Anticoagulant Therapy and Risk of Bleeding in Patients with Atrial Fibrillation of Four Italian Regions: The TYRION Study

Background/Introduction: The new oral anticoagulants (NOA), dabigratran, rivaroxaban, apixaban and edoxaban, have been associated with a lower risk of bleeding in patients with atrial fibrillation (AF) as compared with traditional oral anticoagulants (TOA). In particular, NOA have been shown to reduce the mortality rate caused by haemorrhagic events in the prevention of arterial thromboembolism in nonvalvular AF. However, current literature evidence is conflicting, and some studies highlight a higher frequency of gastrointestinal bleeding with NOA than TOA [1–3]. Aim: Risk evaluation of haemorrhagic events (overall, intracranial and gastrointestinal bleeding) in patients treated with NOA for atrial fibrillation, as compared with TOA. Methods: The present nested case–control analysis was conducted by linking the regional databases of drug reimbursement claims to the regional records of hospital discharge of four Italian Regions (Tuscany, Marche, Lazio and Campania) through an anonymous unique patient identification number. The historical cohort included patients with at least one NOA or TOA prescription from July 1st, 2012 to December 31st, 2017 (intention to treat). The cohort entry was defined by the date of the first NOA or TOA prescription after July 1st, 2013 among patients who were free from any anticoagulant prescription in the previous year. The index date was that of the first bleeding. Cases were patients with at least one haemorrhagic event. In the primary analysis, each case was matched with at least two controls by year of birth, sex, year of cohort entry, and follow-up duration. In the secondary analysis, cases were further stratified into intracranial and gastrointestinal bleeding, and they were matched with at least two controls as well. TOA users were taken as reference group for both analyses. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated by conditional logistic regression for the association of NOA use with haemorrhagic events, adjusted for region. Results: The cohort included 96,247 patients. At the cohort entry, 33,425 and 62,822 patients had received TOA and NOA, respectively. During the follow-up period, 17,576 patients were switched from TOA to NOA, while 1930 from NOA to TOA. A total of 5158 cases of overall bleeding, including 927 events of intracranial bleeding and 1465 events of gastrointestinal bleeding, were recorded. The adjusted ORs for NOA were: 0.537 (95% CI: 0.502–0.574) for overall bleeding, 0.472 (95% CI: 0.401–0.554) for intracranial bleeding, and 0.676 (95% CI: 0.594–0.769) for gastrointestinal bleeding. Conclusions: Adjusted ORs showed a significant protective effect of NOA, as compared with TOA, for overall, intracranial and gastrointestinal bleeding. These results support the evidence of a reduced risk of bleeding events with the use of NOA, as compared with TOA