Oral anticoagulant safety in family practice: prognostic accuracy of Bleeding Risk Scores (from the CACAO study)

To assess bleeding risk of patients treated by oral anticoagulants, several scores have been constructed to assist physicians in the evaluation of the benefit risk. Most of these scores lack a strong enough level of evidence for use in family practice. To assess the predictive prognostic accuracy of 13 scores designed to assess the risk of major or clinically relevant non-major (CRNM) bleeding events in a French ambulatory cohort receiving Vitamin-K antagonists (VKA) or direct oral anticoagulants (DOACs) in a family practice setting. CACAO (Comparison of Accidents and their Circumstances with Oral Anticoagulants) was a multicentre prospective cohort of ambulatory patients prescribed oral anticoagulants. We selected patients from the cohort who had received an oral anticoagulant because of non-valvular atrial fibrillation (NVAF) and/or venous thromboembolism (VTE) to be followed during one year by their GP. The following scores were calculated: mOBRI, Shireman, Kuijer, HEMORR2HAGES, ATRIA, HAS-BLED, RIETE, VTE-BLEED, ACCP score, Rutherford, ABH-Score, GARFIEL-AF, and Outcomes Registry for Better InformedTreatment of Atrial Fibrillation (ORBIT). Prognostic accuracy was assessed by using receiver operating characteristic curves and c-statistics. During 1 year, 3,082 patients were followed. All of the scores demonstrated only poor to moderate ability to predict major bleeding or CRNM in NVAF patients on DOACs (c-statistic: 0.41-0.66 and 0.45-0.58), respectively. The results were only slightly better for patients prescribed VKA (0.47-0.66 and 0.5-0.55, respectively) in this indication. The results were also unsatisfactory in patients treated for VTE. None of the scores demonstrated satisfactory discriminatory ability when used in family practice. ClinicalTrials.gov NCT02376777.