Abstract 4143822: Trial emulation comparing percutaneous left atrial appendage occlusion to direct oral anticoagulants in Medicare beneficiaries

Background: Percutaneous left atrial appendage occlusion (pLAAO) is widely used procedure as an alternative to oral anticoagulants in patients with atrial fibrillation at increased risk of bleeding or prior bleeding events. The efficacy and safety of pLAAO in real-world Medicare patients remain uncertain. Objectives: To compare to pLAAO to DOACs in Medicare beneficiaries with atrial fibrillation. Methods: We used a 15% sample of Medicare claims data (2015-2019) to identify patients who received pLAAO and compared them to patients who received DOACs using trial emulation design. Relying on propensity score matching, we mimicked randomization and assigned patients to receive pLAAO or DOACs in a 1:3 ratio. Confounders in the propensity score model was selected by domain knowledge and machine learning approach (i.e., high dimensional propensity score). Patients were followed for up to 4 years. The primary efficacy outcome was a composite outcome of stroke, TIA or systemic thromboembolism leading to hospital admission. The primary safety outcome was any bleeding leading to hospital admission. We used a Cox proportional hazards regression model to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) for the outcomes between both treatment groups. Results: We identified 3,647 and 10,941, patients who underwent pLAAO and received DOACs, respectively. The primary composite endpoint was similar between pLAAO and DOACs (0.72%/ year vs 0.72%/ year, HR: 0.98, 95% CI: 0.72 – 1.34), figure 1A. The use of pLAAO was associated with significantly more bleeding (7.48%/ year vs 4.98%/ year, HR: 1.38, 95% CI: 1.24 – 1.54), figure 1B. Consistent with the clinical trials, all-cause mortality was not significantly different between both groups, (9.04%/ year with pLAAO vs 10.00%/ year with DOACs, HR: 0.98, 95% CI: 0.90 – 1.07). Conclusion: We successfully emulated the target trial using Medicare claims data. In Medicare beneficiaries with atrial fibrillation, pLAAO use compared to DOACs did not reduce the risk of stroke, TIA or systemic thromboembolism and was associated with more bleeding. The increase in bleeding events in the pLAAO arm is likely due to high upstream anticoagulant use after the procedure.