Outcomes of Upper Gastrointestinal Bleeding Are Similar Between Direct Oral Anticoagulants and Vitamin K Antagonists

Claire Gouriou; Guillaume Bouguen; Pierre Lahmek; Agnes Pelaquier; Ramuntxo Arotcarena; Armand Garioud; Stephanie De Montigny-Lenhardt; Arnaud Pauwels; David Zanditenas; Claire Charpignon; Remi Combes; Stephane Nahon; Vincent Quentin

Disclosures

Aliment Pharmacol Ther. 2021;53(6):688-695. 

In This Article

Results

Patient Characteristics

Among the 2498 patients with a complete file included in the main cohort, 475 (19%) were treated with an oral anticoagulant: 267 with VKA (56.2%)—Fluindione 200 (75%) and Warfarin 67 (25%)—and 208 had DOACs (43.8%)—rivaroxaban 114 (55%), apixaban 73 (35%) and dabigatran 21 (10%). The flowchart is presented in Figure 1 and baseline characteristics of the study population are presented in Table 1. Our cohort was mostly composed of men (65%) with a mean age of 78.7 [±8.2] years and 66.5% were older than 75 years. The mean Charlson, Rockall and Blatchford scores were 3.2, 5 and 11.7 respectively. Baseline characteristics were broadly similar between VKA and DOACs except for the association of kidney failure and cirrhosis, which were more prevalent in the VKA group. Likewise, the Rockall score was slightly superior in the VKA group (P = 0.03). The main reason for anticoagulation was prevention of stroke or systemic embolism in atrial fibrillation in both groups (VKA 52.6% and DOACs 66.8%) followed by VTE (14.7% and 12%). Heart valve represented 13.5% of indications in the VKA group. At least 136 patients (28.6%) had a comedication with an antiplatelet agent. Aspirin intake was described for 100 patients (21.1%) and 55 (11.6%) underwent treatment with another antiplatelet agent, 19 patients had both (4%). The main indication in both groups for Aspirin was secondary prevention of coronary disease (69.6%). Other antiplatelet agents were mainly indicated for a past history of coronary stent or bypass. At admission, 81% patients had melaena and a shock was present in 47 patients (9.9%). UGIB occurred during hospitalisation for 19.8% of the patients without difference between groups (P = 0.79). In the VKA group, mean INR was 2.6 [±1.8] and 23% (n = 62) patients had and INR > 3. Coagulopathy reversion was performed in this group for half of the patients (n = 135, 50.6%): by vitamin K (n = 85, 63%), prothrombin complex (n = 41, 30.3%) or fresh frozen plasma (n = 9, 6.6%). Only 36 (17.3%) patients treated with DOACs had a coagulopathy reversion: with prothrombin complex (n = 19, 52.7%), fresh frozen plasma (n = 9, 25%) or specific antidotes (n = 4, 11%).

Figure 1.

Flow chart. Abbreviations: DOAC, direct oral anticoagulants; UGIB, upper gastrointestinal bleeding, VKA vitamin K antagonist

Endoscopy

Endoscopy was performed for 470 (98.9%) patients. Missing endoscopies were due to contraindication (n = 2), death (n = 1) or emergent transfer of the patient (n = 1) and not documented (n = 1). Gastroscopy was described as normal in 73 patients (15.3%) and presented active bleeding in 117 (24.6%) patients. The endoscopist identified peptic lesions for 286 patients (60.2%) vascular lesions for 44 (9.3%), lesions related to portal hypertension for 43 (9%) and tumour for 27 (5.7%) patients. Bleeding was finally imputed to a peptic origin in 233 cases (49%), vascular origin in 37 cases (7.8%), portal hypertension and tumour in 27 cases (5.7% each). Endoscopic treatment was performed for 128 (26.9%) patients: clip (n = 67), adrenalin injection (n = 65), Hemospray (n = 13), argon (n = 12), ligation (n = 12) and prosthesis (n = 1); permitting bleeding resolution in 95 patients (74.2%). Ninety-eight patients (20.6%) underwent a second look endoscopy mostly indicated for a systematic control (56%), 20% because of an unsatisfying first endoscopy and 20% for rebleeding. All endoscopic features of the bleeding were similar according to the group of anticoagulants.

Mortality and Associated Predictive Factors

Mortality rate at 6 weeks was 12.4% (59 patients): 43 in the VKA group (16.1%) and 16 in the DOACs group (7.8%), P = 0.0059. Death occurred for 67.8% patients (40/59) during hospitalisation with a mean delay for in-hospital mortality of 11.3 [±2.4] days after endoscopy. The causes of death according to the group of oral anticoagulant are presented in Table 2.

Results of univariate and multivariate analysis are presented in Table 3. The Charlson index (OR = 4.02, 95%CI [2.16–7.44], P < 0.0001) and being an in-patient (OR = 2.96 [1.5–5.7], P = 0.0013) remained independently associated with mortality. The type of oral anticoagulation (DOACs vs VKA) had no influence on mortality (OR = 0.53 [0.27–1.04], P = 0.068).

Rebleeding and Associated Predictive Factors

At 6 weeks, a total of 56 patients (11.8%) presented a rebleeding in a mean delay of 14.4 [±7.7] days after the initial endoscopy. Rebleeding rate was similar between groups: VKA 11.2% vs DOACs 12.5%, P = 0.71. For 36 patients (64.3%), this recurrence of bleeding happened within the same hospitalisation, after a mean time of 5.4 [±1.4] days and only half of them (20/36) had a second endoscopy. After discharge, 28 patients presented a further bleeding (VKA n = 15 [5.6%] vs DOACs n = 13 [6.3%], P = 0.77), related to the initial lesion in 64% (n = 18) cases and to a new lesion for 21.4% (n = 6) cases.

By multivariate analysis, concomitant antiplatelet agents intake was associated with a higher rate of rebleeding (OR = 2.55 [1.22–5.35], P = 0.012) whereas this risk was lower among patients treated with beta-blockers (OR = 0.39 [0.21–0.76], P = 0.006).

Non-endoscopic Treatment and Predictive Factors

Non-endoscopic treatment was performed in 18 (3.8%) patients: 10 (3.7%) in the VKA group and eight (3.8%) in the DOACs group, P = 0.95. Fourteen patients underwent surgery, three had radiologic embolisation and one TIPSS.

Only the tumour origin of the bleeding was associated with the need for a non-endoscopic treatment (OR = 6.7 [1.7–26.5], P = 0.007).

Follow-up

The mean duration of post-endoscopy hospitalisation was 10.6 [±1.4] days (VKA 11.1 days vs DOACs 10 days, P = 0.21). Prolongation of hospitalisation was secondary to UGIB for 48 (10%) patients (24 patients in each group). Other main reasons were comorbidities for 121 patients (25.5%): VKA n = 68 (25.5%) vs DOACs n = 53 (25.5%), P = 0.99; and awaiting transfer for 27 patients. After UGIB, oral anticoagulants were reintroduced for 170 patients in the VKA group (64%) and 131 in the DOACs group (63%), P = 0.93. Resumption was more prescribed within 5 days in the VKA group (n = 113, 67%) than in the DAOCs group (n = 72, 55%), P = 0.004. For 78/136 patients (57%) taking at least one antiplatelet agent, treatment was resumed after bleeding, without difference between groups (VKA n = 37/72 vs DOACs n = 41/64; P = 0.14). Antiplatelet agents were restarted within 5 days for 32 patients in the VKA group (87%) and 31 patients in the DOACs group (76%), without difference (P = 0.22). Neither the resumption of antiplatelet agents or anticoagulants nor the time of resumption was associated with the risk of rebleeding.

processing....