Risk of Cancer Comparing Warfarin and Direct Oral Anticoagulants: Population-Based Cohort Studies in England and Hong Kong

Abstract

Previous evidence suggests a potential protective effect of warfarin against cancer, compared to non-users. However, it may be prone to immortal time bias and residual confounding. We aimed to examine the association between cancer and warfarin, compared with active comparator (direct oral anticoagulants). We conducted studies using population-based databases from England and Hong Kong to investigate the association between warfarin and hazard of cancer using a new-user active-comparator cohort design. People with atrial fibrillation aged ≥ 18 years who had first received anticoagulant treatment during 01/01/2011-31/12/2019 were involved. No evidence supported the association between warfarin and hazard of overall cancer, compared with direct oral anticoagulants in both settings (England: hazard ratio [HR] = 1.03, 95% confidence interval [CI] 0.94-1.13; Hong Kong: HR = 0.89, 95% CI 0.79-1.01). A lower hazard of female breast (HR = 0.49, 95% CI 0.30-0.79), ovarian (HR = 0.07, 95% CI 0.01-0.58), and pancreatic (HR = 0.46, 95% CI 0.22-0.96) cancers and a higher hazard of kidney cancer (HR = 3.57, 95% CI 1.64-7.76) were found in Hong Kong, comparing warfarin with direct oral anticoagulants, but these were not replicated in England. This study does not find a protective effect of warfarin against cancer versus direct oral anticoagulants. The risks of site-specific cancers including pancreatic, kidney, and sex-specific cancers between oral anticoagulants shown in the Hong Kong setting only may require further investigation in other independent datasets.