Outcome scoring systems for acute respiratory distress syndrome

Acute respiratory distress syndrome (ARDS) is commonly diagnosed in intensive care units (ICUs), often in association with acute kidney injury. In this study, we compared the predictive value of outcome scoring systems: Acute Physiology and Chronic Health Evaluation IV (APACHE IV), earlier APACHE models, Sequential Organ Failure Assessment (SOFA), the Risk of renal failure, Injury to the kidney, Failure of kidney function, Loss of kidney function, and End-stage renal failure (RIFLE) classification, and Acute Lung Injury score in critically ill patients with ARDS. We retrospectively abstracted data from the medical records of 135 critically ill ARDS patients in two medical ICUs of a tertiary care hospital from December 1999 to June 2006. Overall mortality rate was 65% (88/135). Forward conditional logistic regression identified APACHE IV, alveolar-arterial O₂ tension difference, age, sepsis, and maximum RIFLE (RIFLE_max) score on ICU days 1 and 3 to be independent predictors of hospital mortality. The area under the receiver operating characteristic curve for the APACHE IV score revealed good fit (Hosmer and Lemeshow goodness-of-fit test results) and discriminative power (area under the receiver operating characteristic curve, 0.792 ± 0.038; P < 0.001). The cumulative survival rates at 6-month follow-up after hospital discharge were significantly (P < 0.001) different among ARDS patients with APACHE IV mortality rate 35% or less and APACHE IV mortality rate higher than 35%. The APACHE IV score and RIFLE_max score are predictors of hospital mortality in ARDS patients, with APACHE IV demonstrating desirable properties of prognostic accuracy.