This article is more than 5 years old. Information may no longer be current.
TAVR risk models show moderate discrimination, excellent calibration
Click Here to Manage Email Alerts
Among patients at high risk for poor outcome following transcatheter aortic valve replacement, use of the TAVR Poor Outcome risk models showed moderate discrimination and excellent calibration across risk profiles and devices used, researchers reported in the Journal of the American College of Cardiology.
The researchers evaluated data on 2,830 participants who underwent TAVR as part of the CoreValve U.S. Pivotal Extreme and High Risk trials. Those included in the current study had severe symptomatic aortic stenosis, high or extreme 30-day mortality or morbidity risk, and underwent TAVR vs. surgery.
The researchers constructed the TAVR Poor Outcome risk models, which were two multivariable logistic models to identify patients at high risk for poor outcome. Each model assessed 25 candidate variables and used Harrell’s backward selection. Additionally, the researchers developed a 1-year model, which used a more conservative definition of poor outcome, as well as two reduced clinical models.
Two levels of poor outcome were defined. The worst outcome category was characterized by any of the following: death within 6 months of TAVR, very poor quality of life within 6 months after TAVR or moderate decline in quality of life from baseline to 6 months. The next level of poor outcome was defined as any of the following: death within 1 year of TAVR, poor QOL at 1 year after TAVR or moderate decline in QOL from baseline to 1 year of TAVR, according to the study
At 6 months after attempted TAVR, 882 patients (31.2%) had poor outcomes due to death (n = 498; 17.6%), very poor QOL (n = 328; 11.6%) or worsening of QOL (n = 56; 2%). Of the 2,325 patients with 1-year outcome data available, 1,181 (50.8%) had poor outcomes due to death (n = 703; 30.2%), poor QOL (n = 455; 19.6%) or worsening of QOL (n = 23; 1%), according to the findings.
The models showed moderate discrimination, with a C-index of 0.646, according to the findings. This was comparable to the C-index of 0.661 in the PARTNER derivation cohort. The models showed excellent calibration with experiential outcomes, with an intercept of 0.03, a slope of 0.87 and a coefficient of determination of 96%.
Frailty was observed in 59.8% of patients. In all models except the 1-year clinical model, frailty was linked to a 30% to 40% increase in the odds of a poor outcome when added to the existing models and, additionally, yielded a small but significant improvement in discrimination. Increases in C-indexes in the models were as follows: 0.043 (integrated discrimination improvement [IDI], P = .002) for the 6-month full model, 0.0044 (IDI, P = .001) for the 1-year full model, 0.0036 (IDI, P = .005) for the 6-month clinical model, and 0 (IDI, P = .218) for the 1-year clinical model. Analysis of the individual components of frailty demonstrated that unintentional weight loss was associated with a 50% increased odds of poor outcome, and disability was associated with a 20% increased odds of poor outcome.
“Future research into the usefulness of these models in clinical care is needed to further support their value in the care of patients with severe aortic stenosis,” the researchers concluded.
In a related editorial, Mathew R. Reynolds, MD, of the Harvard Clinical Research Institute, and Jonathan C. Hong, MD, of the division of cardiac surgery, University of British Columbia, Vancouver, noted that there is currently no TAVR risk model that perfectly meets all user needs. However, the risk models evaluated in this study show substantial promise, they wrote.
“There will always be an interest in mortality risk models for high-risk interventions such as TAVR, and some are now available from large and representative patient samples,” Reynolds and Hong wrote. “Arnold et al have shown that with some additional work, we can make some reasonable accurate predictions about survival with improved QOL after TAVR.” – by Jennifer Byrne
Disclosure: The CoreValve U.S. Pivotal Extreme and High Risk trials were sponsored by Medtronic. Please see the full study for a list of the researchers’ relevant financial disclosures.
Click Here to Manage Email Alerts