Comparison of the APACHE II and APACHE III scoring systems in patients with respiratory failure in a medical intensive care unit.

BACKGROUND AND PURPOSE This retrospective study compared the capability of the Acute Physiology and Chronic Health Evaluation (APACHE) II and APACHE III scoring systems to predict outcome and determined the independent predictors of survival in these scoring systems for patients with respiratory failure in a medical intensive care unit (ICU). MATERIALS AND METHODS Seven hundred and eight patients with respiratory failure admitted to the medical ICU throughout a 9-year period were studied. Patients with an ICU stay of less than 24 hours, patients under 12 years of age, and burn and surgery patients were excluded. APACHE scores were calculated at 24 hours after admission. Student's t-test was used to compare the total APACHE scores of survivor and non-survivor groups. Multivariate logistic regression analysis was used to determine which variables were predictors of mortality. The discriminative power of APACHE scores to predict in-hospital mortality was studied by the area under the receiver operating characteristic curves of the APACHE II and APACHE III systems, respectively. RESULTS Both systems showed a significant association between higher scores and higher mortality. The APACHE II system under-predicted the actual hospital mortality rate. The APACHE III systems had a higher discriminative power (area 0.7462) than the APACHE II systems (area 0.6856; p < 0.05). The independent predictors of survival as assessed by APACHE II and III systems were respiratory rate, arterial oxygen pressure, oxygen gradient between alveoli and artery, serum creatinine concentration, and the presence of neurologic abnormalities. CONCLUSIONS The APACHE III systems has greater discriminative power than the APACHE II systems for predicting in-hospital mortality. The variables of oxygenation, mean artery pressure, respiratory rate, serum creatinine concentration, and Glasgow Coma Scale play important roles in predicting survival for patients with respiratory failure.