Elsevier

Annals of Emergency Medicine

Volume 55, Issue 2, February 2010, Pages 201-210.e5
Annals of Emergency Medicine

Neurology/original research
A Multicenter Evaluation of the ABCD2 Score's Accuracy for Predicting Early Ischemic Stroke in Admitted Patients With Transient Ischemic Attack

Presented at the ACEP Research Forum, October 2008, Chicago, IL.
https://doi.org/10.1016/j.annemergmed.2009.05.002Get rights and content

Study objective

We evaluate, in admitted patients with transient ischemic attack, the accuracy of the ABCD2 (age [A], blood pressure [B], clinical features [weakness/speech disturbance] [C], transient ischemic attack duration [D], and diabetes history [D]) score in predicting ischemic stroke within 7 days.

Methods

At 16 North Carolina hospitals, we enrolled a prospective, nonconsecutive sample of admitted patients with transient ischemic attack and with no stroke history, presenting within 24 hours of transient ischemic attack symptom onset. We conducted a medical record review to determine ischemic stroke outcomes within 7 days. According to a modified Rankin Scale Score, strokes were classified as disabling (>2) or nondisabling (≤2).

Results

During a 35-month period, we enrolled 1,667 patients, of whom 373 (23%) received a diagnosis of an ischemic stroke within 7 days. Eighteen percent (69/373) of all strokes were disabling. We were unable to calculate an ABCD2 score in 613 patients (37%); however, our imputed analysis indicated this did not significantly alter results. The discriminatory power of the ABCD2 score was modest for ischemic stroke in 7 days (c statistic 0.59), and fair for disabling ischemic stroke within 7 days (c statistic 0.71). Patients characterized as low risk according to ABCD2 score (≤3) were at low risk for experiencing a disabling stroke within 7 days, with a negative likelihood ratio of 0.16 (95% confidence interval [CI] 0.04 to 0.64) with missing values excluded and 0.34 (95% CI 0.15 to 0.76) when missing values were imputed.

Conclusion

Our analysis suggests the best application of the ABCD2 score may be to identify patients at low risk for an early disabling ischemic stroke. Further study of the ability to determine an ABCD2 score in all patients is needed, along with validation in a large, consecutive population of patients with transient ischemic attack.

Introduction

A systematic review of 18 independent patient cohort studies concluded that the 7-day risk of a completed ischemic stroke among patients receiving a diagnosis of a transient ischemic attack is 5%.1 Because most patients receiving a diagnosis of transient ischemic attack are not at high short-term risk of a completed stroke after their initial event, stroke prediction tools could be valuable in guiding emergency medicine practice and the urgency of transient ischemic attack evaluation. Among transient ischemic attack clinical risk stratification scores, the ABCD2 score has emerged as the most refined tool, consisting of a simple 7-point risk score derived from age (A), blood pressure (B), clinical features (weakness/speech disturbance) (C), transient ischemic attack duration (D), and diabetes history (D).2 However, it lacks independent, prospective validation in an adequately powered, broad patient population.2, 3, 4

Our study evaluates the accuracy of the ABCD2 score for predicting risk of ischemic stroke within 7 days in a large, multicenter population of patients admitted to the hospital with transient ischemic attack. Because no criterion standard exists that distinguishes minor stroke from transient ischemic attack, we included disabling stroke as an important patient-oriented outcome because it is uncontroversially distinguished from transient ischemic attack. Additionally, because a proposed application of the ABCD2 score among emergency medicine practitioners is to identify patients at low risk for early stroke,5 we analyzed the sensitivity of a low-risk ABCD2 score to identify patients with transient ischemic attack and low risk for early stroke.

Section snippets

Study Design

Our study used prospective, nonconsecutive, sample cohort enrollment and data collection, with medical record review at 90 days to determine outcome.

Setting

The North Carolina Collaborative Stroke Registry is one of 4 state registries funded by the Centers for Disease Control and Prevention in the Paul Coverdell National Acute Stroke Registry program. We invited all 25 North Carolina Collaborative Stroke Registry hospitals in 2005 to participate in our ancillary, separately funded project. Sixteen of

Results

During a 35-month period, we enrolled 1,667 patients presenting within 24 hours of symptom onset with a presumptive diagnosis of transient ischemic attack and no history of stroke (Table 1). A total of 373 patients (23%) received a diagnosis of an ischemic stroke within 7 days, of which 353 occurred within 2 days (91%). Eighteen percent (69/373) of all strokes occurring within 7 days were disabling. The majority of all strokes met the outcome classification of an imaging-positive stroke (97%,

Limitations

Despite the strength of our prospective enrollment and the large number of patients included from a variety of hospital settings, our study has several limitations. We limited our analysis to admitted patients with transient ischemic attack, who were urgently evaluated and treated. The number of patients with transient ischemic attack discharged home from the emergency department (ED) at each site and not enrolled is unknown. In an effort to determine the effect of this limitation, the study

Discussion

Several investigators have reported their experience with the ABCD2 score's predecessor rule (the ABCD score), with disparate results,15, 16, 20, 23, 29 which reinforces the need for prospective validation of the refined ABCD2 rule before widespread adoption by emergency physicians. Previous attempts to independently validate the ABCD2 score have been underpowered to reach conclusive results or have considered enrollment or follow-up periods beyond periods pertinent to ED decisionmaking.30, 31,

Appendix

References (43)

  • G. Zou

    A modified Poisson regression approach to prospective studies with binary data

    Am J Epidemiol.

    (2004)
  • R.G. Newcombe

    Two-sided confidence intervals for the single proportion: comparison of seven methods

    Stat Med.

    (1998)
  • D.B. Rubin

    Multiple Imputation for Nonresponse in Surveys

    (1987)
  • J.V. Tu et al.

    Impracticability of informed consent in the Registry of the Canadian Stroke Network

    N Engl J Med.

    (2004)
  • J.A. Kline et al.

    Electronic medical record review as a surrogate to telephone follow-up to establish outcome for diagnostic research studies in the emergency department

    Acad Emerg Med.

    (2005)
  • F. Purroy et al.

    Absence of usefulness of ABCD score in the early risk of stroke of transient ischemic attack patients

    Stroke

    (2007)
  • B.L. Cucchiara et al.

    Is the ABCD score useful for risk stratification of patients with acute transient ischemic attack?

    Stroke

    (2006)
  • P.J. Koudstaal et al.

    Clinical disagreement on the diagnosis of transient ischemic attack: is the patient or the doctor to blame?

    Stroke

    (1989)
  • C.L. Kraaijeveld et al.

    Interobserver agreement for the diagnosis of transient ischemic attacks

    Stroke

    (1984)
  • S.B. Coutts et al.

    Recurrent events in transient ischemic attack and minor stroke: what events are happening and to which patients?

    Stroke

    (2008)
  • R. Sciolla et al.

    Rapid identification of high-risk transient ischemic attacks: prospective validation of the ABCD score

    Stroke

    (2008)
  • Cited by (46)

    • Clinical Policy: Critical Issues in the Evaluation of Adult Patients With Suspected Transient Ischemic Attack in the Emergency Department

      2016, Annals of Emergency Medicine
      Citation Excerpt :

      Therefore, the most compelling rationale to incorporate TIA risk stratification instruments into clinical practice is evidence that when used alone without additional history, physical examination, imaging, or laboratory testing, they may differentiate low-risk patients with TIA for whom advanced workup and specialty consultations can be deferred from those subsets who are at increased short-term risk (ie, 2 to 7 days) for stroke. Six TIA risk stratification instruments have been evaluated in studies that met the inclusion criteria: ABCD,9,11,13,24-30 ABCD2,6-13,17,26,27,31-47 ABCD3,12,27,38 the California,11,13,27,48 the Canadian TIA Score,42 and the Essen Stroke Risk.27 None of these instruments have been assessed in a Class I study.

    • Identification of Emergency Department Patients With Acute Heart Failure at Low Risk for 30-Day Adverse Events: The STRATIFY Decision Tool

      2015, JACC: Heart Failure
      Citation Excerpt :

      The STRATIFY model contains 13 variables. Although this may be too complex to remember, clinical calculators have been widely used in the ED for decision tools dealing with thromboembolic disease, stroke, and atrial fibrillation (67–69). We anticipate this having similar utility.

    • Inter-rater reliability and misclassification of the ABCD<sup>2</sup> score after transient ischemic attack

      2015, Journal of Stroke and Cerebrovascular Diseases
      Citation Excerpt :

      Clinical symptoms (C) and duration (D) displayed the greatest variability among all ABCD2 components. These results are consistent with prior reports.18-20 There was no significant difference among raters from different subspecialties, including neurology versus non-neurology–trained raters.

    View all citing articles on Scopus

    Please see page 202 for the Editor's Capsule Summary of this article.

    Provide feedback on this article at the journal's Web site, www.annemergmed.com.

    Supervising editor: Robert Silbergleit, MD

    Author contributions: AWA conceived the study, obtained research funding, and recruited participating centers. AWA, AMJ, WDR, CHT, and AF designed the trial. AWA, WDR, MFP, and CVM supervised the conduct of the trial and data collection. AWA, MFP, SS, CHT, and AF performed some data collection. MFP and CVM managed the data, including quality control. AMJ, WDR, KMR, and DC provided statistical advice on study design and supervised data analysis. AMJ performed all data analyses. AWA drafted the article, and AMJ, WDR, and KMR contributed substantially to its revision. AWA takes responsibility for the paper as a whole.

    Funding and support: By Annals policy, all authors are required to disclose any and all commercial, financial, and other relationships in any way related to the subject of this article that might create any potential conflict of interest. See the Manuscript Submission Agreement in this issue for examples of specific conflicts covered by this statement. This work was supported by the Emergency Medicine Foundation and the Foundation for Education and Research in Neurological Emergencies, along with a grant from Boehringer Ingelheim Pharmaceuticals, Inc.

    Publication date: Available online June 24, 2009.

    Reprints not available from the authors.

    View full text