Oral anticoagulants and the risk of acute kidney injury: a retrospective real-world cohort study using overlap propensity score weighting

Background Anticoagulant-related nephropathy, initially described in association with warfarin, has increasingly been reported with direct oral anticoagulants (DOACs). However, the comparative risk of acute kidney injury (AKI) within 30 days of treatment initiation between warfarin and DOACs across a broad spectrum of clinical indications remains unclear. Methods This retrospective cohort study included patients newly initiated on DOACs or warfarin between January 2014 and May 2025 at the Third Xiangya Hospital of Central South University. Overlap propensity score weighting, with weights derived from a generalized boosted model incorporating 81 baseline covariates, balanced the treatment groups. Weighted logistic regression compared the risk of AKI within 30 days of OAC initiation, defined by KDIGO criteria. Sensitivity analyses included inverse probability of treatment weighting (IPTW) and logistic regression-derived overlap weighting, each with or without doubly robust covariate adjustment. Results Among 6 876 patients (5 048 DOAC users; 1 828 warfarin users), DOAC initiation was associated with a higher risk of AKI than warfarin after overlap weighting (OR: 1.346, 95% CI: 1.057–1.715; p = 0.016), and with severe AKI (stage 2–3; OR: 1.694, 95% CI: 1.013–2.832; p = 0.044), whereas stage 1 AKI did not differ significantly. The association persisted after additional covariate adjustment (OR: 1.363, 95% CI: 1.068–1.740; p = 0.013) and across alternative propensity score models, although the unadjusted IPTW model did not. By indication, the association was significant for stroke prevention in atrial fibrillation (AF) (OR: 1.566, 95% CI: 1.034–2.369; p = 0.034) but not for venous thromboembolism (VTE) treatment or prophylaxis. Among individual DOACs, rivaroxaban was associated with an elevated risk (OR: 1.392, 95% CI: 1.075–1.804; p = 0.012), whereas edoxaban, dabigatran, and apixaban were not. Lower-dose rivaroxaban initiation was associated with an increased risk (OR: 2.792, 95% CI: 1.524–5.114; p = 0.001), whereas higher-dose and off-label use were not. In prespecified subgroup analyses, ORs were above 1 in most strata, but most comparisons were not significant, and no significant interactions were detected. Conclusion These findings suggest that DOACs, particularly rivaroxaban, may be associated with a higher short-term risk of AKI than warfarin. Given the expanding use of DOACs, medication-specific renal effects warrant further investigation, and close monitoring of renal function may be advisable in high-risk patients after anticoagulation initiation.