The Association Between Hospital Comprehensive Stroke Center Designation Status and Stroke Rate Reporting in the Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy Registry After Transcatheter Aortic Valve Replacement: Implications for Hospital Performance Assessment.

Abstract

Background: The Michigan Structural Heart Consortium showed that transcatheter aortic valve replacement (TAVR) stroke rates were associated with hospital Comprehensive Stroke Center (CSC) designation, suggesting that hospital structural factors affect stroke reporting. We sought to determine whether CSC designation is nationally associated with reported stroke rates and the potential impact on hospital performance and Transcatheter Valve Therapy National Data Registry (TVTR) public reporting.

Methods: Consecutive TVTR cases from January 2016 through June 2023 were included, excluding sites reporting to the TVTR <1 year or performing <20 total TAVR procedures. The primary outcome was in-hospital stroke rate. Secondary outcomes were disabling stroke and 30-day stroke. Major bleeding, death, acute renal failure, and a composite end point were falsification end points to determine whether CSC status was uniquely associated with stroke. Multivariable hierarchical models were created starting with patient covariates and sequentially adding hospital variables to account for determinants of stroke. The impact of CSC status on hospital performance was estimated by correcting CSC hospital stroke rates in the validated TVTR 30-day composite model.

Results: A total of 525 919 TAVRs at 800 sites were included, and 10 963 strokes (2.1%) were reported within 30 days: 8284 (1.6%) in-hospital and 2679 (0.5%) after discharge. A total of 4733 (0.9%) were disabling strokes. Multivariate analysis showed that CSC status and participation in TAVR trials were independent predictors of in-hospital stroke rate (P=0.033 and 0.012) and disabling stroke (P=0.033 and 0.007), but not falsification end points (P=NS). Participation in TAVR trials and operator volume were independent predictors of 30-day stroke (P=0.039 and P=0.041). The impact of CSC status on the 30-day composite model showed performance improvement for 8 of 153 CSC hospitals.

Conclusions: CSC hospital status and participation in TAVR trials are significant predictors unique for stroke. Correcting for the association of CSC status on stroke rate resulted in an estimated 8 of 153 hospitals improving their TVTR 30-day composite model performance. Further work is needed to account for potential ascertainment bias for stroke in public reporting.