Prediction of Perioperative Risk: The Glass May Be Three-Quarters Full

Patients who undergo noncardiac surgery may be at risk for adverse cardiovascular events in the perioperative period. Many studies have shown that readily accessible clinical markers can be used to identify high-risk patients. Such markers typically include factors that are specific to the patient as well as to the planned surgical procedure. When predicting cardiovascular risk, the consulting physician seeks to identify patients who have increased perioperative and long-term risk for adverse events. He or she then tries to decrease short-term and long-term risk by instituting appropriate therapy. Several algorithms have been developed to quantify risk, and although they differ, they also have much in common.

In this issue, Gilbert and colleagues (1) report on the relatively low accuracy of existing perioperative risk indices in predicting adverse cardiac events. The authors examined clinical indicators of perioperative risk in 2035 patients who were referred for preoperative internal medicine consultation before urgent or elective noncardiac surgical procedures at two hospitals. The patients were followed after surgery by the consulting internist, who recorded adverse events. These were defined as unstable angina, myocardial infarction, acute pulmonary edema, and death. Preoperative risk factors were used to assign strata of cardiovascular risk to each patient by using each of four schemes: the 1963 American Society of Anesthesiology classification of physical status (2); the 1977 Goldman multifactorial index of cardiovascular risk (3); the 1986 Detsky index (4), as modified in 1997 by the American College of Physicians (5); and the 1976 Canadian Cardiovascular Society index for grading angina pectoris (6). Receiver-operating characteristic (ROC) curves were …