The role of ABCD2 in predicting stroke recurrence in the Australian setting.

Abstract

Background: To determine the accuracy of the ABCD2 tool in predicting stroke outcome in an Australian cohort of patients. Method(s): This is a retrospective cohort study. The setting is an urban tertiary referral hospital in Melbourne, Australia. All adults presenting with suspected Transient Ischaemic Attack (TIA) who were triaged and managed via the Monash TIA Triaging Treatment (M3T) system between June 2004 and November 2007. The main outcome measure was stroke at 2 days and 90 days. Result(s): 513 patients were managed by the M3T system with TIA confirmed in 324. Complete follow-up was obtained in 481/513 (93.8%) of patients. Stroke occurred in 4/481 (0.83%, CI 0.2-2.1%) patients at 2 days, and 8/481 (1.66%, CI 0.7-3.3%)) patients at 90 days. All strokes occurred in patients with confirmed TIA. The Area Under the ROC (AUC) in the entire M3T cohort was 0.61 (CI 0.43 to 0.80) for stroke at 90 days and 0.63 (CI 0.36 to 0.90) for stroke at 2 days. The AUC in those with confirmed TIA was 0.53 (CI 0.34 to 0.73) for stroke at 90 days and 0.55 (CI 0.26 to 0.84) stroke at 2 days. Conclusion(s): The ABCD2 tool poorly predicts stroke outcome in our Australian cohort of TIA patients.