Validation of severity scoring systems SAPS II and APACHE II in a single-center population

Objective: To validate two severity scoring systems, the Simplified Acute P hysiology Score (SAPS II) and Acute Physiology and Chronic Health Evaluatio n (APACHE II), in a single-center ICU population. Design and setting: Prospective data collection in a two four-bed multidisc iplinary ICUs of a teaching hospital. Patients and methods: Data were collected in ICU over 4 years on 1721 conse cutively admitted patients (aged 18 years or older, no transferrals, ICU st ay at least 24 h) regarding SAPS II, APACHE II, predicted hospital mortalit y, and survival upon hospital discharge. Results: At the predicted risk of 0.5, sensitivity was 39.4 % for SAPS II a nd 31.6 % for APACHE II, specificity 95.6 % and 97.2 %, and correct classif ication rate 85.6 % and 85.5 %, respectively. The area under the ROC curve was higher than 0.8 for both models. The goodness-of-fit statistic showed n o significant difference between observed and predicted hospital mortality (H = 7.62 for SAPS II, H = 3.87 for APACHE II; and C = 9.32 and C = 5.05, r espectively). Observed hospital mortality of patients with risk of death hi gher than 60 % was overpredicted by SAPS II and underpredicted by APACHE II . The observed hospital mortality was significantly higher than that predic ted by the models in medical patients and in those admitted from the ward. Conclusions: This study validates both SAPS II and APACHE II scores in an I CU population comprised mainly of surgical patients. The type of ICU admiss ion and the location in the hospital before ICU admission influence the pre dictive ability of the models.