Prior to the recent advent of stroke-risk prediction rules for use in patients following transient ischemic attack, physicians did not correctly identify the majority of patients at highest risk of stroke following a transient ischemic attack, and their admission decisions were consequently somewhat haphazard (Josephson et al 2008). Before the availability of such stroke-risk prediction rules, the clinical factors associated with physician decisions to hospitalize patients after transient ischemic attacks included prior transient ischemic attack, speech impairment, weakness, gait disturbance, history of atrial fibrillation, presence of symptoms on arrival to the emergency room, and use of ticlopidine.
Since 2000, several similar rules for predicting stroke risk after a transient ischemic attack, which variously estimate stroke risk within 3 months (Johnston and Gress 2000) or within 1 week (Rothwell et al 2005; Selvarajah et al 2008), have been developed. However, risk of stroke following a transient ischemic attack is greatest in the first 2 days following the attack, the period when up to 85% of post-transient ischemic attack strokes occur. Early models also included either hypertension (the ABCD prediction rule) or diabetes (the California prediction rule), but not both, and both have subsequently been short to have prognostic value.
In 2007, the earlier California and ABCD scores provided elements for a more robust prediction rule called the ABCD2 prediction rule, which included both blood pressure and diabetes as prognostic factors (Johnston et al 2007). The ABCD2 prediction rule has been well validated on multiple independent cohorts and is a more accurate predictor of stroke risk than either of the earlier prediction rules.
ABCD2 Scoring Criteria:
|Age||Age 60 or older||1|
|Blood pressure||SBP 140 mm Hg or higher and/or DBP 90 mm Hg or higher||1|
Speech impairment without weakness
|Duration of symptoms||
60 minutes or longer
10 to 59 minutes
Abbreviations: DBP = diastolic blood pressure; SBP = systolic blood pressure
2-Day Stroke Risk by ABCD2 Score:
|Score||2-Day Stroke Risk (%)|
|0 or 1||0.0*|
|2 or 3||1.3*|
|4 or 5||4.1|
|6 or 7||8.1|
*The combined 2-day stroke risk for low-risk patients with scores of 0 to 3 was 1.0%.
ABCD2 scores by themselves are not adequate for diagnosis of transient ischemic attack (and they were never intended for this purpose), although among patients with suspected cerebrovascular events there is a positive association between higher ABCD2 scores and cerebrovascular diagnosis (Quinn et al 2009). Higher ABCD2 scores are associated with vascular lesions on brain imaging and moderate to severe carotid disease but not with atrial fibrillation. Low ABCD2 scores may suggest that noncerebrovascular diagnoses should be further considered.
The original paper on the ABCD2 prediction rule suggested that the rule could be useful in deciding whether to admit a patient with a transient ischemic attack (Johnston et al 2007). Hospitalization can be used, for example, for close monitoring of patients with a transient ischemic attack to ensure rapid administration of tissue plasminogen activator in the event that a stroke occurs soon after the onset of the attack, to ensure rapid evaluation using diagnostic technologies supported by current practice guidelines (Easton et al 2009), and to facilitate urgent application of other appropriate interventions in selected cases (eg, carotid endarterectomy). Hospitalization after transient ischemic attack is reported to be cost-effective if the 24-hour risk of stroke is greater than 4% (Nguyen-Huynh and Johnston 2005), which corresponds to an ABCD2 risk-stratification score of higher than 3 (Johnston et al 2007).
Although no randomized trial has evaluated the benefit of hospitalizing transient ischemic attack patients or the utility of the ABCD2 risk-stratification score in triaging patients for hospitalization, a scientific statement from the American Heart Association, American Stroke Association Stroke Council, and other organizations concluded that hospitalization was "reasonable" if patients with a transient ischemic attack present within 72 hours of onset and any of the following is present (Easton et al 2009):
None of the available prediction rules for stroke risk following a transient ischemic attack incorporate neuroimaging results (Sciolla and Melis 2008; Easton et al 2009). Because neuroimaging has also been shown to have prognostic utility (Easton et al 2009), it is likely that further refinements of prediction rules for stroke risk following a transient ischemic attack are possible and will be forthcoming.