Continuation of direct oral anticoagulants versus warfarin during critical illness: bleeding and clinical outcomes in a multicenter ICU cohort.

A growing proportion of patients admitted to the intensive care unit (ICU) receive chronic oral anticoagulation with direct oral anticoagulants (DOACs) or warfarin. The comparative safety of continuing DOACs versus warfarin during critical illness remains uncertain. We evaluated bleeding and clinical outcomes among ICU patients who continued the same oral anticoagulant during the early ICU course. We conducted a retrospective cohort study of adult ICU admissions across Mayo Clinic sites from 2012 to 2023. Eligible patients were taking a DOAC or warfarin at ICU admission and continued the same anticoagulant during the first three ICU calendar days. The primary outcome was major bleeding during the index hospitalization. Secondary outcomes included ICU and hospital mortality, ICU- and hospital-free days, bleeding subtypes, and need for procedural interventions. Among 6,258 eligible ICU admissions, 2,410 patients continued the same oral anticoagulant during the first three ICU calendar days, including 1,074 continuing DOAC therapy and 1,336 continuing warfarin therapy. In the multivariable analysis, there was no statistically significant difference in major bleeding events, (aOR 1.34, 95% CI 0.89-2.02; p = 0.161) between the DOACs and warfarin groups, respectively. Gastrointestinal bleeding (GI) was more frequent among patients continuing warfarin and remained significant after adjustment (adjusted OR 3.90, 95% CI 1.91-7.94; p < 0.001). Hospital mortality was lower with warfarin use than DOACs (5.6% vs 11.6%; p < 0.001; aOR 0.47, 95% CI 0.33-0.68). Among ICU patients selected to continue their baseline oral anticoagulant, continued DOAC therapy was not associated with higher adjusted odds of major bleeding and was associated with lower GI bleeding events than warfarin, but this bleeding advantage did not translate into lower mortality. This discordance suggests that the mortality in this population is driven largely by illness severity, comorbidity, unmeasured confounding and other non-bleeding related pathways.