Background: Nearly 240,000 U.S. adults are diagnosed every year with transient ischemic attack (TIA). Four to 20 percent of these patients experience a stroke within the following 90 days; one half of those strokes occur within 48 hours. Promptly recognizing patients who are at high risk of progressing to stroke could provide the opportunity for preventive interventions. It also could avoid costly and unnecessary treatment of patients who are at low risk.

The two leading scoring systems for the prognosis of stroke after TIA are the California and ABCD scores. Both scores rely on points associated with clinical factors that are independently predictive of a patient's risk of stroke, but these scores have significant overlap. The California score aims to predict total 90-day risk, but the ABCD score focuses on the more urgent risk of stroke within two days of the TIA. Johnston and colleagues aimed to validate the two scoring systems and to generate a new score that could provide a single standard for clinical application.

The Study: The study included 4,809 patients with TIA divided into two groups. The first group included 1,916 patients whose data were used in the original derivation of the ABCD and California scores. These data also were used to generate a new unified score. The second group included 2,893 patients presenting with TIA at four clinical sites, and their data were used to validate the scoring systems. Two subgroups of patients were drawn from members of a health maintenance organization in California (1,707 who presented to the emergency department and 962 who were diagnosed in an ambulatory clinic). Two other subgroups were patients of family physicians (209) and those from a specialty clinic (315) in Oxford, England. The original treating physician diagnosed the patients with TIA. Subsequent strokes were confirmed by expert review of clinical records and imaging reports. For each participant, the California and ABCD scores were compiled from clinical records. The statistical analysis compared total scores and the performance of each component item as predictors of subsequent stroke. All combinations of the components were tested, and a new prognostic score that combined elements from both scoring systems was developed and validated.

Results: There were 442 participants (9.2 percent) who developed strokes within 90 days of TIA, but only 189 (3.9 percent) occurred within two days. Nearly all strokes (99 percent) were ischemic. Ninety of these patients (20 percent) died from their stroke within 90 days. Both scores predicted stroke risk effectively, but the ABCD scoring system performed slightly better than the California system. However, the difference was not statistically significant.

Examining the components of each scoring system demonstrated that age greater than 60 years was an important prognostic factor, followed by systolic pressure of 140 mm Hg or greater, or diastolic pressure of 90 mm Hg or greater; unilateral weakness or speech impairment; duration of TIA; and diabetes. The researchers allocated points to formulate a new scoring system, ABCD². Each participant was retrospectively scored using the new system to test its ability to predict stroke (see accompanying table). The new score performed better than the California and the ABCD scores. Although the score appeared to perform equally well in patients of different ethnicities, race was not recorded for all participants.

Conclusion: The California and ABCD scores can reliably predict the short-term risk of stroke in patients presenting with TIA. However, the ABCD² scoring system is likely to be the best predictor of stroke. The authors believe this score should be used to identify patients for more aggressive evaluation and urgent intervention.