THE SUPPORT PROGNOSTIC MODEL - OBJECTIVE ESTIMATES OF SURVIVAL FOR SERIOUSLY ILL HOSPITALIZED ADULTS

Objective: To develop and validate a prognostic model that estimates s urvival over a 180-day period for seriously ill hospitalized adults (p hase I of SUPPORT [Study to Understand Prognoses and Preferences for O utcomes and Risks of Treatments]) and to compare this model's predicti ons with those of an existing prognostic system and with physicians' i ndependent estimates (SUPPORT phase II).Design: Prospective cohort stu dy. Setting: 5 tertiary care academic centers in the United States. Pa rticipants: 4301 hospitalized adults were selected for phase I accordi ng to diagnosis and severity of illness; 4028 patients were evaluated from phase II.Measurements: A survival model was developed using the f ollowing predictor variables: diagnosis, age, number of days in the ho spital before study entry, presence of cancer, neurologic function, an d 11 physiologic measures recorded on day 3 after study entry. Physici ans were interviewed on day 3. Patients were followed for survival for 180 days after study entry. Results: The area under the receiver-oper ating characteristics (ROC) curve for prediction of surviving 180 days was 0.79 in phase 1, 0.78 in the phase II independent validation, and 0.78 when the acute physiology score from the APACHE (Acute Physiolog y, Age, Chronic Health Evaluation) III prognostic scoring system was s ubstituted for the SUPPORT physiology score. For phase II patients, th e SUPPORT model had equal discrimination and slightly improved calibra tion compared with physicians' estimates. Combining the SUPPORT model with physicians' estimates improved both predictive accuracy (ROC curv e area = 0.82) and the ability to identify patients with high probabil ities of survival or death. Conclusions: A limited amount of readily a vailable clinical information can provide a foundation for longterm su rvival estimates that are as accurate as physicians' estimates. The be st survival estimates combine an objective prognosis with a physician' s clinical estimate.