The ABCD2 score, used to identify people at risk of stroke after TIA, fails to distinguish between high and low risk patients, warn researchers from the University of Ottawa in Ontario, Canada.
They described the tool as ‘unacceptable’ for clinical use.
Patients with TIA are at increased risk of imminent stroke but the risk varies between individuals.
The ABCD2 test combines clinical scores developed in three studies to predict stroke risk. These include age, BP, clinical features, duration of symptoms and diabetes.
NICE recommends the test to determine whether urgent treatment should be initiated.
Researchers led by Dr Jeffery Perry examined whether the test is an accurate and sensitive predictor of stroke risk after TIA.
They tracked 2,056 patients with TIA or minor stroke from 2007 to 2010.
In this time, 1.8% of patients had a stroke within seven days of a first TIA, and 3.2% within 90 days. An additional 9.8 % had a second TIA between seven and 90 days.
Researchers found that, using an ABCD2 score of four as the threshold for high risk, the test correctly identified just 65.8% of patients at high risk of stroke within seven days.
Similarly, the test correctly identified just 63.1% of patients at high risk of stroke within 90 days.
Dr Perry concluded: ‘We believe that the sensitivities we found are too low to be clinically acceptable.’
The study authors said: ‘The ABCD2 score can be credited for increasing awareness of transient ischemic attack as a medical emergency that carries with it a substantial and modifiable risk for subsequent stroke.’
But they said their study showed that the criteria used to calculate the ABCD2 score ‘are not sensitive enough to classify patients as being at low risk’
They added that doctors 'frequently miscalculated' the score and misclassified patients' risks. This occurred largely due to difficulties allocating points for history of unilateral weakness that had been resolved at point of contact.
Professor Peter Rothwell of Oxford University led a study to create the first form of the tool, ABCD.
He said the present version was the best available predictor of stroke risk and that recent work to find additional predictive measures to improve accuracy had proved fruitless.
He added that the tool’s simplicity was an advantage since it allows fast triage of at risk patients. ‘It’s is as simple as we could make it without compromising on predictive value,’ he said.
Dr Claiborne Johnston of the University of California, San Francisco led research to improve the tool's predictive power by adding an assessment of diabetes - 'D' - to make ABCD2.
He said: ‘I think the most important and worrisome novel finding of the study is the imprecision of the score in the hands of the treating physicians.
'It is surprising that something so simple could be incorrectly applied so frequently. It suggests we need to do a much better job teaching people how to apply the score.’
He added that clinicians needed something ‘much better’ than ABCD2. ‘It may be useful now, at least in the right hands, but we need more reliable tools.’
But he defended the tool’s use, saying: ‘We should never expect any tool to perform perfectly. There is too much chance in this and also too much variability in the predictors of stroke risk.’