Unified Score More Accurately Predicts Stroke After TIA

BOSTON—Although existing prognostic scores for stroke risk after transient ischemic attack (TIA) are well validated, the unified ABCD2 score is more appropriate for use in initial clinical triage, according to findings presented at the 59th Annual Meeting of the American Academy of Neurology.

S. Claiborne Johnston, MD, PhD, Director of the Stroke Center at the University of California, San Francisco, reported that it is important for clinicians to “target the group that is going to have the stroke versus the group that is not, in order to make expensive and potentially risky decisions.”

Dr. Johnston said there are several problems with the existing California and ABCD scores. The California score identifies five factors—age older than 60, diabetes, symptom duration longer than 10 minutes, weakness, and speech impairment. The ABCD score evaluates stroke risk using four factors: Age older than 60; Blood pressure ≥ 140/90 mm Hg; Clinical: unilateral weakness and/or speech disturbance without weakness; and Duration of symptoms. Both scores are validated and highly predictive only in their local populations (Northern California and Oxfordshire, United Kingdom, respectively). The California score optimizes prediction at 90 days, whereas the ABCD score optimizes at seven days. According to Dr. Johnston, predicting risk at two days is more relevant to clinical care.

Because having two separate prediction rules might create confusion, Dr. Johnston’s research team, which created the California score, collaborated with the authors of the ABCD score to combine their derivation cohorts (1,707 patients from Northern California; 209 patients from Oxfordshire) to develop a unified score. They did so by looking at models with different combinations of independent predictors of stroke and selecting the score that was most predictive of two-day stroke risk. They found that the model that best predicted risk of stroke was the ABCD2 score, which Dr. Johnston described as “the ABCD score plus Diabetes.”

In both derivation cohorts, C statistics were superior for the unified score. The ABCD2 score predicted seven-day stroke risk well (C statistics, 0.64 to 0.81), with risk ranging from 0% at a score of 0 to 1, to 11% at a score of 6 to 7. The ABCD2 score also predicted 90-day stroke risk well (C statistics, 0.64 to 0.75), with risk ranging from 1.7% at a score of 0 to 1, to 18% at a score of 6 to 7.

According to Dr. Johnston, another research group has already validated the ABCD2 score. “This was a group that had already validated ABCD, so it wasn’t too hard for them to redo it for ABCD2.”

Dr. Johnston also noted that the ABCD2 score does not consider imaging findings. “I think that that’s going to be an important addition to these scores in the future that may either add to them or change them.”           

NR