Comparison of clinical performance of four gastrointestinal bleeding risk scores in Chinese patients with atrial fibrillation receiving oral anticoagulants.

Gastrointestinal bleeding is the most common bleeding complication during anticoagulant therapy. A reliable bleeding risk score can help the clinician assess risk of bleeding in individual patients and select the anticoagulant regimen. This study retrospectively analyzed the data of patients with atrial fibrillation who received anticoagulant therapy from July 2015 to December 2018 at two centers-the Fujian Medical University Union Hospital and Fuzhou Second Hospital Affiliated to Xiamen University. Demographic data, clinical findings, and laboratory results were collected from the hospital records. Patients were followed up for 6 months. The performance of four bleeding risk scores (New Score, RIETE Score, Cuschieri et al. Score, de Groot et al. Score) for prediction of gastrointestinal bleeding was assessed using the area under the curve. A total of 3462 patients (mean age, 66.3 ± 11.5 years; 59.6% males; 1055 direct oral anticoagulants users and 2407 warfarin users) were followed up for 6 months. While 99/3462 (2.9%) patients had gastrointestinal bleeding. The area under the curves for the New, RIETE, Cuschieri et al., de Groot et al. scores were 0.652 (95% CI 0.576-0.728), 0.862 (95% CI 0.809-0.914), 0.606 (95% CI 0.527-0.685), and 0.873 (95% CI 0.816-0.929), respectively. Among the four BRSs evaluated, the RIETE score and the de Groot et al. score appear to have the good predictive value, while the NEW score and the Cuschieri et al. score did not sufficiently predict gastrointestinal bleeding risk within the study Chinese population.