Background

Worldwide, almost 1 million patients undergo surgery every day and this number is still increasing. Cardiovascular complications occur in approximately 5% of patients after non-cardiac surgery and are a leading cause of morbidity and mortality.1 Early recognition and stratification of patients at high risk for such complications is desirable to direct healthcare towards the most vulnerable patients. Several clinical guidelines recommend the use of preoperative prediction models as risk stratification tools to predict postoperative outcomes such as in-hospital mortality or major adverse cardiac events (MACEs). The revised cardiac risk index (RCRI) is such a prediction model developed over 20 years ago.2 It preoperatively predicts in-hospital MACE using six predictors, namely high-risk surgery (ie, intraperitoneal, intrathoracic or supra-inguinal vascular surgery), history of ischaemic heart disease, history of congestive heart failure, history of cerebrovascular disease, insulin therapy for diabetes mellitus and preoperative serum creatinine ≥2 mg/dL. Although the predictive performance is moderate, especially in vascular surgery patients,3 the score is commonly used in daily practice.

Several studies have attempted to improve the prediction of MACE by either adding biomarkers to the RCRI or by comparing the predictive performance of the RCRI to single biomarkers or to other prediction models. We here summarise the findings of a recent comprehensive Cochrane Review4 that had various aims: (1) to outline which biomarkers have added predictive value to the RCRI to predict MACE or any other postoperative outcome after non-cardiac surgery, (2) to outline whether the predictive performance of biomarkers by itself differs from the predictive performance of the RCRI and (3) to outline whether the predictive performance of the RCRI differs from other prediction models for the prediction of MACE and other postoperative outcomes in noncardiac surgery patients.

Review methods

A comprehensive literature search was set up in Medline and Embase from 1999 (the …