2 and 3. Cluster resampled (by subject) bootstrapping will be used for a model validation and assessment, including assessment of stability of predictors for inclusion in the indices, and for validation of the discriminatory ability of the models. The proposed study and its rationale are directly related to the mission of the National Institute on Aging and relevant to public health. The research addresses important research questions regarding the progression and determinants of change that affect health in old age, variability in rates of change with age in function, and effects of age and comorbidity on the effects of other risk factors. The results will permit us to identify older persons at higher risk of becoming disabled during and after hospitalization, facilitating targeted interventions to prevent or ameliorate this disability.
