Addition of brain and carotid imaging to the ABCD2 score to identify patients at early risk of stroke after transient ischaemic attack: a multicentre observational study

Background The ABCD(2) score improves stratification of patients with transient ischaemic attack by early stroke risk. We aimed to develop two new versions of the score: one that was based on preclinical information and one that was based on imaging and other secondary care assessments. Methods We analysed pooled data from patients with clinically defined transient ischaemic attack who were investigated while in secondary care. Items that contribute to the ABCD(2) score (age, blood pressure, clinical weakness, duration, and diabetes), other clinical variables, carotid stenosis, and abnormal acute diffusion-weighted imaging (DWI) were recorded and were included in multivariate logistic regression analysis of stroke occurrence at early time intervals after onset of transient ischaemic attack. Scores based on the findings of this analysis were validated in patients with transient ischaemic attack from two independent population-based cohorts. Findings 3886 patients were included in the study: 2654 in the derivation sample and 1232 in the validation sample. We derived the ABCD(3) score (range 0-9 points) by assigning 2 points for dual transient ischaemic attack (an earlier transient ischaemic attack within 7 days of the index event). C statistics (which indicate discrimination better than chance at >0.5) for the ABCD(3) score were 0.78 at 2 days, 0.80 at 7 days, 0.79 at 28 days, and 0.77 at 90 days, compared with C statistics for the ABCD(2) score of 0.71 at 2 days (p=0.083), 0.71 at 7 days (p=0.012), 0.71 at 28 days (p=0.021), and 0.69 at 90 days (p=0.018). We included stenosis of at least 50% on carotid imaging (2 points) and abnormal DWI (2 points) in the ABCD(3)-imaging (ABCD(3)-I) score (0-13 points). C statistics for the ABCD(3)-I score were 0.90 at 2 days (compared with ABCD(2) score p=0.035), 0.92 at 7 days (p=0-001), 0.85 at 28 days (p=0-028), and 0.79 at 90 days (p=0.073). The 90-day net reclassification improvement compared with ABCD(2) was 29.1% for ABCD(3) (p=0.0003) and 39.4% for ABCD(3)-I (p=0.034). In the validation sample, the ABCD(3) and ABCD(3)-I scores predicted early stroke at 7,28, and 90 days. However, discrimination and net reclassification of patients with early stroke were similar with ABCD(3) compared with ABCD(2). Interpretation The ABCD(3)-I score can improve risk stratification after transient ischaemic attack in secondary care settings. However, use of ABCD(3) cannot be recommended without further validation.