ARTICLE IN BRIEF
A new meta-analysis suggests that the ABCD2 score — a clinical risk assessment tool commonly used to determine which patients with a transient ischemic attack are at a high risk of early recurrent stroke — does not reliably discriminate between those with high and low risk.
The ABCD2 score, a tool commonly used to triage patients with transient ischemic attack (TIA), does not reliably discriminate between TIA patients who are at high and low risk of recurrent stroke, researchers from the United Kingdom reported in a meta-analysis published in the July 1 online edition of Neurology.
The scale, which factors in and assigns points for five independent risk factors — age, blood pressure, clinical features of TIA, TIA duration, and presence of diabetes — also does not reliably identify patients with carotid stenosis or atrial fibrillation (AF) needing urgent intervention, distinguish true TIA from mimics, or streamline clinic workload, they reported. (See sidebar, “How the ABCD2 Score Is Calculated,” for more information.)
“We found that about one-fifth of the people who have a serious underlying stroke risk factor, like carotid stenosis, and who should have that acted on as quickly as possible, end up in the ‘slow stream’ [for stroke assessment and treatment] because of their ABCD2 score. And they may have their stroke while they're waiting to be assessed,” study author Joanna M. Wardlaw, MD, director of Neuroimaging Sciences and the Brain Research Imaging Centre in the Centre for Clinical Brain Sciences at the University of Edinburgh in Scotland, told Neurology Today.
Similarly, the ABCD2 score may find that people who actually have a mimic of stroke or TIA are at high risk, putting them in the “fast” stream, Dr. Wardlaw said. “So it blocks up the fast stream with people who don't actually have the disease you're interested in at all.”
This is a particularly significant finding for health care professionals in places like the UK, where stratifying patients to faster or slower assessment based on the ABCD2 score is financially incentivized, she said.
Many stroke prevention guidelines recommend specialist assessment within 24 hours of TIA or minor stroke for patients with a score of 4 or more, and within one week for patients with lower scores, Dr. Wardlaw and her colleagues noted. Clinical trials involving the ABCD2 score also tend to use a score of 4 as a “cut point.” However, as Dr. Wardlaw and others have noted, the score was never specifically designed for triage at a specific cut point.
STUDY METHODOLOGY, RESULTS
Dr. Wardlaw and her colleagues conducted a systematic review and meta-analysis of all studies published between January 2005 and September 2014 that reported on patients with TIA, minor strokes, or mimics, as well as risk factors and recurrent stroke rates, stratified by an ABCD2 score of less than or greater than/equal to 4. They then calculated the “effect per 1,000 patients triaged” on stroke prevention services. They sought to examine the score's ability to predict stroke recurrence in patients classified as high or low-risk for stroke recurrence, differentiate patients with mimics from true stroke or TIA, identify carotid stenosis or AF, and its effect on proportions of patients entering fast- or slow-track assessment in the clinic.
The researchers identified 29 relevant studies involving 13,766 TIA patients. The data were heterogeneous and not always of the highest caliber, they noted; 48 percent of the studies calculated the ABCD2 score retrospectively, and few reported on the score's ability to identify TIA mimics or use by non-specialists.
Using the cut point of 4 points or greater, the researchers found the ABCD2 score had a sensitivity of 86.7 percent (95% confidence interval [CI], 81.4-90.7%) for predicting stroke at seven days and 85.4 percent (95% CI, 81.1-88.9) at 90 days, but had only 35.4 percent specificity (95% CI, 33.3-37.6%) at seven days and 36.2 percent specificity (95% CI, 34.0-37.6) at 90 days.
The score did not account for patients with significant carotid stenosis or atrial fibrillation — crucial factors that would influence a patient's need for immediate assessment and treatment, Dr. Wardlaw and colleagues noted. They found that up to 20 percent of patients with an ABCD2 score of less than 4, suggesting a lower risk for early recurrent stroke, had more than 50 percent carotid stenosis or atrial fibrillation.
In addition, the researchers found that 35 to 41 percent of patients with TIA mimics had an ABCD2 score of 4 points or more. “The problem is that about half of the patients that get referred to stroke prevention clinics don't even have a TIA or a minor stroke,” Dr. Wardlaw said. “They have a mimic like a migraine or a bit of viral dizziness. It's a difficult area, and a lot of the assessment hangs on the expert examination and interpretation of the patient's history.”
The researchers calculated that for every 1,000 patients referred to a stroke clinic, the majority — 520, or 52 percent — would have an ABCD2 score of 4 or more, thus requiring fast-track assessment, which suggests the scale would not significantly reduce the workload in a stroke center's fast track channel, Dr. Wardlaw said. And this group would include approximately 171 patients with stroke or TIA mimics.
Given these results, she said, “maybe what we should be doing [instead] is focusing on getting an expert to do a rapid evaluation, get the relevant diagnostic tests done — like carotid imaging, a brain scan if necessary, assessing the heart to see if there's a cardiac source of embolism that needs specific treatment — and not rely too much on the ABCD2 score, as a low score may be falsely reassuring.”
Independent experts who reviewed the study agreed that the ABCD2 score seems to underperform in the clinic, and said that more effort should be made to get patients with suspected TIA to an expert for rapid evaluation to define the underlying TIA or stroke mechanism.
The ABCD2 score “is probably the most commonly used clinical risk stratification tool for TIA patients,” said Kevin M. Barrett, MD, an associate professor of neurology at the Mayo Clinic in Jacksonville, FL, and co-author of an editorial accompanying the study in Neurology. “But this study emphasizes that as a standalone tool, it underperforms. The specificity for diagnosis of true TIA was fairly low using the cut point of greater than or equal to 4. The score doesn't take into consideration true TIA versus a mimic, and it doesn't fully capture the importance of the underlying stroke mechanism, namely atrial fibrillation and carotid stenosis. When used as a triage tool, it also would not significantly decrease the clinic workload, because more than half of the patients would be identified as high-risk and require expedited evaluation based on the score alone.
“The clinical implication is that for patients with transient neurologic symptoms, the true emphasis should be identifying those patients with high-risk mechanisms — carotid stenosis and atrial fibrillation — and as best as possible differentiating between a TIA and a mimic in a timely fashion,” he said. “That's the way we're going to most efficiently capture and implement preventive strategies to reduce the burden of stroke going forward.”
Philip B. Gorelick, MD, MPH, FAAN, a professor of translational science and molecular medicine at Michigan State University College of Human Medicine and medical director of Mercy Health Hauenstein Neurosciences in Grand Rapids, MI, agreed, emphasizing the role of imaging in accurately assessing stroke risk. For instance, he said, “when MRI with diffusion-weighted imaging and extra-cranial carotid artery non-invasive imaging are added to the ABCD2 score, the ability of the score to predict stroke recurrence (i.e., high-risk persons) is enhanced. The focus for the clinician who is dealing with an individual patient should be on defining the underlying TIA or stroke mechanism.”
But another expert noted that the ABCD2 score, despite its limitations, has had a somewhat beneficial effect in the assessment of TIA patients. “It has made people take a careful history, and taking a good history should never be underestimated,” said Shelagh B. Coutts, MD, an associate professor of neurology in the Hotchkiss Brain Institute at the University of Calgary in Canada, and Dr. Barrett's co-author on the Neurology editorial. She agreed with Dr. Barrett that the score misses key clinical variables and ignores important questions, such as whether the TIA was a result of AF, carotid stenosis, or another mechanism such as small vessel disease. “The etiology of a TIA is what's really key, and that's why we emphasized in our editorial that obtaining brain and vascular imaging immediately is the thing that will make the most difference here,” she said.
“The ABCD2 score was never designed to say, ‘Patient A should be seen now, and Patient B later,’” she added. “It was just a score predicting risk. It was never designed to be a triage tool. The score has been used in a way that was never intended.”
“This isn't the end of clinical risk stratification,” Dr. Barrett added. Efforts to develop and validate stroke risk assessment scores that account for carotid stenosis and atrial fibrillation are ongoing. If successful, these could greatly improve triage and treatment of patients with suspected TIA, and could potentially improve patient outcomes over the long term.
“The message here is straightforward,” said Larry B. Goldstein, MD, FAAN, FANA, FAHA, professor and chairman of neurology and co-director of the Kentucky Neuroscience Institute at the University of Kentucky in Lexington. “Treat the patient, not the number.”
EXPERTS: ON THE ABCD2 SCORE FOR PREDICTING STROKE RISK