Difficulties in predicting outcome in cardiac surgery patients

OBJECTIVE: To evaluate a novel combination of preoperative, intraoperative, and postoperative variables (including the Parsonnet, and the Acute Physiology and Chronic Health Evaluation II and III [APACHE II and III] scores) in cardiac surgery patients in order to predict hospital outcome, complications, and length of stay. DESIGN: Prospective survey. SETTING: Adult intensive care unit (ICU) at a tertiary care cardiothoracic surgery center. PATIENTS: All cardiac surgery patients admitted to the ICU over a 1-yr period. INTERVENTIONS: Medical history, Parsonnet score, intraoperative data (including bypass and ischemic times), APACHE II and III scores, complications, and outcome were collected for each patient. MEASUREMENTS AND MAIN RESULTS: One thousand eight patients were entered into the study. The mean Parsonnet score was 7.8 (range 0 to 33), mean APACHE II score 11.8 (range 2 to 33), and mean APACHE III score 42.5 (range 9 to 132). ICU mortality rate was 2.7% and hospital mortality rate was 3.8%. The mean APACHE II predicted risk of dying was 5.31%, which gave a standardized mortality ratio of 0.71. The above scores were all statistically well correlated with hospital mortality. Further, a logistic regression model was developed for the probability of hospital death. This model (which included bypass time, need for inotropes, mean arterial pressure, urea, and Glasgow Coma Scale) had an area under the receiver operating characteristic curve of 0.87, while the Parsonnet score had an area of 0.82 and the APACHE II risk of dying had an area of 0.84. CONCLUSIONS: Cardiac surgery remains a difficult area for outcome prediction. A combination of intraoperative and postoperative variables can improve predictive ability.