Related Experiment Video
Updated: Sep 16, 2025
06:02
Upper-extremity Approach for Secondary Access in Transfemoral Transcatheter Aortic Valve Implantation
Published on: August 8, 2025
186
External Validation of Mortality Prediction Models in Japanese Transcatheter Aortic Valve Replacement Registry
Ryo Shibata1, Tomotsugu Seki1,2, Yuki Takeda1
1Department of Cardiology, Kyoto Prefectural University of Medicine, Kyoto, 602-8566, Japan.
Summary
The Society of Thoracic Surgeons Predicted Risk of Mortality (STS-PRoM) and Netherlands Heart Registration (NHR) models showed the best short-term performance in Japanese transcatheter aortic valve replacement (TAVR) patients. The Optimized CathEter vAlvular iNtervention-TAVI (OCEAN-TAVI) model demonstrated the best long-term performance.
Area of Science:
- Cardiovascular Medicine
- Medical Statistics
- Health Services Research
Background:
- Mortality prediction models (MPMs) are vital for assessing transcatheter aortic valve replacement (TAVR) risk.
- The external validity of existing MPMs in Japanese TAVR populations is not well-established.
- This study aimed to validate short-term and long-term MPMs in Japanese TAVR patients.
Purpose of the Study:
- To evaluate the performance of established short-term (30-day) and long-term (1-year) mortality prediction models in a Japanese cohort undergoing TAVR.
- To assess the discrimination and calibration of these models using data from a prospective multicentre registry.
- To identify the most accurate models for risk stratification in this specific patient population.
Main Methods:
- Analysis of 1756 patients for short-term and 1235 patients for long-term mortality from the KPUM TAVI registry (2016-2023).
- Validation of five short-term MPMs (STS-PRoM, FRANCE-2, OBSERVANT, ACC-TAVI, NHR) and three long-term MPMs (OCEAN-TAVI, Osaka University, TARI).
- Performance assessment using Area Under the Receiver Operating Characteristic Curve (AU-ROC) for discrimination and calibration metrics; intercept recalibration applied to short-term models.
Main Results:
- Short-term mortality was 1.4% (25/1756), and long-term mortality was 14.5% (179/1235).
- For short-term mortality, NHR (AU-ROC: 0.82) and STS-PRoM (0.78) exhibited the highest discrimination; all models initially overestimated risk, improving post-recalibration.
- For long-term mortality, OCEAN-TAVI and Osaka University models showed the highest discrimination (AU-ROC: 0.75), with OCEAN-TAVI demonstrating superior calibration.
Conclusions:
- The STS-PRoM and NHR models demonstrated superior short-term predictive performance in Japanese TAVR patients following recalibration.
- The OCEAN-TAVI model exhibited the best overall long-term performance and calibration in this cohort.
- Future research should investigate machine learning-based models to enhance the accuracy and clinical utility of TAVR risk prediction.
