Design: Two prospective cohort studies, in tandem. The predictive model developed in the initial cohort was subsequently validated in a separate cohort.

Setting: General medical wards of a university teaching hospital.

Patients: For the development cohort, 188 hospitalized general medical patients aged ≥70 years. For the validation cohort, 142 comparable patients.

Measurement and main results: The subjects and their nurses were interviewed twice weekly using standardized, validated instruments. Functional decline occurred among 51/188 (27%) patients in the development cohort. Four independent baseline risk factors (RFs) for functional decline were identified: decubitus ulcer (adjusted relative risk [RR] 2.7; 95% confidence interval [CI] 1.4, 5.2); cognitive impairment (RR 1.7; CI 0.9, 3.1); functional impairment (RR 1.8; CI 1.0, 3.3); and low social activity level (RR2.4; CI 1.2, 5.1). A risk-stratification system was developed by adding the numbers of RFs. Rates of functional decline for the low- (0 RF), intermediate- (1–2 RFs), and high- (3–4 RFs) risk groups were 8%, 28%, and 63%, respectively (p<0.0001).The corresponding rates in the validation cohort, of whom 34/142 (24%) developed functional decline, were 6%, 29%, and 83% (p<0.0001). The rates of death or nursing home placement, clinical outcomes associated with functional decline in the hospital, were 6%, 19%, and 41% (p<0.002) in the development cohort and 10%, 32%, and 67% (p<0.001) in the validation cohort, respectively, for the three risk groups.

Conclusions: Functional decline among hospitalized elderly patients is common, and a simple predictive model based on four risk factors can be used on admission to identify elderly persons at greatest risk.