Predictors of survival following cardiac arrest in patients undergoing noncardiac surgery: a study of 518,294 patients at a tertiary referral center.

BACKGROUND The authors determined the incidence of cardiac arrest and predictors of survival following perioperative cardiac arrest in a large population of patients at a tertiary referral center. METHODS Medical records of patients who experienced cardiac arrest in the perioperative period surrounding noncardiac surgery between January 1, 1990, and December 31, 2000, were reviewed. Logistic regression identified characteristics associated with immediate (>or= 1 h) and hospital survival, with P <or= 0.01 considered statistically significant. RESULTS Cardiac arrest occurred in 223 of 518,294 anesthetics (4.3 per 10,000) during the study period. Frequency of arrest for patients receiving general anesthesia decreased over time (7.8 per 10,000 during 1990-1992; 3.2 per 10,000 during 1998-2000). The frequency of arrest during regional anesthesia (1.5 per 10,000) and monitored anesthesia care (0.7 per 10,000) remained consistent. Immediate survival after arrest was 46.6%, and hospital survival was 34.5%. Twenty-four patients (0.5 per 10,000) had cardiac arrest related primarily to anesthesia. From multivariate analysis, patients who experienced arrest due to bleeding were less likely to survive hospitalization (P = 0.001). Survival was also lower for patients who experienced arrest during nonstandard working hours (P = 0.006) and for patients who had protracted hypotension before arrest (P < 0.001). CONCLUSIONS The overall frequency of arrest for patients receiving anesthesia decreased during the study period. Most arrests were not due to anesthesia-related causes, and most patients experiencing anesthesia-related arrest survived to hospital discharge. Although many factors determining survival may not be amenable to modification, the fact that arrests during nonregular working hours had worse outcomes may indicate that the availability of human resources influences survival.