Validation and refinement of the ABCD2 score: a population-based analysis.

BACKGROUND AND PURPOSE Transient ischemic attacks are a frequent diagnosis in the emergency department setting, yet expert opinion as to the proper follow-up and need for hospitalization differs widely. Recently, an effort has been made to risk-stratify patients presenting with transient ischemic attacks through scoring systems such as the ABCD and ABCD2 scales. The aim of our study was to independently validate these scores using a population-based cohort. METHODS Using the data from the Rochester Stroke and Transient Ischemic Attack Registry and resources of the Rochester Epidemiology Project, medical records of all residents of Rochester, Minn, with a diagnosis of incident transient ischemic attack from 1985 through 1994 were examined (N=284). Patients were scored on the ABCD and ABCD2 scales and new scores were created by adding hyperglycemia and a history of hypertension. The end points of stroke and death were collected previously and were verified through the Rochester Epidemiology Project data. RESULTS Although our study did find that scores >4 had a statistically significant predictive value for future stroke, a substantial proportion of strokes within 7 days (9 of 36 cases [25%]) occurred in patients with low or intermediate risk scores (< or =4) on the ABCD2 scale. Including history of hypertension and hyperglycemia on presentation increased the sensitivity of the score to identify patients who had a stroke within 7 days. CONCLUSIONS Reliance on the ABCD and ABCD2 scores misses some patients who will have a stroke within 7 days of a transient ischemic attack. Adding hyperglycemia and a history of hypertension to the predictive model could be useful, but the value of these additions need to be evaluated further.