Chapter 48
Transient Ischemic Attack

Ian T. Ferguson and Christopher R. Carpenter

Department of Emergency Medicine, Washington University School of Medicine, St. Louis, MO, USA

Highlights

Transient ischemic attack (TIA) precedes 23% of strokes; among emergency department (ED) patients, the post‐TIA stroke rates are 3% at 2 days and 6% at 7 days.
Early registry‐based validations of the ABCD² post‐TIA clinical decision rules were promising, but methodologically rigorous clinical decision rule research has demonstrated that the ABCD² is nonspecific at recommended thresholds and that different cutoff values do not improve the ABCD² prognostic accuracy.
There are no current validated and acceptably accurate TIA prognostic decision aids for the ED population.
Clinical‐plus scores incorporate imaging into the risk stratification, although these scores require further validation for ED use.
Rapid ED‐based diagnostic protocols may be a safe and cost‐effective way to manage patients with suspected TIA.

Background

Historically, the definition of a transient ischemic attack (TIA) included a symptom resolution timeframe of less than 24 hours.¹ This symptom‐based definition has been challenged in favor of a tissue‐based definition with the advent of higher resolution imaging such as magnetic resonance imaging (MRI) with diffusion‐weighted imaging (DWI). TIA is now defined by the American Heart Association (AHA) as “a transient episode of neurological dysfunction caused by focal brain, spinal cord, or retinal ischemia, without acute infarction.”^1,2 This new definition has no time restriction for symptoms, although 70% of all TIAs resolve within 2 hours of onset.³ TIA precedes 23% of strokes, and 200,000–500,000 strokes occur annually.^4,5 The diagnostic and therapeutic management of TIA as a medical emergency is a new concept because TIAs were traditionally not linked to strokes. On average, the annual risk of post‐TIA stroke is 3–4%,⁶ though this is probably a conservative estimate. Johnston et al.⁷ reported a 3‐month risk of 10.5% with half of those occurring within the first 2 days. Emergency department (ED) patients have higher post‐TIA stroke rates than other populations: 3.1% at 2 days and 5.8% at 7 days.⁸ Thus post‐TIA stroke risk stratification models have become a priority amongst neurovascular emergency specialists, and rapid test–treat models are now being evaluated in a variety of healthcare settings.

The symptoms of TIA relate to the brain region suffering ischemia rather than the hemorrhagic, embolic, or atherosclerotic etiology of the low flow state, so causative inferences are not possible at the bedside. Potential etiologies of cerebral ischemia include cardiac emboli from valvular disease or atrial fibrillation (10–15%), large vessel extracranial arterial disease (20–25%), and small vessel intracranial atherosclerosis (10–15%).⁶ The first two etiologies can be identified by echocardiography, telemetry, and carotid Doppler. Unfortunately, in 50% of cases, the cause of TIA remains undefined after diagnostic evaluation.⁷ Observational studies and randomized trials suggest that stroke specialist care, rapid administration of antiplatelet and lipid‐lowering agents, and ED‐initiated diagnostic testing within 24 hours reduce 1‐year post‐TIA stroke rates.^5,8–12 One potential benefit of admitting TIA patients is that in‐hospital thrombolysis using the National Institute of Neurological Disorders and Stroke (NINDS) or European Cooperative Acute Stroke Study III (ECASS‐III) protocols is more likely in the inpatient setting.^13–15 However, debate continues regarding the cost‐effectiveness of hospitalizing TIA patients as opposed to evaluating them in same‐day clinics.^16,17

If TIA patients at increased risk for short‐term stroke could be rapidly and accurately identified in ED settings, cost‐effectiveness might be less controversial, but simply diagnosing TIA remains challenging. One prospective series found that 20% of all suspected TIAs in the ED were TIA mimics, with the two most common being seizures and migraine.⁹ Less frequent causes included hypoglycemia, psychogenic hyperventilation, sepsis, and transient global amnesia.⁹ Consequently, TIAs are frequently misdiagnosed in the ED by neurologists and emergency physicians.^18–20 Features associated with a discordant diagnosis of TIA between neurologists and emergency physicians include headache, involuntary movement, and dizziness, while a high ABCD² score increases the odds of concordance.²¹ MRI with DWI identifies ischemic lesions in one‐third of patients even with symptom resolution <1 hour and increases the risk of future stroke, although this imaging modality is not always readily available.^1,2 In addition, practice variability and resource constraints limit the uniform application of rapid diagnostic and therapeutic pathways on all TIA patients.^22,23

Clinical question

In suspected TIA, can decision aids identify patients at low‐ or high risk of subsequent stroke within 2 or 7‐days?

The American College of Emergency Physicians (ACEP) latest TIA Clinical Policy provides no Level A recommendations for the use of decision aids to identify low‐risk patients with suspected TIA who can safely be discharged from the ED. A Level B recommendation is to not rely on current risk stratification tools such as the ABCD² score for ruling out high‐risk patients with suspected TIA.²⁴

As of this 2016 ACEP policy, six clinical decision instruments aimed at stratifying the risk of patients with TIA for the development of future stroke had been identified. These are the ABCD, ABCD², ABCD³, California Rule, and the Canadian TIA Score.^7,10–13 Registry data were used to derive and validate the California Rule and the original ABCD Rule. The ABCD² rule was derived using components of both the California and original ABCD rules and validated in four different registry‐based cohorts (Table 48.1).^11,14,15. A systematic review of early validation trials for the ABCD² observed negative likelihood ratios (LRs) of 0.24 and 0.35 for risk of stroke in 7 and 90 days respectively using the recommended cut‐off of <3.¹⁶ However, the only trial to follow established methodological rigor for validating a decision instrument prospectively yielded a negative LR of 0.42 for stroke risk in 7 days using the same cut‐off of <3 (Table 48.2)^17,18 Although sensitive, the poor specificity observed for a cut‐off of >2 would mean that 87.6% of all patients evaluated for TIA would require emergent investigations from the ED.¹⁷ Therefore, the ABCD² should not be used to stratify TIA patients as either high or low risk for stroke within 7 days.

Further modification of the ABCD² score using clinical features alone has not improved accuracy,¹⁹ prompting efforts to develop clinical‐plus tools which aim to combine clinical features and imaging findings. A number of these scores have been published, including the clinical‐ and imaging‐based prediction (CIP) model, recurrent risk estimator (RRE), ABCD²‐I, ABCD³‐I, and Canadian TIA score.^12,20–23 The ABCD³‐I requires external validation but showed increased discrimination over the ABCD² score in the initial derivation sample (C statistic 0.92 versus 0.71).¹² The Canadian TIA score incorporates noncontrast computed tomography (CT) and may also be more discriminatory than the ABCD² score, although it also requires external validation. This score has a total score of −3 to 23 and a negative LR of 0.08 with a score of 6 or less.¹³ At this time, neither current risk‐stratification instruments nor noncontrast head CT should be used alone to identify ED patients with TIA at low‐risk or high risk for stroke at either 2 or 7 days.²⁴

Table 48.1 Transient ischemic attack (TIA) risk stratification instruments

Source: Data from [14].

California rule
Age >60 years	1 point
Diabetes mellitus	1 point
Symptoms >10 minutes	1 point
Weakness	1 point
Speech deficit	1 point
ABCD rule
Age >60 years	1 point
Blood pressure >140/90	1 point
Unilateral weakness	2 points
Language disturbance	1 point
Without weakness duration
>60 minutes	2 points
10–59 minutes	1 point
ABCD² rule
Age >60 years
Blood pressure >140/90	1 point
Diabetes	2 points
Unilateral weakness	2 points
Language disturbance	1 point
Without weakness duration
>60 minutes	2 points
10–59 minutes	1 point