Beyond Anticoagulant Exposure: Determinants of Mortality After Gastrointestinal Bleeding Across DOAC, Warfarin and Non-Anticoagulated Patients.

BACKGROUND AND AIMS: Reported outcomes after gastrointestinal bleeding (GIB) in anticoagulated patients are inconsistent, and comparisons with non-anticoagulated patients are limited. We assessed whether anticoagulation status was independently associated with mortality after GIB and identified predictors of death.

METHODS

We analysed 11,387 consecutive adults with acute GIB over 25 years (n = 500 direct oral anticoagulant [DOAC], n = 353 warfarin, n = 10,534 non-anticoagulated). All-cause mortality at 30 days and 1 year was modelled using overlap-weighted logistic regression, with complete mortality follow-up through the Israeli National Population Registry.

RESULTS

Anticoagulated patients were older (median age 81.7 [DOAC] and 75.1 [warfarin] vs. 58.4 years), had greater comorbidity and higher crude mortality than non-anticoagulated patients. After overlap weighting, neither DOAC nor warfarin was significantly associated with mortality (30-day OR 1.61, 95% CI 0.81-3.21 for DOAC and 1.30, 95% CI 0.56-3.02 for warfarin; 1-year OR 0.90, 95% CI 0.62-1.31 and 1.08, 95% CI 0.69-1.69). At 30 days, mortality was associated with age, haemodynamic shock (OR 4.25) and Internal Medicine ward admission (OR 4.55); at 1 year, with age, renal failure, Internal Medicine ward admission and inversely with male sex. Anticoagulation was resumed at discharge in 37.8% of survivors.

CONCLUSIONS

Neither DOAC nor warfarin was significantly associated with 30-day or 1-year mortality after GIB, including when bleeding severity and admitting ward were excluded from the outcome model. Thirty-day mortality was associated with severity at presentation, whereas 1-year mortality was associated with underlying patient characteristics. Resumption was not significantly associated with mortality or readmission for recurrent bleeding.