ABCD 2---too easy to be true?

Clinicians have long relied on the ABCD 2 score to help determine which patients who present to the ER with TIA can be sent home for completion of their work up as an out patient. But the tool had never been prospectively validated. A new study in CMAJ tried to do just that. The result? If you use the traditional cut off of a score of 5 the sensitivity is lousy. If you use AHA's recommended score of 2 the sensitivity improves considerably but the specificity is poor.


A related Medscape piece offers some insightful comments from one of the study authors and new perspectives on today's stroke care:

The investigators found that an ABCD2 score of more than 5 had low sensitivity (31.6%; 95% confidence interval [CI], 19.1% – 47.5%) for predicting subsequent stroke at 7 days. For predicting stroke at 90 days, its sensitivity was 29.2% (95% CI, 19.6% – 41.2%). "These sensitivities are too low to be clinically acceptable," Dr. Perry said...

"The proposed threshold by the American Heart Association, which is a score greater than 2 to indicate high risk, was very sensitive; however, it classified all but a few patients as high risk, so it is not very discriminating for early stroke," Dr. Perry noted.

Which means you might as well admit just about everybody, if not to the hospital to a stroke obs unit, if you have one, or hold them in the ER long enough to do more extensive imaging, more than just a CT scan.


How would the score perform if integrated with aggressive imaging in the ER? That's a question for further study.