3A-02 - Concomitant use of direct oral anticoagulants and COX-2 selective non-steroidal anti-inflammatory drugs and the risk of gastrointestinal bleeding in patients with non-valvular atrial fibrillation: multinational cohort study

Background: Non-steroidal anti-inflammatory drugs (NSAIDs) selectively inhibiting cyclooxygenase-2 (COX-2) reduce the risk of gastrointestinal bleeding (GIB) compared to non-selective NSAIDs. However, their benefits among patients with non-valvular atrial fibrillation (NVAF) who are concomitantly using direct oral anticoagulants (DOACs) have not been assessed.

Objectives: To assess whether concomitant use of DOACs and COX-2 selective NSAIDs (celecoxib, diclofenac, etodolac, mefenamic acid, meloxicam, rofecoxib) is associated with a decreased risk of GIB versus concomitant use of DOACs and non-selective NSAIDs (fenoprofen, flurbiprofen, ibuprofen, indomethacin, ketoprofen, ketorolac, nabumetone, naproxen, oxaprozin, oxyphen-butazone, phenylbutazone, piroxicam, sulindac, tenoxicam tolmetin, tiaprofenic acid) in patients with NVAF.

Methods: We used the UK´s Clinical Practice Research Datalink and Québec claims data to assemble a two cohorts of patients with NVAF who initiated concomitant use of DOACs and NSAIDs (2011-2020). Cohort entry date was the first day of concomitant use. Using an as-treated exposure definition, we assessed the risk of GIB associated with concomitant use of DOACs and COX-2 selective NSAIDs versus concomitant use of DOACs and non-selective NSAIDs. Propensity score (PS) based inverse probability of treatment weighting (IPTW) was used for confounding control. PS were calculated separately for different strata based on the order of initiation of DOACs and NSAIDs and then pooled together. Patients were allowed to contribute multiple episodes of concomitant use. We estimated site-specific hazard ratios (HRs) and 95% confidence intervals (CIs) via Cox models and pooled them together using random-effect models. We further stratified by specific DOAC (apixaban, rivaroxaban) repeating IPTW for each stratum. In sensitivity analyses, we assessed the impact of exposure and outcome misclassification, selection bias, and residual confounding.

Results: The two study cohorts included overall 37,833 patients treated with DOACs and COX-2 selective NSAIDs (n=17,105) or non-selective NSAIDs (n=20,728). Most patients (>80%) contributed one episode of concomitant use. Compared to concomitant use of DOACs and non-selective NSAIDs, concomitant use of DOACs and COX-2 selective NSAIDs was associated with a decreased risk of GIB (pooled HR 0.63; 95 % CI, 0.46-0.87; I2=56%). The effect was more pronounced with apixaban (pooled HR, 0.38; 95% CI, 0.21-0.70; I2=64%) than rivaroxaban (pooled HR, 0.62; 95% CI, 0.38-0.99; I2=47%). Sensitivity analyses yielded consistent estimates (HRs from 0.61 to 0.83).

Conclusions: COX-2 selective NSAIDs seem to retain their beneficial effects with respect to GIB during concomitant use with DOACs.