Apixaban showed a lower 2-year risk of complicated upper gastrointestinal bleeding (UGIB) vs edoxaban and rivaroxaban, but not dabigatran, in patients without prior anticoagulant exposure, according to an industry-independent, Asian population–based cohort study conducted by researchers from the Chinese University of Hong Kong (CUHK).
“Asian populations exhibit distinct risk profiles for UGIB due to a higher prevalence of Helicobacter pylori infection, different socioeconomic factors, and genetic variations influencing anticoagulant metabolism and bioavailability,” noted the researchers. “[However,] Asian data on complicated UGIB risks among direct oral anticoagulants [DOACs] are limited.” [Clin Gastroenterol Hepatol 2026;doi:10.1016/j.cgh.2026.04.016]
One of the largest Asian population–based cohorts
The current industry-independent, population-based, retrospective cohort study included 58,229 patients with new exposure to DOACs (apixaban, n=22,359; dabigatran, n=20,639; edoxaban, n=2,068; rivaroxaban, n=13,163) in Hong Kong during 2011–2020. Baseline parameters were balanced between DOACs by inverse probability of treatment weighting. The cohort’s mean age range was 71.6–76.5 years. The most common indications for DOAC prescription were atrial fibrillation, cardiac arrhythmia, and thromboembolism.
“To the best of our knowledge, this study is one of the largest Asian population–based cohorts that specifically evaluated the incidence and severity of UGIB in DOAC users,” highlighted the researchers.
The primary endpoint was complicated UGIB, defined as any UGIB-related death, haemoglobin drop of >2 g/dL, blood transfusions, emergency endoscopic procedures, radiological or surgical intervention, or 30-day rebleeding. After 2 years of observational follow-up, 271 cases of complicated UGIB occurred, with a median time-to-event of 4.7 months.
Compared with apixaban, both edoxaban and rivaroxaban were associated with significantly higher risks of complicated UGIB, with sub-distribution hazard ratio [S-HR] of 1.891 for edoxaban (95 percent confidence interval [CI], 1.091–3.275; p=0.025; false discovery rate [FDR]–adjusted p=0.039) and 1.441 for rivaroxaban (95 percent CI, 1.047–1.983; p=0.026; FDR-adjusted p=0.039). The number-needed-to-harm (NNT) was 244 and 427, respectively.
In contrast, dabigatran was not associated with a higher risk of complicated UGIB (S-HR, 1.05; 95 percent CI, 0.751–1.469) vs apixaban.
Elderly and renal impairment subgroups
In 21,449 patients aged ≥80 years, dabigatran (S-HR, 1.774; 95 percent CI, 1.197–2.630), edoxaban (S-HR, 2.326; 95 percent CI, 1.066–5.072) and rivaroxaban (S-HR, 1.899; 95 percent CI, 1.230–2.933) were all associated with higher risks of complicated UGIB than apixaban.
Similarly, in 15,611 patients with reduced estimated glomerular filtration rate (eGFR) of <60 mL/min/1.73 m2, both edoxaban (S-HR, 2.977; 95 percent CI,1.315–6.741) and rivaroxaban (S-HR,1.688; 95 percent CI, 1.002–2.841) were associated with higher risks of complicated UGIB than apixaban.
Among 32,113 patients receiving DOACs at lower doses, dabigatran (S-HR, 1.533; 95 percent CI, 1.019–2.305), edoxaban (S-HR, 2.504; 95 percent CI, 1.169–5.362) and rivaroxaban (S-HR, 1.678; 95 percent CI, 1.042–2.700) were all associated with higher risks of complicated UGIB vs apixaban.
Apixaban: Relatively better safety profile as 1st DOAC
“Our findings demonstrated that apixaban may have a relatively better safety profile when considering DOAC initiation in our Chinese population, especially in vulnerable patients requiring dose reduction due to advanced age or renal impairment,” conclude the researchers.