Abstract 13391: Enhancing Population-Based Risk Estimation with Institution-Calibrated Models: Implications for Quality Improvement and TAVR
Objective: Large population-based risk models, like those of STS, are valuable for clinical practice benchmarking, but are being used to predict individual patient risk for TAVR vs. Surgical repair. For this, their institution-specific accuracy is unknown. To improve accuracy of risk estimation for new therapies like transcatheter aortic valve replacement (TAVR) and better target institution-specific areas for quality improvement we investigated 1) accuracy of the Society of Thoracic Surgeons (STS) models to predict institution-specific risk after aortic valve replacement (AVR), 2) contrast strength of risk factors between those of STS models and the institution, and 3) additional risk factors not accounted for by STS models.
Methods: From 1/2000-1/2010, 4,971 patients underwent AVR, isolated (n=2,660) or with coronary artery bypass grafting (AVR+CABG; n= 2,311). Hospital mortality predicted by STS risk models was compared with observed mortality. Differences between STS model coefficients (risk factor strength) and those specific to our institution were calculated using logistic regression. After adjusting for patients’ STS score, additional risk factors were sought.
Results: Observed mortality was less than predicted for AVR (1.2% vs. 2.9%, P<.0001) and AVR+CABG (2.3% vs. 4.9%, P<.0001). The trend of underestimation and over estimation was also seen for other STS defined end points. In many instances, risk factor strength differed substantially from that of STS models (Figure). Hepatic dysfunction (elevated bilirubin), renal failure (higher creatinine) and some other factors were among additional institution-specific risk factors.
Conclusions: Use of institution-calibrated and enhanced models may be more valuable than unmodified STS models to improve accuracy of risk estimation for procedures like TAVR and to target institution specific areas for quality improvement.
- © 2012 by American Heart Association, Inc.