Outcomes of Direct Oral Anticoagulant- and Warfarin-Associated Hemorrhage: A Single Center Retrospective Cohort Study

Background:Direct oral anticoagulants (DOAC) are now a mainstay of anticoagulant therapy. However, outside of clinical trials, reports on treatment of DOAC-related bleeding are limited. We report a retrospective cohort study of the management and outcomes of oral anticoagulant-associated bleeding in a single medical center in the United States.

Methods: Patients ≥ 18 years admitted to one of four Emory Hospitals for bleeding while taking warfarin or DOAC (rivaroxaban, apixaban, edoxaban, or dabigatran) for atrial fibrillation, atrial flutter, or venous thromboembolism between October 2010 and June 2017 were identified using ICD9-10 codes from the clinical data warehouse (CDW). Patients were excluded if they had additional cardiac indications for warfarin, in-hospital or peri-procedural bleeding, admitted with ICD codes for abnormal coagulation without bleeding, or were transferred from another hospital. For patients with multiple admissions, only the initial bleeding episode was included. Patient characteristics, bleeding site, treatment, and outcome (length of stay [LOS] or death) were obtained from the CDW. Data from a random sample (20% of patients) was verified through chart review demonstrating that the ICD codes for bleeding have a positive predictive value of 94.7%. Death events were confirmed using the National Death Index. Data was summarized and compared between groups (DOAC vs warfarin) using standardized differences and Fisher exact test. Poisson regression was used to calculate risk ratios in the use of blood products and reversal agents, and in-hospital and 30-day mortality.

Results:Of the 277 patients included in the data abstraction and analysis, 89 (32.1%) received a DOAC and 188 (67.8%) received warfarin. The mean age was similar between groups (DOAC 76.4 years vs warfarin 73.6 years). Patients in the DOAC group compared to the warfarin group were more likely to be of white (70.8% vs 50.5%) vs African American race (24.7% vs 44.1%). The ISTH criterion for major bleed was met by 47 (52.8%) and 108 (57.4%) in the DOAC and warfarin groups, respectively. Within the DOAC group, most were on rivaroxaban (60.7%). Patients taking warfarin compared to DOAC were more likely to have a Charlson Comorbidity Index (CCI) of 2+ (63.9% vs 49.4%). There was no significant difference in bleeding sites between groups. Gastrointestinal was the most common site of bleeding in both groups (80.9% DOAC, 71.3% warfarin). Intracranial hemorrhage was reported in 12.4% of DOAC and 8.5% of warfarin group.

Patients in the warfarin group were more likely to receive a blood product during hospitalization for anticoagulation-related bleeding event compared to those receiving DOACs (66.5% vs 46.1%, adjusted RR 0.68 [95% CI 0.47, 0.97], Table 1). The difference is mainly driven by the higher rate of FFP administration in the warfarin group. As expected, vitamin K use was also higher in the warfarin group compared to the DOAC group. Utilization of non-activated PCC was low in both DOAC (6.7%) and warfarin (3.7%) groups without statistically significant difference. Of 17 patients on dabigatran, 4 (23.5%) were admitted after FDA approval of idarucizumab; 2/4 had a major bleed but idarucizumab was not used.

In-hospital death (Table 2) occurred in 5.6% of patients in the DOAC group compared with 2.1% in the warfarin group. The Kaplan-Meier curve (Figure 1) showed a trend for higher survival probability in warfarin-related bleeding event compared to DOAC (P=0.052). However, in-hospital and 30-day mortality were not statistically significantly different between groups after adjustment for age, race, and CCI (Table 2). There was no significant difference in median hospital LOS (3.19 days DOAC vs 3.97 days warfarin).

Conclusions:We report experience of anticoagulant-related hemorrhage in a large metropolitan healthcare system. Despite > 50% with major bleeding and lack of specific antidote for DOAC, we found an overall low in-hospital mortality and short LOS which was not significantly different between DOAC or warfarin users. Total blood product administration was higher in warfarin compared to DOAC-related bleeding. Hemostatic or reversal agents were rarely, if ever used, for management of DOAC-related bleeding. This study highlights that as the complexity of anticoagulants and their reversal increases, the need to optimize hospital systems for anticoagulation reversal strategies also increases.

View largeDownload slide

View largeDownload slide

Close modal