Surgical risk models estimate operative outcomes while controlling for heterogeneity in 'case mix' within and between institutions. In cardiac surgery, risk models are used for patient counselling, surgical decision-making, clinical research, quality assurance and improvement, and financial reimbursement. Importantly, risk models are only as good as the databases from which they are derived; physicians and investigators should, therefore, be aware of shortcomings of clinical and administrative databases used for modelling risk estimates. The most frequently modelled outcome in cardiac surgery is 30-day mortality. However, results of randomized trials to compare conventional surgery versus transcatheter aortic valve implantation (TAVI) indicate attrition of surgical patients at 2-4 months postoperatively, suggesting that 3-month survival or mortality might be an appropriate procedural end point worth modelling. Risk models are increasingly used to identify patients who might be better-suited for TAVI. However, the appropriateness of available statistical models in this application is controversial, particularly given the tendency of risk models to misestimate operative mortality in high-risk patient subsets. Incorporation of new risk factors (such as previous mediastinal radiation, liver failure, and frailty) in future surgical or interventional risk-prediction tools might enhance model performance, and thereby optimize patient selection for TAVI.
ASJC Scopus subject areas
- Cardiology and Cardiovascular Medicine