USING SEVERITY MEASURES TO PREDICT THE LIKELIHOOD OF DEATH FOR PNEUMONIA INPATIENTS

OBJECTIVE: To see whether predictions of patients' likelihood of dying in-hospital differed among severity methods. DESIGN: Retrospective co hort. PATIENTS: 18,016 persons 18 years of age and older managed medic ally for pneunonia; 1,732 (9.6%) in-hospital deaths. METHODS: Probabil ity of death was calculated for each patient using logistic regression with age, age squared, sex, and each of five severity measures as the independent variables: 1) admission MedisGroups probability of death scores; 2) scores based on 17 admission physiologic variables; 3) Dise ase Staging's probability of mortality model; the Severity Score of Pa tient Management Categories (PMCs); 4) and the All Patient Refined Dia gnosis-Related Groups (APR-DRGs). Patients were ranked by calculated p robability of death; 5) rankings were compared across severity methods . Frequencies of 14 clinical findings considered poor prognostic indic ators in pneumonia were examined for patients ranked differently by di fferent methods. RESULTS: MedisGroups and the physiology score predict ed a similar likelihood of death for 89.2% of patients. In contrast, t he three code-based severity methods rated over 25% of patients differ ently by predicted likelihood of death when compared with the rankings of the two clinical data-based methods (MedisGroups and the physiolog y score). MedisGroups and the physiology score demonstrated better cli nical credibility than the three severity methods based on discharge a bstract data. CONCLUSIONS: Some pairs of severity measures ranked over 25% of patients very differently by predicted probability of death. R esults of outcomes studies may vary depending on which severity method is used for risk adjustment.