Background and purpose Most patients with stroke-like symptoms screened by advanced imaging for proximal occlusion will not have a thrombus accessible by neurointerventional techniques. Development of a sensitive clinical scoring system for rapidly identifying patients with an emergent large vessel occlusion could help target limited resources and reduce exposure to unnecessary imaging.
Methods This historical cohort study included patients who underwent non-contrast CT and CT angiography in the emergency department for stroke-like symptoms. NIH Stroke Scale (NIHSS) criteria were extended to include resolved symptoms and dichotomized as present or absent. Combinations of NIHSS criteria were considered as tests for proximal occlusion.
Results Proximal cerebral vascular occlusion was present in 19.2% (100/522) of the population and, of these, 13% (13/100) had an NIHSS score of 0. The presence on examination or history of diminished consciousness with inability to answer questions, leg weakness, dysarthria, or gaze deviation had 96% sensitivity and 39% specificity for proximal occlusion. If implemented in this population, the use of CT angiography would have been decreased by 32.4% (169/522 patients) while missing 0.76% with proximal occlusions (4/522). Half of those missed (2/4) would have been identified as large vessel infarcts on non-contrast CT, while the remainder (2/4) were transient ischemic attacks associated with carotid stenosis.
Conclusions In this cohort, specific NIHSS criteria were highly sensitive for emergent large vessel occlusion and, if validated, may allow for clinical screening prior to advanced imaging with CT angiography.
- CT Angiography
Statistics from Altmetric.com
Treatment of acute ischemic stroke is focused on minimizing the time to recanalization of the occluded vessel, which has been associated with improved outcomes.1–3 The recent success of second-generation mechanical thrombectomy devices has been attributed in part to decreased time to reperfusion for stent retrievers compared with earlier devices.3 Triage of patients presenting with stroke-like symptoms often includes advanced imaging (non-contrast CT (NCCT)/CT angiography (CTA)/CT perfusion (CTP)) to quickly identify those with large vessel occlusions who may benefit from mechanical thrombectomy.3 However, nearly half of the strokes screened by CTA have no proximal occlusion or are hemorrhagic.4 ,5 Universal screening of patients suspected of stroke with CTA is not logistically feasible for many practices and it carries the added cost and inherent risk of contrast enhanced CT.
Ideally, a clinical score could exclude a subset of patients presenting with focal signs and symptoms who do not have large vessel occlusion, thus allowing scarce resources and advanced imaging to be targeted to those who may benefit from emergent intravascular treatment. The NIH Stroke Scale (NIHSS) has been validated as a marker for stroke severity and outcome prediction,6–8 with higher scores predictive of a proximal occlusion and worse outcomes.9–12 However, the negative predictive value (NPV) is low, making it a poor screening tool for triaging stroke patients for advanced imaging.4 ,10 ,11 Recent trials such as MR CLEAN have shown the benefit of mechanical thrombectomy in patients with NIHSS scores as low as 2.13 The presence of collateral flow can result in a range of severity of presenting clinical symptoms with up to 10% of proximal occlusions having an NIHSS V.0.4 As blood pressure and collateral perfusion change, so may the severity of symptoms and NIHSS score. This emphasizes the need for any developed clinical scoring system to screen for both presenting signs and resolved symptoms.
Ischemia affecting certain vascular territories results in predictable signs and symptoms. While the NIHSS score has proved a poor tool for ruling out proximal occlusion, individual components may be of use. Four individual NIHSS components (level of consciousness, gaze, motor leg, and extinction/inattention) have been shown to be independent predictors of proximal occlusion.11 We hypothesized that the presence of any of these criteria on examination (physical signs) or history (resolved symptoms) would have a sensitivity exceeding 95% for proximal large vessel occlusion.
Materials and methods
A historical cohort study was conducted for consecutive patients who presented to the Erlanger Hospital Emergency Department (Chattanooga, Tennessee, USA) between 1 January 2012 and 31 July 2012 meeting the following inclusion criteria: age at least 18 years, stroke-like signs or symptoms, evaluation by board-certified neurologist within 12 h of presentation and prior to imaging, and CTA of the head and neck. No additional exclusion criteria were applied. Stroke-like symptoms were defined as focal neurologic signs or symptoms (persistent or resolved) or altered mental status suspicious for an ischemic event. Patients who received CTA for isolated headache (rule-out aneurysm) or craniofacial trauma (rule-out vascular damage) were not included.
Gender, age, NIHSS score, and discharge neurologic diagnosis were collected to characterize the population. For each patient, 13 exposures of interest (NIHSS criteria: level of consciousness overall/questions/commands, best gaze, visual, facial palsy, motor arm, motor leg, limb ataxia, sensory, best language, dysarthria, and neglect) were dichotomized to present (NIHSS component score >0) or absent based on the pre-imaging history and physical examination by a neurologist. An NIHSS criterion was considered present if documented as a physical sign on examination or resolved symptom on history. Motor arm and motor leg were each considered as single criteria that were positive if either limb had weakness on examination or history. Baseline neurologic deficits were not counted as positive if they were documented as present prior to symptom onset. The NIHSS has been previously validated for retrospective chart review where examinations were not performed with the NIHSS in mind.14–16
The primary outcome of acute occlusion or ≥70% stenosis in a proximal vessel (internal carotid artery, middle cerebral artery 1st or 2nd segment, vertebral artery, basilar artery) was determined by review of the CTA report and, where not specified, by review of the imaging by a board-certified neuroradiologist (BWB or SDQ) blinded to presenting signs and symptoms. Proximal occlusion was dichotomized to present or absent and excluded chronic occlusions.
In addition to the hypothesized set of criteria, other combinations of NIHSS criteria were considered as tests for proximal occlusion. A test was positive if the patient had at least one of the test's criteria on history or examination. Sensitivity, specificity, positive predictive value (PPV), and NPV were calculated for each combination of NIHSS criteria. All tests with sensitivity exceeding 95% were then sorted by specificity. This study received Institutional Review Board approval.
Of 522 patients (mean age 63±15.2 years, range 21–98; 46.9% male) presenting to the emergency department with stroke-like symptoms warranting neurology consultation, 36.6% (191/522) were acute ischemic infarcts, 22.8% (119/522) were transient ischemic attacks (TIAs), and 40.6% (212/522) were of non-thrombotic etiology (eg, seizure, migraine, conversion disorder). Occlusion of a proximal cerebral vessel was present in 19.2% (100/522) of the population, comprising 45.5% (87/191) of ischemic infarcts and 10.9% (13/119) of TIAs. Eighty-seven of the 100 patients with proximal occlusion had focal signs on examination while 13 had only a history of symptoms (NIHSS score 0). The mean (SD) NIHSS score for the population was 4.1 (4.9) (range 0–27; median 2, IQR 0–6). Patients with ischemic infarct without proximal occlusion had a mean (SD) NIHSS score of 5.3 (4.2) (range 0–18; median 4, IQR 2–7), while those with proximal occlusion had a mean (SD) NIHSS score of 10.5 (6.0) (range 0–27; median 10, IQR 6–16).
Our preselected test with four criteria failed to meet the required 95% sensitivity (see table 1); however, several other combinations of NIHSS criteria were found to exceed this threshold. The most specific tests with three, four, and five criteria are listed in table 1. While tests with more than five criteria had high sensitivity, they lacked specificity and would be impractical to use as a quick screening metric. The test with greatest specificity (of those exceeding 95% sensitivity) was composed of four criteria and differed from our hypothesis by including dysarthria rather than neglect. This screening test would have prevented 32.4% (169/522) of all CTAs performed in the emergency department for patients with stroke-like symptoms warranting a neurology consultation, while missing 0.76% (4/522) of the population with proximal occlusion. Of these, half (2/4) would have been detected by NCCT with hyperdense artery sign or cytotoxic edema, while the remainder were TIAs with carotid stenosis. Tests with more than four criteria had increased sensitivity at the cost of specificity. Conversely, no tests with fewer than four criteria met the threshold of 95% sensitivity.
Acute stroke places a large burden on the USA healthcare system, with an annual incidence of 795 000 and combined direct and indirect cost of $34.4 billion in 2010. Improvements in early detection and treatment led to a 34.8% decrease in the annual death rate between 1998 and 2008; however, stroke remains the fifth leading cause of death in the USA with an annual mortality of 128 978 in 2013.5 ,17 Proximal vessel occlusions are associated with mortality rates as high as 80% without reperfusion,18–20 and often necessitate endovascular intervention to achieve recanalization.21 As earlier recanalization is known to improve outcomes, many stroke centers follow the American Stroke Association guidelines to select diagnostic tests to minimize time to treatment and perform advanced imaging in any patient being considered for endovascular intervention.2 ,3 ,22 Currently, those ‘being considered for endovascular intervention’ include all patients with focal signs or symptoms as determined by the evaluating clinician. The majority of these will not have proximal occlusion.
In our institution, 80.8% of CTAs performed for stroke-like symptoms in patients receiving neurology consultation did not show proximal occlusion. Our mean NIHSS score of 4.1 was lower than the established global mean of 5–6 for patients presenting to the emergency department with stroke, which is explained by the fact that 40.6% of our population underwent CTA for non-ischemic symptomatology deemed ‘stroke-like’ on initial evaluation. Our subgroup of ischemic infarcts had a mean NIHSS score of 5.3 with 45.5% proximal occlusions, which is consistent with prior studies.4 ,23 However, real-world CTA screening for proximal thrombi is not limited to patients with ischemic events, and the challenge is to provide clinicians with a tool to safely narrow the pool of patients being considered for endovascular intervention.
The NIHSS is an obvious choice for this purpose, having been validated as a marker for stroke severity and proved accurate when used by non-neurologists in an emergent setting.6–8 ,24 However, multiple studies have found that threshold scoring fails to accurately rule out proximal occlusion.10 ,11 For example, one study with digital subtraction angiography (DSA) (N=43) found an NIHSS score of ≥10 to be indicative of proximal occlusion, however 36.4% of patients with an NIHSS score of <10 still had proximal occlusion.10 Another study using DSA (N=226) showed that an NIHSS score of ≥12 had a PPV of 91% for proximal occlusion but only 81% sensitivity.11 The use of DSA in these studies probably resulted in selection bias, leading to NIHSS scores (mean 14.710 and median 1411) much higher than the global average of 5–6 for patients presenting to the emergency department with acute ischemic stroke.23 A subsequent study with a more representative population (N=699, mean NIHSS score 7.5) who received advanced imaging in the emergency department for ischemic stroke found that an NIHSS score ≥10 had an 81% PPV but only 48% sensitivity. All patients with NIHSS score ≥2 would have to undergo CTA to detect 90% of proximal occlusions. Furthermore, 6.5% of patients with proximal occlusion had NIHSS scores of 0.4 Identification of these patients is especially important as TIAs with occlusion have a 46% chance of recurrent stroke within 90 days (compared with 5.8% without occlusion) and an even higher rate of diffusion-weighted imaging lesions at 30 days.25 The authors of that study conclude that “excluding patients presenting with low NIHSS scores from [advanced imaging] would lead to a failure to diagnose the majority of arterial lesions amenable for alternative therapies”.4
Our study demonstrates that the presence of specific signs or symptoms, rather than a score representing severity of symptoms, may have utility in determining which patients should go to advanced imaging for suspected proximal occlusion. This metric, which we call the Emergent Large Vessel Occlusion (ELVO) score, would be simple and easily implemented by emergency departments and pre-hospital medical staff. To achieve 96% sensitivity, only four criteria were necessary (see table 1); five criteria allowed for 99% sensitivity. In the absence of four specified criteria, nearly one-third of CTAs could have been deferred in this population. This percentage, along with the NPV, is likely to increase when the population is expanded to include all patients receiving advanced imaging for stroke-like symptoms rather than only those meriting a neurology consultation. We required neurologic evaluation to increase the accuracy of our data, however the population most likely to benefit from such a screening tool is the subset of patients sent for CTA by the emergency physician or mid-level provider. That population is likely to be disproportionately composed of non-ischemic pathology and non-proximal ischemic events.
As with any retrospective study, these results lack generalizability beyond this cohort. The cohort design allowed accurate characterization of NPV and PPV for patients receiving a neurology consultation; however, when the population is expanded, these values will change. Additionally, our predetermined hypothesis did not meet the sensitivity threshold, and those tests that were found to exceed 95% sensitivity will require prospective validation, which is currently ongoing as part of the prospective Advanced Imaging Metric For Acute Stroke Triage (AIM FAST) study.
While the ELVO score fell short of 100% sensitivity in this historical cohort, the false negatives would have been captured by subsequent investigation with NCCT or dedicated carotid imaging. The purpose of this screening metric is to quickly triage patients who potentially have ELVOs requiring immediate intervention. Those with TIAs or focal signs/symptoms not meeting our criteria should undergo standard evaluation with NCCT, followed by cardiac and carotid evaluation if indicated. Symptomatic moderate to severe carotid stenosis is at high risk of progression to stroke and should receive carotid endarterectomy or stenting within 14 days.26 ,27 However, the procedure need not be emergent as there is an increased perioperative risk if done within 2 days.28 Carotid duplex ultrasound has been recommended as first-line imaging in these patients, followed by contrast-enhanced MR angiography if necessary or CTA if there is a contraindication to MRI.29
Patients presenting without specific signs or a history of specific symptoms are unlikely to have proximal arterial occlusion. Several subsets of criteria based on the NIHSS had a high sensitivity and NPV for excluding large vessel occlusion in our historical cohort, requiring validation in a currently ongoing prospective study. Use of such an easily administered scoring system may allow a more targeted approach to patient selection for advanced stroke imaging to identify patients with ELVO for interventional therapy. Importantly, the use of this scoring system would significantly reduce the radiation, contrast exposure, and cost of performing advanced imaging on patients whose symptomatogy is non-ischemic in nature.
Contributors RDM: study design, data collection, data analysis, manuscript authoring, editing, submission. JCJ: study design, data collection, manuscript authoring, editing. SLV: data collection, manuscript authoring, editing. SDQ: study design, analysis, editing. BWB: study design, manuscript authoring, editing. All authors have provided final approval of the version to be published and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Competing interests None declared.
Ethics approval Ethics approval was obtained from the Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.