Related Experiment Video
Updated: Aug 6, 2026
Technique and Patient Selection Criteria of Right Anterior Mini-Thoracotomy for Minimal Access Aortic Valve Replacement
Published on: March 26, 2018
The Society of Thoracic Surgeons Risk Model for Surgical Aortic Valve Replacement After Transcatheter Aortic Valve
Robert B Hawkins1, Levi Bonnell2, J Hunter Mehaffey3
1Department of Cardiac Surgery, University of Michigan, Ann Arbor, Michigan.
Background:
Recent evidence suggests cardiac surgery after transcatheter aortic valve replacement (TAVR) carries additional risk. Herein we describe validation of a risk model for surgical aortic valve replacement (SAVR) after prior TAVR and risk distribution over time.
Methods:
Patients undergoing SAVR (7/2014-3/2025) were extracted from the Society of Thoracic Surgeons (STS) Adult Cardiac Surgery Database. Patients were excluded for no prior TAVR, concomitant arch/descending aortic procedures, other major surgery, or missing operative mortality data. Patients undergoing proximal aortic or other valve surgery were included. De novo, parsimonious models were built for all 8 STS outcomes.
Results:
Among 651,398 SAVR patients during the study period, 5,708 had SAVR after TAVR with 2,298 (40.3%) isolated SAVR-after-TAVR and 3,410 (59.7%) concomitant procedures. The risk model for SAVR-after-TAVR had excellent discrimination for operative mortality (AUC 0.82) and morbidity or mortality (AUC 0.73). Calibration was excellent across risk deciles for all demographic and surgical subgroups. Of isolated SAVR-after-TAVR cases that might qualify for redo-TAVR (excluding endocarditis, root enlargement and emergent cases), 53.3% (753/1,334) had predicted risk of mortality <3%, and the operative mortality rate decreased over time from 13.1% to 3.5%. Given improvement over time, era was included as an interaction term in the model for isolated SAVR cases, leading to stable observed-to-expected ratios.
Conclusions:
Mortality rates for SAVR-after-TAVR have decreased. The risk model has excellent performance and demonstrates a large cohort of TAVR explant patients can be classified as low risk. The new risk model allows for accurate risk estimates to inform clinical decision-making.
