This site is not optimized for Internet Explorer 8 (or older).

Please upgrade to a newer version of Internet Explorer or use an alternate browser such as Chrome or Firefox.

Predicting Quality of Life at 1 Year After Transcatheter Aortic Valve Replacement in a Real-World Population

Sunday, October 14, 2018

Submitted by


Source Name: Circulation: Cardiovascular Quality and Outcomes


Suzanne V. Arnold, David J. Cohen, David Dai, Philip G. Jones, Fan Li, Laine Thomas, Suzanne J. Baron, Naftali Z. Frankel, Susan Strong, Roland A. Matsouaka, Fred H. Edwards, J. Matthew Brennan

Despite the benefit for most patients, some patients have poor outcome after transcatheter aortic valve implantation (TAVI). Currently it is hard to predict which patient is predisposed to poor outcome at one year. This paper investigates the performance of a previously developed prediction model of poor outcome after TAVI.

Arnold and colleagues previously built a prediction model of poor outcome using data from high-risk TAVI trials. The model included the preoperative Kansas City Cardiomyopathy Questionnaire (KCCQ), mean aortic valve gradient, usage of home oxygen, creatinine level, atrium fibrillation, and atrial fibrillation. Poor outcome was defined as death, poor quality of life (KCCQ-OS <60), or moderate worsening in quality of life (decrease of >10 points in KCCQ-OS) at one year. In the current paper they set out to validate the model in lower-risk real-world dataset of >13000 TAVI patients from the TVT registry. The model was validated based on discrimination and calibration, and was recalibrated for this real-world population.

Poor outcome decreased from 42.0% in 2012 to 37.8% in 2015. Initially, the model performed poorly with moderate discrimination but poor calibration. After recalibration, the model performed well with a C index of 0.65, but excellent calibration.

The recalibrated model can be used to identify patients who are less likely to benefit from TAVI and to help patients prepare for the recovery phase after the procedure.

Unfortunately, risk models are inherently imperfect, as was shown by the only moderate discrimination. This means it cannot be used to accurately predict who will or will not have a poor outcome. Rather, it helps to estimate the chances of good recovery. This information will be useful for patients in order to prepare for the recovery and have realistic expectations.

Add comment

Log in or register to post comments