Incidence and Predictors of Bleeding in Patients With Cancer and Atrial Fibrillation

Despite patients with cancer having a higher incidence of atrial fibrillation (AF), little is known about the predictors of outcomes in this population. This study aimed to assess the incidence and predictors of bleeding in patients with AF and cancer. The study population comprised 16,056 patients from a Spanish health area diagnosed with AF between 2014 and 2018 (1,137 with cancer). Competing risk analysis were used to evaluate the association of cancer and bleeding. Discrimination and calibration of bleeding risk scores were assessed by the concordance statistic and the Brier score, respectively. During a median follow-up of 4.9 years, the incidence of bleeding in patients with cancer was 13.2 per 100 patients/year. After multivariate adjustment, a significant association between cancer and bleeding was detected (subdistribution hazard ratio [sHR] 1.18, 95% CI 1.07 to 1.30, p = 0.001), specifically in patients with active cancer or previous radiotherapy. Early age, male gender, diabetes, and anticoagulation were independent predictors of bleeding. However, only anticoagulation with vitamin K antagonist (sHR 1.36, 95% CI 1.03 to 1.78, p = 0.026), not with direct oral anticoagulants (sHR 1.25, 95% CI 0.84 to 1.85, p = 0.270), was associated with bleeding. Discrimination and calibration of Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile INR, Elderly, and Drugs/alcohol concomitantly (HAS-BLED), AnTicoagulation and Risk factors In Atrial fibrillation (ATRIA), and Hepatic or renal disease, Ethanol abuse, Malignancy, Older (age ≥75 years), Reduced platelet count or function, Rebleeding risk, Hypertension, Anaemia, Genetic factors, Excessive fall risk and Stroke (HEMORR₂HAGES) scores were poor in patients with cancer (concordance statistic <0.6 and Brier score >0.1). In summary, cancer was associated with an increased risk of bleeding in patients with AF. The predictive ability of bleeding risk scores was poor in this population. Anticoagulation with vitamin K antagonist but not with direct oral anticoagulants, was an independent predictor of bleeding in patients with cancer.