Anticoagulation Stewardship in the Era of Practice-Changing Evidence: A Narrative Review of Recent Trials and Their Bedside Translation.

Background/Objectives: The evidence base for anticoagulation moved substantially between 2023 and 2026: a reframed atrial fibrillation (AF) guideline, new randomized data on extended treatment of cancer-associated venous thromboembolism (VTE), the first adequately powered head-to-head comparison of two direct oral anticoagulants (DOACs), the first randomized evidence for andexanet alfa, and mixed results for the factor XI inhibitor class. Real-world uptake remains incomplete. To review the 2023-2026 evidence changing anticoagulation practice, identify where bedside practice lags behind it, and outline the institutional adjustments needed to translate it into routine care. Methods: Narrative review. PubMed, EMBASE, and the Cochrane Library were searched from 2000 to January 2026, with hand-searching of society guidelines and regulatory documents; eligible sources were randomized trials, pre-specified subgroup analyses, prospective registries, systematic reviews, and society-endorsed guidelines. A targeted update in March 2026 captured practice-changing trials published after the primary search date, identifying COBRRA (published 11 March 2026). This is a narrative, non-systematic review reported in accordance with SANRA: no protocol was registered, topic selection is interpretive, and no formal risk-of-bias assessment or quantitative synthesis was undertaken. Recommendations are labeled by evidence basis. Results: Seven developments warrant attention: (1) the 2023 AF guideline's ≥2%/year anticoagulation threshold, with sex-specific application and a 1-2%/year shared decision zone; (2) API-CAT-reduced-dose apixaban non-inferior to full dose for extended cancer VTE prevention (2.1% vs. 2.8%; adjusted subhazard ratio 0.76, 95% CI 0.41-1.41; pre-specified non-inferiority margin 2.00), with less bleeding (12.1% vs. 15.6%); (3) COBRRA-apixaban substantially safer than rivaroxaban in acute non-cancer VTE (clinically relevant bleeding 3.3% vs. 7.1%; RR 0.46, 95% CI 0.33-0.65; major bleeding 0.4% vs. 2.4%; RR 0.16, 95% CI 0.06-0.40), without efficacy trade-off; (4) ANNEXA-I-a modest hemostatic advantage for andexanet alfa over usual care (67.0% vs. 53.1%) at the cost of increased thrombosis (10.3% vs. 5.6%); (5) an asymmetric factor XI landscape in which class-level conclusions remain premature; (6) new special population data (ELDERCARE-AF, AZALEA sub-analyses by kidney function); and (7) a maturing but uneven stewardship evidence base. Conclusions: Three years of trials have produced concrete implications for how anticoagulation should be practiced now. The interval between publication and routine implementation remains the largest practical problem in the field, and institutional stewardship structures are a major mechanism for closing it.