The following is the summary of “Predicting short-term outcomes after transcatheter aortic valve replacement for aortic stenosis” published in the February 2023 issue of Heart by Savitz, et al.

Aortic stenosis transcatheter aortic valve replacement (TAVR) has seen rapid growth in the number of patients who qualify for treatment in recent years. However, there are still questions about how to properly define risk for subpar clinical and patient-centered outcomes. Their mission was to create models that predict clinical and patient-centered outcomes 30 days following TAVR in a large, varied community-based population. At Kaiser Permanente Northern California, an integrated healthcare delivery system, researchers tracked all adult patients who had TAVR between 2013 and 2019 for the following 30 day outcomes: death from any cause, improvement in quality of life, hospitalization for any reason, Emergency departments (ED) visit for any reason, hospitalization for heart failure, and ED visit for heart failure (HF). 

Investigators used the Society of Thoracic Surgeons (STS)/American College of Cardiology (ACC) TVT Registry and electronic health records to create prediction models utilizing gradient boosting machines for linked demographic, clinical, and other data. The area under the curve (AUC) for model discrimination and related calibration plots was used to determine the accuracy of our models. Logistic regression was used to assess the relationship between predictors and quality of life, whereas Cox proportional hazards regression was used to assess the relationship between predictors and all other outcomes. They found 1,564 people who qualified for TAVR and gave them the procedure. Between 1.3% (HF hospitalizations) and 15.3% (other bad events), the chances of poor 30-day post-TAVR outcomes varied widely (all-cause ED visits). Discrimination was only moderate for death (AUC 0.60) and quality of life (AUC 0.62) in the most discriminative models, but it was better for HF-related ED visits (AUC 0.76).

Additionally, calibration was outcome dependent. Crucially, the STS risk score only predicted death and all-cause hospitalization independently. Older age was the only independent predictor of ED visits due to HF, and there was no significant association between race/ethnicity and any outcome. Predicting short-term clinical and patient-centered outcomes after TAVR remains difficult despite the use of extensive data from the STS/ACC TVT Registry and electronic health records. Better predictors of post-TAVR outcomes are needed to facilitate individualized therapeutic decision making and monitoring measures, therefore they should continue their work in this direction.