Objective: To compare the ability of two methods--Acute Physiology and Chronic Health Evaluation (APACHE II) and Mortality Prediction Model (MPM)--to predict hospital outcome for a large group of intensive care patients from Britain and Ireland.
Design: Prospective, multicenter, cohort study.
Setting: Twenty-six general intensive care units in Britain and Ireland.
Patients: A total of 8,724 patients admitted to the study.
Interventions: None.
Measurements and main results: Probabilities of hospital death for patients were estimated by applying APACHE II and MPM. Predicted risks of hospital death were compared with observed outcomes using four methods of assessing the overall goodness of fit. APACHE II performed better than MPM; the calibration curve for APACHE II lay closer to the line of perfect predictive ability. Lemeshow-Hosmer chi-square statistics were 81 and 57 for APACHE II, and 2515 and 1737 for MPM. The overall correct classification rate for APACHE II was 79%, and this classification rate was 74% for MPM, applying a decision criterion of 50%. The area under the receiver operating characteristic curve was 0.83 with APACHE II and 0.74 with MPM. Even after modifications to the MPM for the assessment of coma, the performance of APACHE II was superior.
Conclusions: APACHE II demonstrated a higher degree of overall goodness of fit, which was superior to MPM for groups of intensive care patients from Britain and Ireland.