To the Editor,
Bleeding is a major driver of morbidity and mortality following transcatheter aortic valve replacement (TAVR), ,, particularly among patients with atrial fibrillation. , Observational analyses have hinted at distinct bleeding profiles across direct oral anticoagulants (DOACs), but comparative evidence specific to post-TAVR populations is sparse. To address this gap, we examined major bleeding outcomes associated with apixaban versus rivaroxaban in a large, multicenter, real-world cohort.
We conducted a retrospective cohort study using the TriNetX research network. Adults who underwent TAVR from 2010 to 2024 and were discharged on either apixaban or rivaroxaban for AF were included. Individuals who transitioned between agents, or who received warfarin at discharge were excluded. Propensity-score matching (1:1) was performed across 75 demographic, clinical, laboratory, and treatment variables. Postmatching covariate balance was robust, with standardized differences <10% for all variables, yielding 2,157 patients in each treatment group.
The primary endpoint was major bleeding. Secondary endpoints included all-cause mortality, ischemic stroke, intracranial hemorrhage, thromboembolism, mechanical valve–related complications, and cardiovascular or heart-failure hospitalization. Median follow-up was 1.9 IQR 2.8. Time-to-event analyses were conducted using Cox proportional hazards models.
Before matching, patients receiving rivaroxaban were slightly younger and more often male, with higher rates of permanent atrial fibrillation and prior myocardial infarction. In contrast, the apixaban group had a greater burden of comorbidities, including chronic kidney disease, heart failure, and prior ischemic stroke, as well as lower estimated glomerular filtration rate. All baseline characteristics achieved standardized differences below 10%.
Major bleeding occurred more often with rivaroxaban (592 events, 8.48%/year) than with apixaban (486 events, 7.76%/year), yielding a significantly increased risk (HR [hazard ratio] 1.14, 95% CI [confidence interval] 1.01 to 1.29, p = 0.03). Rates of intracranial hemorrhage were similar (HR 0.97, p = 0.85). No significant differences were observed in all-cause mortality (HR 0.97, p = 0.68), ischemic stroke (HR 0.96, p = 0.65), thromboembolism (HR 0.99, p = 0.86), AF–related encounters (HR 1.02, p = 0.70), or cardiovascular hospitalization (HR 0.93, p = 0.10) ( Figure 1 ).
Propensity score–matched analysis of patients undergoing TAVR from the TriNetX database (n = 2,157 per group). Over a median follow-up of 1.9 years, apixaban was associated with similar rates of death, ischemic stroke, and thromboembolism, but lower major bleeding compared with rivaroxaban, with no difference in intracranial hemorrhage or cardiovascular-related hospitalization.
In this large, propensity-matched cohort, apixaban was associated with a significantly lower risk of major bleeding than rivaroxaban among TAVR patients, with no observed differences in ischemic or thromboembolic events. These findings align with prior data in AF populations suggesting a more favorable gastrointestinal bleeding profile with apixaban.
Several pharmacologic factors may underlie the divergent bleeding risks. Rivaroxaban’s once-daily dosing produces higher peak anticoagulant levels, whereas apixaban’s twice-daily regimen yields more consistent plasma concentrations. Rivaroxaban also demonstrates greater bioavailability and extended luminal exposure contact time within the gastrointestinal tract, which may contribute to mucosal injury. Variability in factor Xa inhibition kinetics, renal elimination, or patient selection may further modulate risk.
Major bleeding after TAVR imposes substantial clinical and economic burden. Prior costs analyses estimate direct medical expenditures of approximately $10,000 to $43,374 per major bleeding event, with clinically relevant nonmajor bleeding adding $1,000 to $5,000 in outpatient resource use. Although the observed lower bleeding risk with apixaban is consistent with data from the broader atrial fibrillation population, comparative evidence specific to patients undergoing TAVR has remained limited. In this high-risk setting, where advanced age, frailty, and comorbidity burden amplify bleeding susceptibility, our findings provide clinically relevant, procedure specific data. Given the 14% relative reduction in major bleeding observed with apixaban, averting 100 major bleeding events per 10,000 treated patients could translate into $1 to 4.5 million in avoided direct medical expenditures. These estimates do not capture downstream societal burdens including functional decline, caregiver needs, or long-term and reduced independence, which further magnify the overall impact. As such, preferential use of apixaban may offer both clinical and economic advantages in this high-risk population.
Strengths of this analysis include its large, heterogeneous real-world sample and extensive propensity adjustment, allowing for a focused comparison of DOAC safety in a population underrepresented in randomized trials However, the observational design using a web-based electronic health record platform is inherently subject to selection bias and residual confounding, including the possibility of unmeasured variables despite robust matching. Outcome ascertainment relied on diagnostic and procedural coding, and bleeding events could not be independently adjudicated, introducing potential misclassification. In addition, detailed information on valve type was not available, and data on concomitant antiplatelet therapy and imaging parameters were inconsistently captured. In the absence of randomized head-to-head trials of DOACs after TAVR, these results provide meaningful evidence to inform anticoagulant selection in patients with heightened bleeding vulnerability.
Among TAVR patients requiring long-term oral anticoagulation, apixaban was associated with a significantly lower risk of major bleeding compared with rivaroxaban, while thromboembolic and mortality outcomes remained comparable. While these results align with observations from the general atrial fibrillation population, they extend prior knowledge by specifically informing anticoagulant choice in the post-TAVR setting. These findings support preferential selection of apixaban and underscore the need for dedicated randomized trials to define optimal anticoagulation strategies after TAVR.
