Paper
For many elderly patients, an acute medical illness requiring hospitalization is followed by a progressive decline, resulting in high rates of mortality in this population during the year following discharge. However, few prognostic indices have focused on predicting posthospital mortality in older adults.To develop and validate a prognostic index for 1 year mortality of older adults after hospital discharge using information readily available at discharge.Data analyses derived from 2 prospective studies with 1-year of follow-up, conducted in 1993 through 1997.We developed the prognostic index in 1495 patients aged at least 70 years who were discharged from a general medical service at a tertiary care hospital (mean age, 81 years; 67% female) and validated it in 1427 patients discharged from a separate community teaching hospital (mean age, 79 years; 61% female).Prediction of 1-year mortality using risk factors such as demographic characteristics, activities of daily living (ADL) dependency, comorbid conditions, length of hospital stay, and laboratory measurements.In the derivation cohort, 6 independent risk factors for mortality were identified and weighted using logistic regression: male sex (1 point); number of dependent ADLs at discharge (1-4 ADLs, 2 points; all 5 ADLs, 5 points); congestive heart failure (2 points); cancer (solitary, 3 points; metastatic, 8 points); creatinine level higher than 3.0 mg/dL (265 micromol/L) (2 points); and low albumin level (3.0-3.4 g/dL, 1 point; 6 points). In the validation cohort, 1-year mortality was 4% in the lowest-risk group, 19% in the group with 2 or 3 points, 34% in the group with 4 to 6 points, and 64% in the highest-risk group. The area under the receiver operating characteristic curve for the point system was 0.75 in the derivation cohort and 0.79 in the validation cohort.Our prognostic index, which used 6 risk factors known at discharge and a simple additive point system to stratify medical patients 70 years or older according to 1-year mortality after hospitalization, had good discrimination and calibration and generalized well in an independent sample of patients at a different site. These characteristics suggest that our index may be useful for clinical care and risk adjustment.