(P-464) Transcatheter and Surgical Aortic Valve Replacement: Patient Characteristics and Outcomes

Background: Transcatheter and surgical aortic valve replacement (TAVR and SAVR, respectively) are medical device procedures for patients with aortic valve stenosis or insufficiency (AVSI). TAVR is less invasive and generally preferred for patients prone to post-surgical complications.

Objectives: The objectives of this research are (1) to characterize differences in AVSI patients who undergo TAVR and SAVR and (2) to compare post-procedural occurrence of death, atrial fibrillation (AF), and adverse bleeding events (ABE) between TAVR and SAVR patients in the real-world hospital setting.

Methods: Patients from a large health system in the OMNY Health Device Database with ICD-10 codes for AVSI were indexed at TAVR or SAVR procedure date. Patients were characterized at index. Propensity score (PS) models were constructed separately for each outcome using variables independently associated with the outcome. Time-to-event analyses and Cox regression were employed to derive unadjusted and PS overlap weight (PSOW)-adjusted hazard ratios (HR) and confidence intervals (CI) for each outcome. Analyses were repeated including only patients with aortic valve stenosis (AVS).

Results: A total of 268 patients (82 TAVR, 186 SAVR) were included. Compared to SAVR patients, TAVR patients were more likely to be female (61% vs. 30%), healthy weight (37% vs. 19%), ≥ 70 years old (85% vs. 35%), have a history of cancer (11% vs. 4%), congestive heart failure (96% vs. 32%), chronic obstructive pulmonary disease (20% vs. 9%), frailty (41% vs. 1%), gastrointestinal bleeding (32% vs. 4%), hypertension (59% vs. 41%), pacemaker use (15% vs. 9%), mitral valve (15% vs. 8%), peripheral artery disease (23% vs. 5%), pneumonia (11% vs. 3%), and less likely to have had coronary artery bypass grafting (12% vs. 28%). PS models were highly discriminative (c-index > 0.9) indicating minimal overlap between cohorts. As such, PSOW-adjusted HR estimates (TAVR vs. SAVR) exhibited large degrees of variability for death (HR: 0.67; 95% CI: 0.16-2.87), AF (HR: 0.94; 95% CI: 0.18-4.79), and ABE (HR: 1.52; 95% CI: 0.12-19.2). Results were similar for AVS patients.

Conclusions: These results help provide insights into which factors are associated with the receipt of TAVR vs. SAVR among AVSI patients in the real-world hospital setting. Comparative analyses were limited by the great degree of underlying differences between TAVR and SAVR cohorts, which manifested themselves as highly reduced effective sample sizes derived from the PSOWs. Additional analyses with an amplified number of patients in the common support region would be insightful to understand the relative performance of TAVR and SAVR on real-world outcomes.