The American College of Chest Physician score to assess the risk of bleeding during anticoagulation in patients with venous thromboembolism: reply

Abstract

Essentials The risk of bleeding influences the duration of anticoagulation (AC) after venous thromboembolism. We assessed the ACCP bleeding risk score in an inception-cohort of patients receiving AC. 53% were categorized at high-risk, but their bleeding rate was low during long-term AC. ACCP score had low predictive value for bleeding. Summary Background The American College of Chest Physicians (ACCP) guideline proposes a score to decide on extended anticoagulation after an unprovoked venous thromboembolism (VTE). Methods We investigated the ACCP score to predict bleeding risk in an inception cohort of 2263 patients on long-term anticoagulation (1522 treated with vitamin K antagonists [VKAs] and the remaining with direct oral anticoagulants [DOACs]) belonging to the Italian START2 Register. Results More than half the patients were categorized as high risk; nevertheless, a higher proportion received anticoagulation for > 1 year compared with those in the low-risk category. For 3130 years (median 12 [interquartile range 6, 24] months), 48 bleeding outcomes occurred (1.53%/year) in the cohort (1.7%/year and 0.95%/year in high- and low-risk categories, respectively). The c-statistic of the ACCP score was 0.55 (0.48-0.63), 0.50 (0.42-0.58) and 0.56 (0.48-0.64) in low-, moderate- and high-risk categories, respectively. The bleeding incidence was higher during the first 90 days of treatment (3.0%/year) than afterwards (1.2%/year; relative risk (RR), 2.5 [1.3-4.7]), and similar among the three categories. The bleeding rate was not different during the initial 3 months of treatment in patients receiving VKAs or DOACs; it was, however, lower in the latter patients in the subsequent period (0.5%/year vs. 1.4%/year, respectively). Conclusion The bleeding rate during extended treatment was rather low in our patients. ACCP score had insufficiently predictive value for bleeding and cannot be used to guide decisions on extended treatment. New prediction tools for bleeding risk during anticoagulant treatments (including DOACs) are required.