ABCD2 Score Predicts Severity Rather Than Risk of Early Recurrent Events After Transient Ischemic Attack

Background and Purpose-The ABCD(2) score predicts the early risk of stroke after transient ischemic attack (TIA). However, data on the severity of recurrent events would also be useful. Do patients with high scores also have more severe early recurrent strokes, perhaps further justifying hospital admission? Do patients with low scores have a low early risk of recurrent TIA as well as recurrent stroke? Methods-We completed a prospective, population-based study in Oxfordshire, England, of 500 consecutive patients presenting with TIA from April 1, 2002, by using multiple methods of case ascertainment (Oxford Vascular Study). Recurrent TIA, minor stroke, and major stroke (National Institutes of Health Stroke Scale score >3 at the time of first assessment) were identified by face-to-face follow-up. Predictive value was expressed as the area under the receiver operating characteristic curve. Results-Of 500 patients with TIA, 55 had a recurrent TIA (11.0%; 95% CI, 8.3% to 13.7%) and 50 had a recurrent stroke (10.0%; 95% CI, 7.5% to 12.0%) within 7 days. The ABCD(2) score was highly predictive of major recurrent stroke (area under the receiver operating characteristic curve=0.80; 95% CI, 0.72 to 0.87, P<0.0001), weakly predictive of minor stroke (area under the receiver operating characteristic curve=0.57; 95% CI, 0.43 to 0.71, P=0.26), and inversely related to risk of recurrent TIA (area under the receiver operating characteristic curve=0.37; 95% CI, 0.29 to 0.44, P<0.001) (overall heterogeneity, P<0.0001). The score predicted stroke-related disability, length of stay for recurrent stroke, and hence, overall acute hospital care costs. Conclusions-The ABCD(2) score predicts severity of recurrent events after TIA, high scores being associated with major recurrent stroke and low scores with high rates of recurrent TIA. These findings have implications for cost-benefit analyses of policies on hospital admission for patients with high scores and for the advice given to patients with low scores. (Stroke. 2010;41:851-856.)