IMPORTANCE: It is unclear whether the beneficial effects of nonsteroidal anti-inflammatory drugs (NSAIDs) that selectively inhibit cyclooxygenase 2 (COX-2) in decreasing the risk of gastrointestinal (GI) bleeding are retained in patients with nonvalvular atrial fibrillation (NVAF) concomitantly treated with direct oral anticoagulants (DOACs). OBJECTIVE: To assess whether concomitant use of DOACs and COX-2-selective NSAIDs is associated with a decreased risk of GI bleeding vs concomitant use of DOACs and nonselective NSAIDs in patients with NVAF. DESIGN, SETTING, AND PARTICIPANTS: This multinational cohort study used electronic medical records from the UK and claims data from Quebec between January 1, 2011, and December 12, 2020. Adult patients (aged ≥18 years) with NVAF who initiated concomitant use of DOACs and NSAIDs during the study period were included. Cohort entry was the first day of concomitant use. Patient data were followed until a study outcome, death, or treatment discontinuation or switch. All analyses were conducted between November 19, 2024, and July 25, 2025. EXPOSURES: Concomitant use of DOACs and COX-2-selective NSAIDs or concomitant use of DOACs and nonselective NSAIDs. MAIN OUTCOMES AND MEASURES: The primary outcome was GI bleeding, and the secondary outcome was non-GI bleeding. Inverse probability of treatment weighting was used to control for confounding. Cox models were used to estimate site-specific hazard ratios (HRs) and 95% CIs, which were pooled using random-effects models. Further stratification was performed for potential effect modifiers. RESULTS: The study cohort included 30 240 patients with NVAF who initiated concomitant use of DOACs and NSAIDs (mean [SD] age, 72.1 [9.2] years, 17 146 male [56.7%]). They contributed 37 833 episodes of concomitant use of DOACs and NSAIDs (45.2% COX-2-selective NSAIDs, 54.8% nonselective NSAIDs). Concomitant use of DOACs and COX-2-selective NSAIDs was associated with a decreased risk of GI bleeding (pooled weighted HR, 0.63 [95% CI, 0.46-0.87]; I2 = 56%) and non-GI bleeding (pooled weighted HR, 0.54 [95% CI, 0.40-0.74]; I2 = 0%) vs concomitant use of DOACs and nonselective NSAIDs. Female patients had a greater decreased risk of GI bleeding associated with concomitant use of DOACs and COX-2-selective NSAIDs (pooled weighted HR, 0.50 [95% CI, 0.31-0.80]; I2 = 0%) than male patients (HR, 0.85 [95% CI, 0.55-1.32]; I2 = 78%). There was no major effect measure modification by age, order of drug initiation in concomitant use, baseline bleeding risk, or individual DOACs. CONCLUSIONS AND RELEVANCE: These findings suggest that COX-2-selective NSAIDs may retain their beneficial effects regarding GI bleeding during concomitant use with DOACs. Future studies should examine the association of COX-2 selectivity with stroke risk among patients treated with DOACs.