Vitamin-K Antagonists vs. Direct Oral Anticoagulants on Severity of Upper Gastrointestinal Bleeding: A Retrospective Analysis of Italian and UK Data

Background: Gastrointestinal bleeding (GIB) is one of most frequent and significant challenges for emergency physicians and gastroenterologists. Mortality for upper (U) GIB is high, especially in the elderly and comorbid patients. However, there is scant evidence in the literature concerning an assessment of warfarin (VKA) and direct oral anticoagulants (DOACs) in terms of upper gastrointestinal bleeding (UGIB) severity. Aims: Using data from two different settings (Italy and the UK), we aimed to compare the impact of VKA and DOACs on the severity of UGIB. Methods: Retrospective bicentric study on adult patients under VKA or DOACs admitted either to the emergency department at the Gemelli Hospital in Rome, Italy or University College Hospital in London, UK, with suspected UGIB from 01/01/2017 to 31/12/2018. Univariate analysis with Fisher's exact test, and analysis of variance (ANOVA) were used. Results: 106 patients (62 M/44 F; mean age 71.2 +/- 16.9 yrs) were enrolled and divided into the VKA group (N = 57; M: 56%, mean age: 64.9 +/- 21.3 yrs) and the DOAC group (N = 49; M: 61%; mean age: 77.6 +/- 12.5 yrs). At univariate analysis, the VKA group presented two endoscopic diagnoses more frequently than the DOAC group (26% vs. 8%, p < 0.05), were more frequently endoscopically treated (44% vs. 22%, p < 0.05), rescoped (12% vs. 2%, p = 0.048) and hospitalized (79% vs. 53%, p = 0.01) with a longer length of stay, LOS (VKA: 58% > 5 days vs. DOAC: 68% < 5 days, p = 0.01). There was no difference in terms of hemoglobin level on admission, however the requirement of blood transfusions was higher in the VKA group (60% vs. 41%, p = 0.041). One third of the VKA group showed a lower platelet count than the DOAC group (33% vs. 8%, p = 0.01). No statistically significant differences for in-hospital mortality were observed. For the ANOVA, the type of anticoagulant used was the only significant predictor of need to rescope (p = 0.041) and a significant co-predictor for a LOS > 5 days (p = 0.009; as well as cirrhosis, p = 0.013 and age, p = 0.005). Conclusions: Our outcomes revealed a more severe UGIB in patients on VKA, but the impact of comorbidities (i.e., more cirrhotic patients in the VKA group) cannot be disregarded. DOAC subgroup descriptive analysis, even though on a little cohort, showed higher bleeding severity for rivaroxaban.