In clinical practice, direct oral anticoagulants (DOACs) are often started earlier (≤ 7 days) than in randomized clinical trials after stroke. We aimed to develop a nomogram model incorporating time of DOAC introduction ≤ 7 days of stroke onset in combination with different degrees of stroke radiological/neurological severity at the time of treatment to predict the probability of unfavorable outcome. We conducted a multicenter prospective study including 344 patients who started DOAC 1-7 days after atrial fibrillation-related stroke onset. Computed tomography scan 24-36 h after stroke onset was performed in all patients before starting DOAC. Unfavorable outcome was defined as modified Rankin Scale (mRS) score > 2 at 3 months. Based on multivariate logistic model, the nomogram was generated. We assessed the discriminative performance by using the area under the receiver operating characteristic curve (AUC-ROC) and calibration of risk prediction model by using the Hosmer-Lemeshow test. Onset-to-treatment time for DOAC (OR: 1.21, p = 0.030), NIH Stroke Scale (NIHSS) score at the time of treatment (OR: 1.00 for NIHSS = 0-5; OR: 2.67, p = 0.016 for NIHSS = 6-9; OR: 26.70, p < 0.001 for NIHSS = 10-14; OR: 57.48, p < 0.001 for NIHSS ≥ 15), size infarct (OR: 1.00 for small infarct; OR: 2.26, p = 0.023 for medium infarct; OR: 3.40, p = 0.005 for large infarct), and age ≥ 80 years (OR: 1.96, p = 0.028) remained independent predictors of unfavorable outcome to compose the nomogram. The AUC-ROC of nomogram was 0.858. Calibration was good (p = 2.889 for the Hosmer-Lemeshow test). The combination of onset-to-treatment time of DOAC with stroke radiological/neurological severity at the time of treatment and old age may predict the probability of unfavorable outcome.
Keywords: Atrial fibrillation; Direct oral anticoagulants; Nomogram; Outcome; Stroke.