Bleeding Risk in Patients 80 Years and Older Taking a Direct Oral Anticoagulant

In people 80 years and older taking a direct oral anticoagulant (DOAC), who is at increased risk for major hemorrhage?

DOACs have a somewhat lower bleeding risk than older vitamin K antagonists such as warfarin. DOACs are also more convenient to take, which make them well suited for older patients. Current guidelines recommend DOACs over warfarin for patients who require anticoagulation for atrial fibrillation.^1,2 However, the risk of major hemorrhage increases with age, and older patients may consider alternative therapies, such as left atrial appendage closure, if they have a significantly higher risk of bleeding.³ Therefore, a score to predict bleeding risk in patients 80 years and older is needed to guide shared decision-making.

Several risk scores have been developed and prospectively validated to predict the likelihood of major bleeding in patients taking anticoagulation, including the HAS-BLED, HEMORR₂HAGES, ATRIA, and ORBIT risk scores.^4–7 However, these risk scores were developed in younger populations with mean ages of 71 to 76 years; in some cases, they have a large number of predictors that increases the burden for clinicians to apply them.⁵

These scores have failed to validate well. For example, a systematic review that compared the HAS-BLED, ATRIA, and HEMORR₂HAGES risk scores found sensitivities of 41% to 53% for HAS-BLED, 23% for ATRIA, and 27% for HEMORR₂HAGES in the detection of major bleeding.⁸ A subsequent network meta-analysis identified 18 studies that evaluated the accuracy of nine risk scores in 321,888 patients, including the ATRIA, ORBIT, HAS-BLED, and HEMORR₂HAGES.⁹ These risk scores had a sensitivity of less than 50% for detecting hemorrhage.

Another limitation of these older risk scores is that they were largely derived in patients taking warfarin or another vitamin K antagonist. Because care has shifted to DOACs, it is important that risk scores are developed and evaluated in patients who take a DOAC. The 10-item DOAC score was derived in 5,684 patients with a mean age of 71 years who were taking the DOAC dabigatran (Pradaxa).¹⁰ In prospective validation using a new group of patients, the DOAC score demonstrated modest overall accuracy for bleeding (C statistic = 0.67) and slightly outperformed the HAS-BLED (C statistic = 0.63). In a second validation study, the HAS-BLED (C statistic = 0.65) and DOAC (C statistic = 0.62) scores also had modest overall accuracy.¹¹ The DOAC score is discussed in more detail in a previous Point-of-Care Guide.¹²

Most recently, French investigators developed and internally validated a risk score in a population of people 80 years and older with a heavy burden of comorbidities who were taking the DOAC rivaroxaban (Xarelto) for atrial fibrillation.¹³ The A4C risk score uses five factors to predict bleeding risk: age, anemia, low albumin, amiodarone use, and low creatinine clearance. It was developed in 408 participants and validated in 439. The C statistic for the new score was 0.73 in the derivation population and 0.66 in the validation population. The researchers also evaluated the accuracy of the ATRIA, HEMORR₂HAGES, HAS-BLED, and RE-LY risk scores in the same population and found that they have worse overall accuracy, with C statistics ranging from 0.50 to 0.61 in the validation population.