Machine learning based analysis of the risks of combination aspirin and rivaroxabane antithrombotic therapy with nonsteroid anti-inflammatory drugs after arthroplasty

• Aspirin and rivaroxaban are largely interchangeable for thromboembolism prophylactics. • Concomitant use of these anticoagulants with ketorolac may increase bleeding risk. • Concomitant use with celecoxib was associated with a lower bleeding risk. • Celecoxib is the preferable NSAID for postoperative analgesia in patients on thromboprophylaxis. Thromboprophylaxis is essential following major joint surgery. While aspirin is widely discussed as a postoperative prophylactic alternative to heparin or rivaroxaban, it is often co-administered with NSAIDs for pain relief. The safety of such combinations remains unclear. This study uses multivariate machine learning (ML) analysis to compare bleeding risks between aspirin and rivaroxaban, and between their respective combinations with ketorolac or celecoxib. We conducted a retrospective study of 1164 patients who underwent primary total knee or joint arthroplasty. ML was applied to model the risk of bleeding based on the clinical factors. The analysis revealed a minimal difference in bleeding risk between aspirin (odds ratio [OR] = 1.042, 95 % confidence interval [CI] = 0.762–1.426, 1.000) and rivaroxaban (OR = 0.959, 95 % CI = 0.701–1.312, p = 1.000). Concomitant use of ketorolac with either anticoagulant non-significantly increased the risk of bleeding (OR = 1.077, 95 % CI = 0.787–1.473, p = 0.337), in contrast, the combination with celecoxib was associated with a reduced risk (OR = 0.751, 95 % CI = 0.549–1.028, p = 0.0008). Multivariable logistic regression demonstrated the highest predictive value (AUC = 0.68), along with the best F1 score and Matthews correlation coefficient (0.32 and 0.22, respectively). Aspirin and rivaroxaban appear to be interchangeable with respect to bleeding risk. While combining these anticoagulants with ketorolac did not significantly affect bleeding risk, the combination with celecoxib was associated with a lower risk. Therefore celecoxib may be preferable NSAID for analgesia in this context.