Acute Vestibular Syndrome: Does my patient have a stroke? A Systematic and Critical Review of Bedside Diagnostic Predictors

Transcription

1 Acute Vestibular Syndrome: Does my patient have a stroke? A Systematic and Critical Review of Bedside Diagnostic Predictors Alexander A. Tarnutzer MD 1, Aaron L. Berkowitz PhD 1, Karen A. Robinson PhD 2, Yu-Hsiang Hsieh PhD 3 and David E. Newman-Toker MD, PhD 1,3 1. Department of Neurology, The Johns Hopkins University School of Medicine, Baltimore, MD, USA 2. Department of Medicine, The Johns Hopkins University School of Medicine, Baltimore, MD, USA 3. Department of Emergency Medicine, The Johns Hopkins University School of Medicine, Baltimore, MD, USA Running title: Acute vestibular syndrome word count for the text: 6040 (excluding boxes, figures, tables and legends) character count for the title: 49 (plus 57 characters for the suggested subtitle) number of figures: 2 number of boxes: 2 number of tables: 3 number of appendices: 4 number of references: 104 Corresponding author: David E. Newman-Toker, MD, PhD, Associate Professor, Department of Neurology, The Johns Hopkins Hospital, Meyer 8-154, 600 North Wolfe Street, Baltimore, MD toker@jhu.edu

2 Key points: There are hundreds of thousands of patients who suffer from acute vestibular syndrome (AVS) annually in North America, and the most common causes appear to be vestibular neuritis and ischemic stroke in the brainstem or cerebellum. Vertebrobasilar ischemic stroke often presents with dizziness, including AVS, and may closely mimic vestibular neuritis and related benign peripheral vestibular disorders. Current best evidence indicates that obvious focal neurological signs are absent in more than half of all AVS presentations that result from posterior fossa ischemic stroke. Cranial CT identifies hemorrhage but detects fewer than 1 in 6 acute strokes. Brain MRI with diffusion-weighted imaging is falsely negative in 10-20% in the first hours. Expert opinion suggests the initial approach to evaluating AVS patients for stroke should employ a combination of focused history and physical examination techniques. Best available evidence, albeit limited, indicates a three-component bedside oculomotor examination (H.I.N.T.S. horizontal Head Impulse test, Nystagmus, Test of Skew) identifies stroke with high sensitivity and specificity in AVS, ruling out stroke more effectively than MRI with diffusion-weighted imaging. The Case A 45-year-old man presents to the emergency department with 18 hours of new, continuous vertigo, nausea, vomiting, and unsteady gait. He prefers to lie motionless. He denies auditory or neurological symptoms, headache, neck pain, or recent trauma. There is no relevant past medical or exposure history, including no cerebrovascular or vestibular disorders, no recent or remote ear surgery, and no smoking or other vascular risk factors. He takes no medications. He has no family history of vestibular disease or recurrent dizziness. While obtaining the history, you note his eyes jerking horizontally. Neither neurology consultation nor neuroimaging is

3 readily available to you. In this patient with acute dizziness, how should you focus your exam to efficiently stratify the patient s risk of stroke as a cause? Dizziness/vertigo is the third most common major medical symptom reported in general medical clinics 1 and accounts for approximately 3-5% of visits across care settings. 2 This translates to 10 million ambulatory visits per year in the US for dizziness 3 with roughly 25% of these to emergency departments. 2 Many patients have transient or episodic symptoms lasting seconds, minutes, or hours, but some have prolonged dizziness persisting continuously for days to weeks. 4 When dizziness or vertigo develops acutely, is accompanied by nausea or vomiting, gait instability, nystagmus, and head-motion intolerance, and persists for a day or more, the clinical condition is known as acute vestibular syndrome (AVS). 5,6 Note that transient dizziness has a differential diagnosis distinct from AVS (Figure 1), and the diagnostic approach to transient dizziness should differ accordingly. 7 This review focuses on AVS. Most patients with AVS have an acute, benign, self-limited, presumed viral or post-viral vestibular condition usually called vestibular neuritis, but sometimes referred to as vestibular neuronitis, labyrinthitis, neurolabyrinthitis, or acute peripheral vestibulopathy. 5,6 Some authors distinguish labyrinthitis from vestibular neuritis based on the presence of auditory symptoms accompanying the AVS presentation, but this distinction is inconsistently applied and the terms are often used interchangeably. Here we include labyrinthitis and vestibular neuritis together as peripheral causes of AVS i.e., pathology localized to the inner ear (labyrinth) or eighth cranial (vestibular) nerve, as distinguished from pathology affecting central nervous system vestibular connections referred to as central. We further define isolated AVS (with or without hearing loss) as occurring in the absence of focal neurologic signs such as hemiparesis, hemisensory loss, or

4 gaze palsy. Although peripheral causes are more common, dangerous central causes, particularly ischemic stroke in the brainstem or cerebellum, can mimic benign peripheral causes closely. 6,8-12 Regional variation in diagnostic practice is probably commonplace, 3 but little is known about factors influencing dizziness diagnosis (e.g., access to technology, availability of specialists, nature of training, cultural or linguistic differences). There is growing evidence that many AVS patients are misdiagnosed and that frontline physicians around the world are eager for diagnostic guidelines. 18,19 The evidence base for diagnosis of patients with dizziness is limited. 20 We are unaware of any systematic reviews, practice parameters, or fully-validated clinical decision rules applicable to unselected patients with acute, prolonged dizziness offering evidence-based guidance for diagnosis and management of AVS. Narrative reviews highlight the importance of accurately assessing the risk of dangerous disorders, particularly posterior fossa ischemic strokes, and emphasize the utility of a focused history and physical examination in these patients. 5,21-23 To generate an evidence-based approach to diagnostic assessment of AVS patients, we performed a systematic search and synthesis of the medical literature (Box 1), focusing on bedside diagnostic predictors. Where evidence derived from our systematic review is incomplete, we cite expert opinion and critically review related evidence to support or refute such opinion. Evidence failing to meet strict inclusion criteria was considered part of the critical review. Epidemiology and Differential Diagnosis We found no direct studies of the incidence of AVS as a clinical presentation. Vestibular neuritis, likely the most common cause of AVS, has an estimated annual incidence of 3.5 per 100,000 population based on a single retrospective survey of neuro-otology clinics in Japan. 24

5 Data from a nationally-representative sample of US emergency departments indicate that, of the 2.6 million visits for dizziness or vertigo annually, peripheral AVS (i.e., vestibular neuritis or labyrinthitis) is diagnosed in about 6%, corresponding to ~150,000 US visits per year. 2,16 Another 4% receive a cerebrovascular diagnosis and 22% leave the emergency department with no etiologic diagnosis (i.e., diagnosed with dizziness or vertigo ), 2,16 many of whom likely presented with AVS-type symptoms. From a German general population survey, 11% of those reporting dizziness complained of symptom duration greater than a day. 4 Data from a US-based study of consecutive emergency department patients with dizziness 25 indicate that, of patients with any dizziness in the 24 hours prior to their visit, 27% (n=47/175) still had dizziness in the emergency department that had not remitted since it began (Newman-Toker, unpublished). Thus, we estimate that ~10-20% of patients presenting with acute dizziness have AVS, corresponding to ~ ,000 emergency department visits per year in the US alone. In our systematic review we found no studies of all AVS presentations, and only three enrolled relatively unselected, consecutive populations. 6,8,10 The most common diagnoses in these AVS studies are listed in Table 1. Central mimics of vestibular neuritis (sometimes called pseudoneuritis 10 ) were predominantly cerebrovascular (83%) and demyelinating (11%). Two studies prospectively enrolled high-risk-for-stroke patients using age or vascular risk factors as entry criteria, 6,8 likely leading to over-representation of cerebrovascular patients; the third used a case-control design and did not report the method for sampling controls. 10 All other studies included in the systematic review (the remaining 7 of 10) focused on acutely dizzy patients with a diagnosis of stroke. 9,26-32 None of the included studies was large enough to identify rare but important causes such as Wernicke syndrome. 33 Box 2 shows a suggested differential diagnosis

6 for AVS adapted from narrative reviews written by specialists in the field of vestibular disorders. 23,34 Our systematic review data do not allow a definitive statement about the relative prevalence of vestibular neuritis versus stroke among unselected AVS presentations. We can, however, calculate a rough range for the proportion of AVS presentations with stroke, using data on the annual incidence of stroke and prevalence of dizziness among stroke victims. Of 795,000 strokes per year in the US, 35 ~18% are located in the posterior fossa, 36 and ~50-70% are likely associated with dizziness as a prominent symptom. 21,37 Thus, we estimate there are ~70-100,000 strokes per year in the US with dizziness as a prominent symptom. Considering the approximate incidence of AVS presentations calculated above (~ ,000 per year US) we estimate the true fraction of stroke in AVS to be roughly 25 ±15%. Clinical History Predictors Related to the Chief Symptom (Table 2) Although not evidence-based, classic teaching 62 and current practice 18 in the US divides dizziness into four types based on symptom quality, each said to predict the underlying etiology: (i) vertigo (false sense of spinning or motion), (ii) presyncope, (iii) unsteadiness, and (iv) nonspecific/other dizziness. 63,64 This differs slightly from European practice 25 and international consensus research criteria 65 that define dizziness and vertigo separately. In keeping with current North American practice, 18 for this review we use dizziness as an umbrella term that includes vertigo as a subset, recognizing, however, that recent data suggest drawing any linguistic distinction between dizziness and vertigo probably has little diagnostic value. The type or quality of dizziness is typically the main focus for directing diagnostic inquiry in patients with dizziness, 66 as described in numerous textbooks and review articles 23 and

7 reported in a multicenter clinical practice survey of over 400 emergency physicians. 18 We identified no studies that met inclusion criteria for our systematic review examining dizziness type as a predictor of etiology in AVS. However, a critical review of the literature suggests this is not an evidence-based practice. 23 An emergency department-based study in over 300 consecutive patients with acute dizziness has shown the type of dizziness to be an imprecise metric more than half of patients are unable to reliably report which symptom type most accurately reflects their experience. 25 More importantly, symptom quality does not appear to be a trustworthy predictor of underlying etiology. In a population-based study of emergency department dizziness not focused specifically on AVS, patients with unsteadiness as part of their symptom complex were at slightly higher risk of stroke, but vertigo versus other dizziness predicted stroke with equal likelihood. 13 These results accord with disease-based studies (reviewed elsewhere 25 ) that indicate the type of dizziness is inconsistently described by patients with disorders known to cause AVS such as vestibular neuritis, 67 posterior fossa stroke, 37,42 and cerebellar hemorrhage. 68 To meet the clinical definition of AVS, total duration should exceed 24 hours of continuous dizziness. This excludes most disorders in which dizziness typically presents with transient episodes lasting seconds, minutes, or hours, such as benign paroxysmal positional vertigo, Menière disease, vestibular migraine, cardiac arrhythmia, or transient ischemic attack. Thus, these disorders rarely remain diagnostic considerations in AVS patients beyond the first few hours. Vestibular migraine may be an exception, since two large case series suggest that up to 27% of patients report dizziness persisting for more than 24 hours. 43,44 Some experts suggest that abrupt onset of dizziness in AVS favors a vascular cause, whereas vestibular neuritis produces symptoms more gradual in onset, over hours. 5 While we

8 found no studies comparing onset timing across disorders, two uncontrolled, retrospective case series identified in our critical review reported that dizziness was typically abrupt in onset among patients with a cerebrovascular cause of acute vertigo. 38,39 We identified no studies in our systematic review reporting on prodromal dizziness. A critical review of other evidence on transient isolated episodes of prodromal dizziness preceding an AVS presentation suggests a rate of about 25% in vestibular neuritis 40,41 and a similar rate in patients with vertebrobasilar stroke. 26,38,39 As a harbinger of stroke, these episodes presumably represent transient ischemic attacks. While a single episode of transient dizziness within a few days prior to AVS presentation is probably non-specific (reported with high frequency in both vestibular neuritis and stroke), recurrent episodes lasting seconds to minutes over the preceding weeks or months may favor a cerebrovascular cause. Repetitive events have been reported in two retrospective case series of cerebrovascular dizziness patients; 38,39 in the larger series, 29% (n=12/42) of vertebrobasilar stroke patients had prodromal, isolated attacks of vertigo for up to two years prior to the stroke. 39 Recent literature indicates most strokes after transient ischemic attacks occur within the first 90 days of the index event. 69 Excess stroke risk after an emergency department discharge with a benign dizziness (mis)diagnosis declines rapidly within 60 days and returns to baseline by 6 months. 17 Therefore, repetitive events over a more protracted period could suggest a lower-risk pathomechanism such as episodic hemodynamic insufficiency superimposed on slowly-progressive vasculopathy. In patients with transient dizziness, there is little controversy that dizziness triggers represent a key source of diagnostic information (e.g., for those with benign paroxysmal positional vertigo or orthostatic hypotension which are position-provoked 7 ). A clinical practice survey of emergency department physicians 18 and a case-control study comparing generalist to

9 specialist knowledge 45 suggest that overgeneralization of this rule to include AVS leads to the common misconception that exacerbation of symptoms with head movement (as with the Dix- Hallpike positional test for benign paroxysmal positional vertigo) predicts a peripheral cause. Although our systematic review identified no data on triggers as a predictor of the underlying cause of AVS, specialists agree that AVS patients tend to be intolerant of head motion during the acute stage, regardless of the underlying localization as peripheral or central. 45 In keeping with expert opinion, exacerbation of symptoms or signs with head motion or positional testing has been reported in both vestibular neuritis 41 and stroke. 70 Clinical History Predictors Related to Associated Symptoms and Risk Factors (Table 2) Narrative reviews generally suggest that the presence of neurologic symptoms (or signs) indicates a central cause of AVS, and some authors have gone so far as to suggest that their absence indicates a peripheral cause. 71,72 Most studies in our systematic review considered only those with isolated AVS, excluding patients with major neurologic symptoms beyond the core features of AVS (dizziness/vertigo, nausea/vomiting, gait disturbance). Those that included patients with general neurologic symptoms or signs (e.g., diplopia, numbness) found them to be strongly associated with central causes, but their absence was a relatively poor predictor of peripheral etiology. 6,27,30 Proportionality of dizziness, neurovegetative, and gait/postural symptoms has sometimes been cited by experts as typical of peripheral vestibular disorders, while central syndromes are said to occasionally be associated with disproportionate symptoms (e.g. severe gait impairment with a mild subjective sense of dizziness or vertigo). 21,23 Our critical review identified some weak evidence to support this contention. Authors of uncontrolled case reports and small case

10 series hypothesize that the presence of vomiting 42,73,74 or imbalance/gait unsteadiness 42 out of proportion to the degree of dizziness or nystagmus may be a marker for brainstem or cerebellar pathology, including stroke. The largest series we found addressing these associations (66 cerebellar infarcts) noted disproportionate vomiting or gait disturbance to be particularly common in patients with superior cerebellar artery (SCA) strokes (n=30). 42 Among 40% (n=12/30) with vomiting at stroke onset, only 42% (n=5/12) had vertigo; in the same study, 67% (n=20/30) had inability to walk or stand unaided, but only 37% of the total (n=11/30) had vertigo. 42 The study did not indicate, however, that all patients presented with AVS. Expert opinion grounded in well-established neuroanatomy suggests that auditory symptoms usually point to a peripheral localization in patients with dizziness (reviewed in Newman-Toker 23 ). Unfortunately, when translated into a bedside prediction rule, this anatomic fact may lead to the mistaken diagnosis of a more benign cause. 58 The vascular supply to the inner ear derives from the posterior cerebral circulation, and combined audio-vestibular presentations are known to occur in those with cerebrovascular disease. 22 Patients with auditory symptoms have generally been excluded from major studies of AVS or vestibular neuritis. 58 None of the studies included in the systematic review adequately assessed auditory symptoms as a specific predictor of peripheral versus central causes. Nevertheless, recent evidence suggests that auditory symptoms in AVS patients can and do result from inner ear ischemia, often in association with frank posterior circulation infarction in the anterior inferior cerebellar artery (AICA) territory. 21,27 In our systematic review we identified three studies from one research group reporting on transient dizziness, hearing loss, and/or tinnitus as a premonitory TIA that can precede audio-vestibular loss due to AICA infarction in up to 42% of cases Our critical review also identified a recent study in which patients with sudden hearing loss were found to be

11 at increased stroke risk (OR 1.64, 95% CI, 1.31 to 2.07) relative to appendectomy patients (general hospital population control) during the five year period following hospitalization. 75 Although not sought out as part of that study s design, the hearing loss presumably co-occurred with vertigo or vestibular symptoms in some patients. Narrative reviews suggest that craniocervical pain may accompany dizziness in posterior fossa strokes as a result of the stroke itself (mass effect or direct involvement of pain-sensitive structures) or its underlying cause (e.g., vertebral artery dissection or aneurysm). 21 Two studies included in our systematic review addressed this issue. Headache or neck pain was present in a minority of AVS presentations (38%, n=9/24 8 and 29%, n=29/101 6 ); a statistically-significant association with central causes (38% vs. 12%, p<0.05) was found in the larger study. A critical review of related literature indicates that vestibular migraine is a diagnostic consideration in almost any patient with dizziness and headache. 76 None of the systematically reviewed studies of AVS presentations diagnosed migraine as a cause, but our critical review found that young patients with vertebral artery dissection and posterior fossa stroke may be at particularly high risk of being misdiagnosed with migraine. 14 A recent systematic review indicates that dizziness, headache, and neck pain are the three most common symptoms of vertebral artery dissection. 46 The gradual onset of pain over hours and the disappearance of pain within hours typify migraine 77 and the headaches of vestibular migraine are often relatively mild. 78 By contrast, craniocervical pain was sudden in 57% (n=43/75), and severe in 60% (n=41/68) 48,51,52 of patients diagnosed with vertebral artery dissection, though diagnostic ascertainment bias might have enriched these proportions. Headache duration in patients with vertebral artery dissection lasted 2-35 days (n=18, mean 8.3 days, median 3 days) in one series, 49 and a sustained headache (i.e., >72 hours) was found in 100% (n=12) in another. 50 The absence of pain appears to be

12 diagnostically inconclusive with regard to ruling out vertebral artery dissection; roughly one in four dissection patients presents without head or neck pain. 46 When obtaining the medical history from a patient with AVS, one should identify any major known, concurrent illnesses (e.g., multiple sclerosis, HIV-AIDS, metastatic malignancy, severe malnutrition, major depression) or potentially-relevant exposures (e.g., viral syndrome, ear surgery, bacterial otitis media, or ototoxic medications [especially aminoglycoside antibiotics or chemotherapy agents]) in addition to assessing vascular risk factors. Our systematic search identified no evidence on the predictive value of any of these variables, though common sense dictates that certain ones (e.g., recent middle ear surgery) are more likely to be predictive than others (e.g., recent viral syndrome). The manuscripts included in our systematic review offered limited evidence regarding risk factor history as predictors of underlying etiology in AVS. As described above, stroke risk factors 6 and perhaps age 8 have been used in study design to increase the proportion of AVS patients with stroke, apparently successfully. 11 Though age is a known risk factor for stroke, there is evidence from our systematic review that stroke should still be considered in younger patients. In the largest prospective study of AVS patients, 25% (n=15/69) with ischemic stroke (and three of four patients with vertebral artery dissection as the underlying cause) were under 50 years of age. 6 From our critical review, younger patients with AVS may be more likely to be misdiagnosed than their older counterparts, with a small series of missed cerebellar infarctions finding that 50% (n=7/14) who presented with dizziness/vertigo were under age 50, all but one of whom had vertebral artery dissection or occlusion. 14 In related literature from our critical review, recent head or neck trauma is a known risk factor for vertebral artery dissection (estimated odds ratio 3.8 for minor trauma 55 ). Since vertebral artery dissection is a known cause of AVS, 6 a

13 history of trauma should spark concern for underlying dissection. 55 However, roughly half of symptomatic vertebral artery dissections occur without an identifiable history of trauma, 48 so the absence of trauma is insufficient to exclude the diagnostic possibility. Physical Examination Predictors (Tables 2,3) The presence of general neurologic findings has occasionally been touted by authors as key feature distinguishing peripheral and central causes of AVS. 71,72 The true prevalence of focal neurological signs in AVS patients is difficult to estimate, since symptoms or signs suggesting a central disorder influence patient selection in most studies. We identified six case series in our systematic review reporting focal neurological signs in stroke patients presenting with acute dizziness. 6,26-31 In aggregate, focal neurological signs were present in 80% (n=185/230), but diagnostic ascertainment bias in patients with obvious neurologic signs likely makes this proportion a substantial overestimate. The best available data come from the largest prospective study of AVS presentations (n=101). 6 Authors found obvious general neurological (e.g., facial palsy, sensory loss, limb ataxia, hemiparesis) or oculomotor (e.g., internuclear ophthalmoplegia, gaze palsy, vertical nystagmus) signs in 51% of 76 patients with central AVS, but in none of 25 with peripheral AVS. 6 Notably, some of these signs are not considered focal and might not have been obvious clinical findings to non-neurologists (e.g., severe truncal ataxia as opposed to moderate; nystagmus with a predominantly vertical or torsional, as opposed to horizontal, vector). Posterior circulation stroke symptoms and signs are generally underappreciated by patients, 80 physicians, 54 and even standardized stroke rating scales. 81 Some experts suggest that severe difficulty walking or inability to sit or stand unaided is a sign of central pathology in AVS. 21,82 This assertion is supported by some evidence from both

14 our systematic and critical reviews. The prospective study mentioned above found severe truncal ataxia (inability to sit with arms crossed unaided) as the only obvious neurologic sign in 29% of strokes and more frequent in stroke than vestibular neuritis (33% vs. 0%, p<0.001). 6 A retrospective, population-based study (not reporting the duration of dizziness) found imbalance/gait unsteadiness as part of the presenting complaint to be associated with stroke in acute dizziness (odds ratio 3.71, 95%CI ). 13 Our systematic review identified one other study demonstrating a high proportion of severe imbalance (71%, n=17/24) in AVS of cerebrovascular origin. 9 These results generally accord with several other studies of central AVS identified by our critical review, which, in aggregate, reported gait unsteadiness in 55% (n=185/334). 42,68,70,83 It remains unknown whether diagnostic ascertainment bias might have played a role in these apparently high frequencies. The ten studies identified by our systematic review focused largely on bedside tests of vestibular and oculomotor function as assessed through careful eye movement examination (see Online Appendix 3 for video links). A normal horizontal head impulse test of vestibulo-ocular reflex function 84 was the single best bedside predictor of peripheral versus central causes (Table 3). By itself, the head impulse result essentially matched MRI with diffusion-weighted imaging sensitivity for detecting stroke, with comparable specificity. 6 In AVS, an abnormal head impulse usually indicates a peripheral vestibular lesion while a normal response virtually confirms a stroke. 11 However, the studies included in our systematic review suggest that about 15% of patients with stroke or other central etiology might be mistakenly considered to have vestibular neuritis due to an abnormal head impulse if other eye findings are not considered. Based on known anatomic considerations, we anticipated heterogeneity in head impulse sensitivity for stroke depending on the precise posterior circulation vascular territory involved.

15 Prospectively-defined subgroup analysis by stroke vascular territory (Table 3) revealed that falsely abnormal head impulse findings (i.e., pointing to a peripheral vestibular lesion), which reduce the bedside test s sensitivity for stroke detection, are almost exclusively associated with anterior-inferior cerebellar artery infarctions. Anterior-inferior cerebellar artery stroke often results in an ischemic lesion of the labyrinth or eighth nerve root entry zone in the pons that mimics vestibular neuritis both anatomically and physiologically, although it can sometimes be recognized by the presence of associated unilateral hearing loss. 21,22 By contrast, the head impulse test is almost a perfect predictor when considering the more common posterior-inferior cerebellar artery stroke, which generally affects only the cerebellum or lateral medulla, sparing the structures required for an intact vestibulo-ocular response as assessed by the head impulse test. 6,21 Another demonstrated bedside predictor of central AVS is direction-changing horizontal nystagmus on lateral gaze (i.e., right-beating nystagmus in right gaze and left-beating nystagmus in left gaze, with or without nystagmus when the patient looks straight ahead); 5 this type of nystagmus generally reflects dysfunction of gaze-holding structures located in the brainstem and cerebellum. 85 Direction-changing horizontal nystagmus was reported in 6 of 10 included studies and correctly identified central causes with high specificity (92%) but low sensitivity (38%) (Table 3). Vertical ocular misalignment of vestibular cause ( skew deviation or skew ) during the alternate cover test was found to be a third bedside predictor for central AVS. Skew results from a right-left imbalance in otolithic/graviceptive inputs from the vestibular system to the oculomotor system, and, with rare exceptions, is generally central in origin. 86 Skew was reported in two of the included studies and, similar to direction-changing nystagmus, correctly identified central causes with high specificity (98%), but low sensitivity (30%) (Table 3).

16 We identified only one study in our systematic review that directly tested the combination of the three bedside oculomotor findings called H.I.N.T.S. by the authors (horizontal Head Impulse test, evaluation of Nystagmus, and Test of Skew). 6 Test results predicting stroke on the H.I.N.T.S. battery were labeled using a second acronym, I.N.F.A.R.C.T. (Impulse Normal or Fast-phase Alternating or Refixation on Cover-Test). This clinical prediction rule was developed based on neuroanatomic and physiologic principles combined with empiric analysis of results from the first 43 of 101 AVS patients studied 6 and other published literature on AVS 21 before being validated on the next 58 patients. Authors found that the presence of any one dangerous sign was 100% sensitive (n=76/76) and 96% specific (n=24/25) for stroke (negative likelihood ratio 0.00, 95%CI ), surpassing MRI with diffusion-weighted imaging for urgently ruling out stroke in AVS patients. According to the authors, these clinical tests can be performed in approximately one minute at the bedside. 6 Although not originally reported as a three-component battery, H.I.N.T.S. results from an earlier, independent study 10 were later published in a critical review of cerebellar stroke diagnosis (sensitivity 91% [n=21/23], specificity 78% [n=31/40]). 21 This earlier study did not repeat MRI scans in clinically-diagnosed vestibular neuritis cases to capture any false negative early MRIs, likely leading to some underestimation of H.I.N.T.S. battery specificity. Risk of diagnostic misclassification notwithstanding, the pooled sensitivity, specificity, and negative likelihood ratio of H.I.N.T.S. for stroke (98%, 85%, 0.02, 95% CI ) still suggest that bedside eye exam likely outperforms acute, early MRI with diffusion-weighted imaging to rule out stroke. Experts suggest that the vector, pattern, and fixation characteristics of spontaneous nystagmus can help distinguish peripheral from central causes of AVS. 5,87 In our systematic and critical review we found no substantive evidence to support or refute these assertions. Further

17 studies comparing spontaneous nystagmus characteristics (and bedside means to assess them such as Frenzel goggles 88 or penlight-cover test 87 ) in AVS patients are needed to determine their diagnostic utility in this context. We also found insufficient data to evaluate the usefulness of examining smooth pursuit eye movements, saccades, or optokinetic nystagmus. As alluded to above, experts agree that specific provocative testing procedures such as the Dix-Hallpike (Nylén-Bárány) positional maneuver are crucial in the evaluation of transiently dizzy patients to elicit symptoms (i.e., dizziness/vertigo) and signs (i.e., nystagmus), 7,82 but generally unhelpful diagnostically in patients with AVS as they fail to differentiate between central and peripheral causes. 45,82 Other forms of provocative testing to elicit nystagmus for diagnostic purposes (e.g., mastoid vibration, valsalva, head shaking 88 ) have also been described for use in patients with transient, mild, or residual symptoms, but have not been sufficiently studied as diagnostic predictors of underlying etiology. Based on evidence of frequent misdiagnosis and misconceptions in patients presenting with acute dizziness, 13,14,16,18,45,89 it seems prudent to recommend that none of these provocative maneuvers (including Dix-Hallpike) be routinely applied to patients with AVS in the acute care setting. Both physiologic and empiric evidence indicates provocative testing will lead to a non-specific worsening which could be misinterpreted as diagnostic of a peripheral vestibular disorder, 16,18 before stroke has been adequately excluded. Although not the primary focus of our systematic search, we identified several studies reporting on laboratory-based audio-vestibular function tests applied in acutely dizzy patients. These included pure-tone audiometry, auditory evoked potentials, and various oculomotorvestibular metrics using quantitative electro-oculography (Online Appendix 4). In aggregate,

18 there was insufficient evidence to support the routine use of any of these tests as a means to distinguish peripheral from central causes of AVS in the acute phase. Neuroimaging in the acutely dizzy patient None of the studies enrolling patients presenting with AVS of unknown cause compared different neuroimaging modalities across all patients. Our critical review found that although the predominant form of neuroimaging for patients with dizziness in the emergency department remains CT, 2 substantial evidence suggests this approach is grossly inadequate. 16 A rigorous, randomized-order diagnostic test study has shown that the sensitivity of CT for all acute ischemic stroke is only 16% compared to 83% for MRI with diffusion-weighted imaging with masked expert readers. 90 Imaging in AVS must assess acute lesions in the posterior fossa, a region where CT performs particularly poorly due to bone-related artifacts. 91 Although modern CT identifies acute intracranial hemorrhage with high sensitivity, 90 only about 1% of all AVS presentations are due to hemorrhage (Table 1). Not surprisingly, CT has been shown to have a low diagnostic yield in the assessment of emergency department patients with dizziness. 92,93 Although frequently viewed as the gold standard and final arbiter for stroke diagnosis, MRI with diffusion-weighted imaging can be misleading up to at least 48 hours after symptom onset in patients with AVS due to ischemic stroke. 6 Our systematic review found only one study in which patients with initially-negative MRI but evolving signs suspicious for central cause underwent repeat MRI; false negatives were confirmed by delayed MRI in 12% (n=8/69). 6 Combined with figures reported in two prior studies of MRI with diffusion-weighted imaging in acute vertebrobasilar strokes from our critical review (23% false negatives; n=47/206), 94,95 the aggregate posterior-fossa MRI with diffusion-weighted imaging sensitivity in the first ~24 hours

19 after symptom onset is estimated at 80% (n=220/275). Taking an aggregate MRI specificity of 97% from the two studies reporting on patients without stroke (n=198/205), 6,90 this gives an estimated negative likelihood ratio of 0.21 (95%CI ), making MRI less potent as a rule-out tool than the three bedside oculomotor predictors described above. 6 Consequences of diagnostic failure in AVS While details of management and treatment of AVS are beyond the scope of this review and are considered in detail elsewhere, 21,96 a brief discussion of prognosis and the relevance of diagnostic failure based on our critical review is warranted. Patients with cerebrovascular AVS are at risk for additional strokes and secondary complications of the initial stroke, particularly ischemic swelling following large cerebellar infarction. Ischemic stroke of the cerebellum 97 and cerebellar hemorrhage 98 can be fatal without close monitoring and urgent neurosurgical intervention at the time of clinical deterioration. A critical review on diagnosis and initial management of cerebellar stroke found 10-20% of patients deteriorate in the days following the event and swelling peaks on the third day post infarction. 21 Thus, stroke patients mistakenly thought to have peripheral AVS may appear clinically stable at the time of discharge from the emergency department yet be at risk for complications days later. Dizziness is the emergency department symptom most often linked to missed diagnosis of stroke, 89 with 35% of cerebrovascular events in patients with any dizziness (and 44% in those with isolated dizziness) missed at first medical contact. 13 The impact of these errors on patient health is incompletely known, but available data suggest those misdiagnosed are at particularly high risk for poor outcomes. The one small series of misdiagnosed cerebellar strokes (14 of 15 presenting with dizziness, vertigo, or imbalance) found 40% (n=6/15) ended in death 14 compared

20 to 5% (n=15/282) in the largest series of all-comers with cerebellar infarction 83 (p<0.0001, Fisher s exact). Although the small series of missed cerebellar infarctions was potentially biased by patient selection, the difference in outcomes occurred despite the fact that misdiagnosed patients presented with milder symptoms (normal alertness in 93% (n=14/15) 14 versus 69% (n=195/282) 83 of all comers with cerebellar stroke, p=0.046, Fisher s exact). If adverse case selection were to blame, we would expect initial presentations to have been more severe, since milder presentations are associated with a substantially better prognosis (among 282 patients, the rate of being bedridden, vegetative, or dead was 2% in alert patients vs. 13% in confused patients vs. 57% in obtunded or comatose patients, p<0.0001, Cochran-Armitage trend test). 83 The case revisited Although only symptomatic for 18 hours so far, your patient will soon probably fit the full clinical picture of AVS (continuous vertigo lasting more than a day, accompanied by nausea or vomiting, head-motion intolerance, nystagmus, and unstable gait). The most likely etiology is vestibular neuritis, but dangerous causes of AVS, such as cerebellar stroke, must be considered. The absence of vascular risk factors and age <50 reduce the risk of vertebrobasilar atherosclerosis, but vertebral artery dissection remains a concern, mitigated incompletely by the absence of craniocervical pain. There are no auditory symptoms to raise concerns about inner ear ischemia. The patient s general neurologic exam is normal, including no limb ataxia or dysmetria, compatible with a peripheral cause. He feels unsteady when standing but is able to sit with arms crossed unaided, also compatible with a peripheral lesion. Eye examination reveals direction-fixed, left-beating horizontal nystagmus worse in left gaze and no skew deviation on alternate cover testing, all compatible with your leading potential diagnosis of vestibular neuritis.

21 You avoid the Dix-Hallpike maneuver, recognizing its lack of diagnostic utility in AVS. However, you astutely test his vestibulo-ocular responses using the head impulse test and find normal responses during rapid head rotation in either direction. The normal head impulse result in the setting of AVS strongly favors a stroke, despite the lack of other symptoms, signs, or risk factors. You admit the patient to the neuro-critical care unit for close monitoring, pending neuroimaging and stroke consultation. MRI obtained the next morning reveals a large left posterior inferior cerebellar artery infarction (Figure 2). A complete search for cerebrovascular risk factors identifies a left vertebral artery dissection as the underlying cause of this stroke. The patient is observed in the hospital while antithrombotic agents are initiated, and is discharged uneventfully after the critical three-day window has passed during which complications related to swelling are most likely. He makes a complete neurologic recovery, his dissection heals, and at 2-year follow-up, he has suffered no new strokes. Limitations We found only a limited number of rigorously-designed studies that prospectively assessed AVS patients acutely, including only one that used superior reference standards both to rule in and rule out stroke. 6 Variability in diagnostic reference standards across studies is a limitation, although studies used in analysis of key oculomotor exam findings all used neuroimaging to diagnose or exclude stroke and met or exceeded an acceptable quality standard (Appendix 1). The dearth of studies with superior reference standards (especially to rule out stroke) means that diagnostic misclassification might still partially confound our results. The bedside tests investigated varied across studies, resulting in relatively small numbers of patients sampled for each test. The highest-quality studies were in higher-risk AVS subgroups,

22 leading to more imprecise estimates of test specificity, despite relatively robust estimates of sensitivity. Evidence to support the three-component battery (H.I.N.T.S.) is limited to two studies with only 184 patients. 6,10 Bedside oculomotor tests were performed by experienced subspecialist neuro-otologists, limiting the immediate generalizability of these findings to care settings lacking access to this clinical expertise. While current best evidence supports this approach and a strong recommendation for use (Table 2), the H.I.N.T.S. battery has not yet been fully validated as a clinical prediction rule, 99 and the generalizability of these findings to routine clinical practice (particularly with generalist rather than specialist examiners) remains to be confirmed. Studies reviewed were conducted predominantly in the emergency department or acute inpatient setting; findings might not generalize well to non-urgent, ambulatory care settings as a result of case mix differences in cause or illness severity. This is most likely to be a concern if patients are seen after the acute phase (>72 hours post symptom onset), when eye movement findings tend to evolve as a result of adaptive neural mechanisms. Conclusions Acute persistent dizziness is a common presentation to the emergency department, and potential etiologies range from benign to deadly. The main differential diagnosis is between vestibular neuritis and ischemic stroke. Although physical exam is more likely to be helpful than history once AVS is identified, red flags for stroke probably include multiple transient prodromal episodes over weeks or months; auditory symptoms; and headache, neck pain or recent trauma. Best evidence suggests that nearly two-thirds of stroke patients present without focal neurological signs that would be readily apparent to a non-neurologist (and one-third present without signs readily apparent to a neurologist). In such cases, a three-component bedside

23 oculomotor examination (H.I.N.T.S. to I.N.F.A.R.C.T.) can rule stroke in and out with high accuracy that probably exceeds that of urgent MRI with diffusion-weighted imaging. While expertise in these bedside techniques may currently be limited to specialists, our review suggests that clinical skill development in eye examination is probably warranted for generalists who routinely assess patients with acute dizziness. Training emergency department and other frontline physicians in the H.I.N.T.S. exam (or providing a device-based solution such as quantitative video-oculography 100 ) may be necessary if dangerous causes of AVS are to be promptly and accurately diagnosed. Negative cranial MRI with diffusion-weighted imaging obtained in the first hours after symptom onset should not be considered final in the presence of any dangerous H.I.N.T.S. or a high baseline prior probability for a vascular etiology of AVS. Future studies should seek to confirm these findings in completely unselected AVS populations (with and without auditory symptoms), including the utility of bedside tests when performed by emergency department physicians or other generalists.

24 Competing interests The authors report no conflicts of interest. Contributions Alexander A. Tarnutzer, Aaron L. Berkowitz and David E. Newman-Toker conceived of the study and conducted the systematic review of the literature. Karen A. Robinson executed the literature search. Alexander A. Tarnutzer abstracted and summarized the results. Alexander A. Tarnutzer, David Newman-Toker and Yu-Hsiang Hsieh performed the statistical analysis. Alexander A. Tarnutzer and David Newman-Toker drafted the article. All of the authors participated in the revisions of the draft with respect to content and form in preparation of the final version submitted for publication. Sources of funding This work was supported partly by a grant from the Agency for Healthcare Research and Quality (AHRQ HS ). Alexander A. Tarnutzer s effort in preparation of this article was supported by the Swiss National Science Foundation (PBZHP ).

25 Boxes Box 1 - Systematic Literature Search To identify observational studies reporting on the clinical features, diagnostic evaluation, and differential diagnosis of AVS, we conducted a search using PubMed and also examined the bibliographies from eligible articles. Details of the search strategy appear in Online Appendix 1.* We excluded articles that lacked original patient data contained no symptom data about dizziness provided no information about diagnostic accuracy for acute central or peripheral vestibulopathies did not evaluate patients in the acute stage of disease involved patients under age 18 years reported on fewer than five subjects. Our search identified 779 unique citations, of which 15 articles reporting details from 10 prospective or retrospective studies (there were no meta-analyses) met our inclusion criteria and whose diagnostic reference standard for ruling in or ruling out stroke was rated adequate (medium-quality) or superior (high-quality) (see Online Appendix 2 for definitions). Articles in Online Appendix 2 were selected for inclusion in the systematic review. Whenever possible and as appropriate, we contacted authors to gather patient-level data for abstraction. * Online Appendices available at

26 Box 2 - Differential Diagnosis of AVS based on Expert Opinion BENIGN* or LESS URGENT causes Common Causes (>1% of AVS) vestibular neuritis multiple sclerosis Uncommon (<1%) or Unknown Frequency viral labyrinthitis herpes zoster oticus (Ramsay Hunt) acute traumatic vestibulopathy med ototoxicity (e.g., aminoglycosides) acute disseminated encephalomyelitis CNS side effects (e.g., antiepileptics) prolonged Menière syndrome attack prolonged vestibular migraine attack episodic ataxia syndrome attack Presumed Possible Causes atypical infection (otosyphilis, lyme) degenerative cerebellar ataxia cerebello-pontine angle neoplasm (e.g., vestibular schwannoma, metastases) drug intoxication (e.g., alcohol, illicits) DANGEROUS* and MORE URGENT causes Common Causes (>1% of AVS) brainstem or cerebellar infarction brainstem or cerebellar hemorrhage Uncommon (<1%) or Unknown Frequency labyrinthine stroke bacterial labyrinthitis/mastoiditis Wernicke syndrome (B1 deficiency) Miller Fisher syndrome brainstem encephalitis (e.g., listeria, herpes simplex/zoster, paraneoplastic) Presumed Possible Causes cerebral infarction or hemorrhage subarachnoid hemorrhage/aneurysm severe anemia or hypoxia carbon monoxide toxicity electrolyte (e.g., hyponatremia) endocrine (e.g., hypothyroidism) decompression sickness CNS medication toxicity (e.g., lithium) AVS = acute vestibular syndrome; CNS = central nervous system * Any disease causing dizziness/vertigo can be considered a dangerous medical problem if the symptoms occur in dangerous circumstances (e.g., highway driving or free-rock climbing). Furthermore, the high vagal tone that accompanies some vestibular disorders can provoke bradyarrhythmias in susceptible individuals, including during the examination. 101 Nevertheless, although they may be quite disabling to patients during the acute illness phase, diseases classified here as Benign or Less Urgent Causes rarely produce severe, irreversible morbidity or mortality (unlike their Dangerous counterparts). The frequency of labyrinthine infarction is difficult to estimate given that the current reference standard test for confirming the diagnosis (i.e., autopsy with temporal bone histology) is rarely performed. Recent studies, however, suggest that patients with sudden deafness, with or without vertigo, are at increased risk of stroke, suggesting a possible vascular mechanism. 27,75 Presumed possible causes are conditions known to cause acute dizziness, but it remains unknown whether they can present with a clinically-complete or clinically-predominant AVS picture. AVS secondary to cerebrovascular disease is strongly associated with posterior fossa infarcts. 6 Transient dizziness or vertigo may be present in up to 10% of patients with right-hemispheric stroke 102 and isolated cases of insular or parietal infarction may present with vertigo as the lead symptom. 103,104 However, the precise duration of symptoms and extent of clinical pattern match with AVS in most of these cases remains unknown.

27 Tables Table 1: Common Central and Peripheral Causes of AVS based on Systematic Review* Diagnosis Study Norrving et al (n=24) Cnyrim et al (n=83) Kattah et al (n=101 total [62 isolated ]) Total Isolated AVS AVS n=208 n=169 Peripheral Causes Vestibular neuritis (100%) 82 (100%) Central Causes Posterior fossa cerebrovascular events [37] 104 (83%) Ischemic stroke [36] 99 (cerebellum or (79%) brainstem) Hemorrhage [1] 5 (cerebellum or (4%) brainstem) Multiple sclerosis [0] 14 (11%) Other central/equivocal [0] 8 (6%) 68 (78%) 66 (76%) 2 (2%) 12 (14%) 7 (8%) Abbreviations: AVS = acute vestibular syndrome; NR = not reported * Note that only studies with an adequate (medium-quality) or superior (high-quality) reference standard to rule in or rule out stroke were considered for this table (see Online Appendix 1 for definitions). Only one study (Kattah, 2009) offered a superior reference standard for ruling out stroke. Note that patients in the three studies reported here were not completely unselected. As a result, little can be said from these studies alone about the relative prevalence of peripheral versus central causes in unselected AVS presentations; proportions of peripheral and central causes are therefore reported separately. Two of the studies focused on higher-risk-for-stroke AVS presentations (Norrving, 1995 enrolled only patients over age 50; Kattah, 2009 required all patients to have at least one vascular risk factor). The third (Cnyrim, 2008) employed a case-control design that included all pseudoneuritis presentations but only a sample of vestibular neuritis patients for comparison. Total AVS includes all AVS patients for the available studies. Isolated AVS includes only those patients without obvious focal neurologic signs. Numbers of isolated AVS presentations are reported in brackets for the one study reporting AVS patients with additional neurologic signs as part of the total. 6

28 No patients with labyrinthitis (i.e., peripheral AVS with auditory symptoms) were reported in these three studies. However, two studies expressly excluded patients with auditory symptoms and the third (Kattah, 2009) excluded those with a history of recurrent auditory symptoms. Authors describe 6 confirmed infarctions at the time of presentation and 1 patient with an occluded vertebral artery who returned with a stroke four months after his initial presentation. The 7 other cases reported in Cnyrim, 2008 included 2 cases of brain metastases, 1 case of Arnold-Chiari malformation with a iatrogenic lesion after surgical decompression, and 4 cases with nonspecific white matter lesions (personal communication with the lead author). The other case reported in Kattah, 2009 suffered from acute intoxication with carbamazepine.

29 Table 2. Clinical findings as predictors of stroke versus vestibular neuritis in patients presenting with AVS Clinical Finding Predictor Magnitude Review Evidence Note Type Level* Chief Sx Dizziness/vertigo gradual onset neuritis large (?) critical 4 Dizziness is gradual onset (hours to days) in very few stroke cases; 38,39 little is known about onset characteristics in neuritis, but some note gradual onset 5 Dizziness/vertigo abrupt onset Stroke small (?) critical 4 Dizziness is abrupt onset (seconds to minutes) in a large proportion of stroke cases; 39 little is known about onset characteristics in neuritis, but some awake with new Sx Multiple prodromal dizziness episodes Stroke large critical 3b Multiple prodromal episodes in the 6 months prior to onset of AVS, particularly in weeks prior to AVS, have been reported only as TIAs preceding stroke 38,39 Single prodromal dizziness episode No - critical 3b Single episode of prodromal dizziness in the 48 hours preceding the onset of AVS is reported in 1 in 4 cases, regardless of cause (vestibular neuritis vs. stroke) 26,38-41 Dizziness type (vertigo vs. others) No - critical 2b Dizziness type is imprecisely reported by patients; 25 disease-based studies indicate overlap in type reported for different diseases, 23,25 including posterior fossa stroke 13,42 Dizziness/vertigo total duration >1d No - critical 4 Dizziness duration varies across etiologic causes; continuous duration >24 hours for new, acute Sx eliminates most transient causes (e.g., BPPV) except migraine 43,44 Dizziness/vertigo triggers No - critical 5 Triggers predict causes in patients with brief episodes, but provocative factors (e.g., worse with head motion) do not distinguish central vs. peripheral causes of AVS 18,45 Associated Sx Painless (no neck pain or headache) Neuritis small NLR both 2a Head/neck pain present in a minority of AVS whether due to stroke or neuritis; frequency in peripheral AVS lower; pain absent in 1 of 4 vertebral dissections 46 Craniocervical pain (e.g., headache, neck pain) Stroke moderate PLR both 2a Craniocervical pain is reported frequently in cerebellar infarctions and vertebral dissections; 6,8,21 frequency in peripheral AVS is lower; characteristics of pain as sudden, severe, 48,51,52 or sustained (>72 hours) 50 may increase likelihood stroke Neurologic Sx including diplopia Stroke large (?) critical 4 Double vision, slurred speech, trouble swallowing, hoarseness and other posterior circulation Sx suggest stroke, though rare exceptions probably do occur 53 Non-proportional Sx (gait or nausea) Stroke unknown critical 4 Disproportionate gait/postural imbalance or nausea/vomiting in excess of the symptomatic experience of dizziness/vertigo is described with central causes 6,9,13,21 Auditory Sx (e.g., hearing loss) stroke (?) unknown both 4 Several case series have found auditory Sx (particularly hearing loss) as a frequent Sx in AICA infarctions; 22,27 frequency of benign causes ( labyrinthitis ) unknown

30 Risk Factors Age < 50 Neuritis small (?) both 3b 25% of AVS stroke patients are < 50; 6 younger AVS patients often have vertebral artery dissection as a cause; 6 stroke misdiagnosis is more common in those < 50 14,54 Age > 50 Stroke large both 3b Age is a known risk factor for stroke and has been used to enrich AVS research populations with stroke; 8 isolated AVS > 50 is probably still more often neuritis 8 Other vascular risk factors Stroke moderate both 3b Patients with at least one vascular risk factor appear to have an increased risk of stroke 6 (estimated prevalence in isolated AVS ~30-50% versus ~10-20% without) Trauma history Stroke moderate critical 4 Recent head/neck trauma is a risk factor for vertebral artery dissection (PLR 3.8); 55 in AVS, trauma (even minor) should spark concern for dissection, even without pain No trauma No - critical 4 > 50% of symptomatic vertebral artery dissections occur without an identifiable history of trauma, most in patients without known collagen-vascular disorders 46,55 Comorbid medical illness or surgery Physical Exam General neurologic signs absent All neurologic signs absent Dangerous HINTS ( I.N.F.A.R.C.T. ) absent on exam Normal head impulse test Any neurologic signs present Other vestibularoculomotor signs Variable - critical 5 Patients with a history of malnutrition (Wernicke syndrome), immunocompromise (encephalitis, herpes zoster oticus), ear surgery (mastoiditis), or drug exposure (aminoglycoside ototoxicity) are presumably more likely to have a related cause Neuritis Neuritis Neuritis Stroke small NLR systematic 2a Not counting oculomotor findings or severe gait or truncal ataxia, absence of neurologic signs, excludes only about 1 of 5 strokes causing AVS 6 moderate systematic 2a Not counting HINTS exam, absence of all neurologic Sx or signs, including severe NLR gait or truncal ataxia, excludes approximately 2 of 3 strokes causing AVS 6 very large systematic 1b The benign triad of abnormal unilateral head impulse test (away from nystagmus fast NLR 0.02 (see text) GRADE phase), direction-fixed dominantly horizontal nystagmus, and absent vertical ocular strong+ misalignment (skew) virtually excludes stroke in AVS without auditory Sx 6,10,21 large PLR 18 systematic (Table 3) 1a GRADE strong+ In a patient with AVS, a normal horizontal head impulse test in both directions is a strong predictor of stroke as a cause, even if all other findings suggest neuritis 6,10,11 Stroke very large (?) systematic 2a Any new neurologic signs, including severe gait or truncal ataxia, are probably strong predictors of central causes, usually stroke, and generally exclude neuritis 53 Unknown - both 4/5 Characteristics of spontaneous nystagmus (vector, fixation suppression, pattern, appearance with provocation) and abnormal visual smooth pursuit are understudied

31 Abbreviations: AICA anterior inferior cerebellar artery; AVS acute vestibular syndrome; BPPV benign paroxysmal positional vertigo; HINTS head impulse, nystagmus, test of skew; INFARCT impulse normal, fast-phase alternating, refixation on cover test; NLR negative likelihood ratio; PLR positive likelihood ratio; Sx symptoms; TIA transient ischemic attack * Level of evidence was defined by a single rater (DNT) according to the Oxford Centre for Evidence-based Medicine Levels of Evidence for Diagnosis. 56 The highest level of evidence is 1a (based on systematic review of level 1 studies with homogeneity) and the lowest level of evidence is 5 (expert opinion without explicit critical appraisal or based on physiology or first principles ). Evidence failing to meet strict inclusion criteria for the systematic review was considered part of the critical review; studies that achieved a higher evidence rating yet were not included in the systematic review were generally studies where the AVS population was incompletely defined (e.g., AVS patients included, but along with other non-avs vertigo or dizziness patients, with predictors not defined separately for the two groups). In addition, a GRADE strength recommendation for adoption of evidence related to the clinical predictors is provided for exam elements that achieved a Level 1 rating (see footnote below for details). Refers to proportionality of vestibular (dizziness/vertigo), neurovegetative (nausea/vomiting), and gait/postural symptoms. Non-proportional symptoms would be, for example, severe gait impairment with a mild subjective sense of dizziness or vertigo. Auditory symptoms have not been adequately studied as a specific predictor of stroke versus non-stroke causes of AVS. The vascular supply to the inner ear derives from the posterior cerebral circulation, and combined audio-vestibular presentations appear to be common among those with cerebrovascular disease. 22 It is known that some uncommon, non-vascular peripheral vestibular disorders, such as herpes zoster oticus (Ramsay Hunt syndrome), also produce combined audio-vestibular presentations (i.e., AVS with hearing loss or tinnitus) that can be abrupt in onset. 57 Unfortunately, patients with auditory symptoms have generally been excluded from major studies of AVS or vestibular neuritis. 58 This includes even the highest quality study 6 from our systematic review, which excluded those with a prior history of auditory symptoms (though not those with new, acute symptoms, which were identified in 2 of 69 strokes and 0 of 25 neuritis cases). 6 Future studies of AVS should prospectively enroll patients with and without auditory symptoms. Vascular risk factors included smoking, hypertension, diabetes, hyperlipidemia, atrial fibrillation, eclampsia, hypercoagulable state, recent cervical trauma, or prior stroke or myocardial infarction. 6 GRADE level of evidence for recommending use of the HINTS approach to AVS diagnosis or normal head impulse test finding alone as a stroke predictor is Level B (moderate). This quality assessment is based on homogeneous and consistent results from multiple research groups for individual oculomotor findings (including the normal head impulse test) and one high-quality cross-sectional study in an appropriate, unbiased population with diagnostic uncertainty and direct comparison of HINTS test battery results with an appropriate reference standard. 59 The level of evidence is downgraded from high quality for indirectness (diagnostic accuracy study without health outcomes measurement [-1], sparse specificity data [-1], and uncertain generalizability beyond subspecialists [-1]); these effects are mitigated by a very strong association/effect size (+2). 59,60 In aggregate, therefore, the quality of evidence for HINTS and the head impulse according to GRADE is Level B (moderate). Considering the balance of benefits, risks, and alternatives for stroke diagnosis (i.e., routine use of insensitive CT or more expensive, less readily available, and less sensitive MRI), the GRADE recommendation for adequately trained providers to use the HINTS approach to stroke diagnosis (or a normal head impulse to predict stroke) in patients with AVS is Strongly For (+). 61

32 Table 3: Pooled test properties of key bedside predictors in AVS from studies with an adequate (medium) or superior (high) reference standard for final diagnosis n[studies] n[subjects] Sensitivity Specificity NLR PLR total, peripheral, (95% CI*) (95% CI*) (95% CI*) (95% CI*) central Peripheral Central Bedside diagnostic tests Normal h-hit (all central) 6,9,10,29 4, 2, PICA stroke vs. peripheral 6,9,29 3, 2, AICA stroke vs. peripheral 6,9,29 3, 2, ( ) ( ) 0.99 ( ) ( ) ( ) ( ) 0.01 ( ) NC 0.40 ( ) NC Direction-changing N 6,8-10,27,31 6, 3, Skew deviation 6,10 2, 2, ( ) 0.30 ( ) 0.92 ( ) 0.98 ( ) 0.68 ( ) 0.71 ( ) 4.51 ( ) ( ) AVS = acute vestibular syndrome; CI = confidence interval; h-hit = horizontal head impulse test; NC = not calculated; NLR = negative likelihood ratio; PLR = positive likelihood ratio * Calculated by the method described by Simel. 79 Upper-bound values of sensitivity or specificity that were calculated to be greater than 100% using this method were assigned a value of 1.0. Includes all beside tests where at least two studies provided data for either peripheral or central causes of AVS. Bedside examination of smooth pursuit eye movements, optokinetic nystagmus, saccades, positional nystagmus, head-shaking nystagmus, vibration-induced nystagmus, and hearing loss are not shown because of insufficient (<2) study numbers to offer pooled results. Includes 27 patients with central causes other than ischemic stroke (14 demyelination; 5 brainstem or cerebellar hemorrhages; 8 other). The 95% CI of the specificity and the PLR could not be calculated for the prospective subgroup analysis since specificity for the h-hit was 1.0 (i.e., there were no false positive h-hit results in this pooled sample), resulting in a division by zero for the calculation of PLR and 95% CI of PLR. Beside-clinical and electro-oculography-based data describing nystagmus were pooled for these estimates since not all studies reported the two methods of identifying the same clinical findings as independent results.

33 Figure legends Figure 1 Initial differential diagnosis of the acutely dizzy patient based on the duration of symptoms. The most common causes of both transient and continuous dizziness are shown. Abbreviations: AED = antiepileptic drug; AVS = acute vestibular syndrome; BPPV = benign paroxysmal positional vertigo; MS = multiple sclerosis; TIA = transient ischemic attack. Figure 2 This patient with AVS presented with a pseudoneuritis clinical picture of left-beating horizontal nystagmus, absent skew, but normal horizontal head impulse test result. The patient was found to have a large left-sided stroke encompassing the territory of the posterior inferior cerebellar artery on T2-weighted (left) and diffusion-weighted imaging (right) MRI. Images courtesy of Jorge C. Kattah, MD

34 REFERENCES [1] Kroenke K, Mangelsdorff AD. Common symptoms in ambulatory care: incidence, evaluation, therapy, and outcome. Am J Med. 1989;86: [2] Newman-Toker DE, Hsieh YH, Camargo CA, Jr., et al. Spectrum of dizziness visits to US emergency departments: cross-sectional analysis from a nationally representative sample. Mayo Clin Proc. 2008;83: [3] Kruschinski C, Hummers-Pradier E, Newman-Toker D, et al. Diagnosing dizziness in the emergency and primary care settings [letter to the editor]. Mayo Clin Proc. 2008;83: [4] Neuhauser HK, von Brevern M, Radtke A, et al. Epidemiology of vestibular vertigo: a neurotologic survey of the general population. Neurology. 2005;65: [5] Hotson JR, Baloh RW. Acute vestibular syndrome. N Engl J Med. 1998;339: [6] Kattah JC, Talkad AV, Wang DZ, et al. HINTS to diagnose stroke in the acute vestibular syndrome: three-step bedside oculomotor examination more sensitive than early MRI diffusionweighted imaging. Stroke. 2009;40: [7] Newman-Toker DE, Camargo CA, Jr. 'Cardiogenic vertigo'--true vertigo as the presenting manifestation of primary cardiac disease. Nat Clin Pract Neurol. 2006;2: [8] Norrving B, Magnusson M, Holtas S. Isolated acute vertigo in the elderly; vestibular or vascular disease? Acta Neurol Scand. 1995;91:43-8. [9] Lee H, Sohn SI, Cho YW, et al. Cerebellar infarction presenting isolated vertigo: frequency and vascular topographical patterns. Neurology. 2006;67: [10] Cnyrim CD, Newman-Toker D, Karch C, et al. Bedside differentiation of vestibular neuritis from central "vestibular pseudoneuritis". J Neurol Neurosurg Psychiatry. 2008;79: [11] Newman-Toker DE, Kattah JC, Alvernia JE, et al. Normal head impulse test differentiates acute cerebellar strokes from vestibular neuritis. Neurology. 2008;70: [12] Duncan GW, Parker SW, Fisher CM. Acute cerebellar infarction in the PICA territory. Arch Neurol. 1975;32: [13] Kerber KA, Brown DL, Lisabeth LD, et al. Stroke among patients with dizziness, vertigo, and imbalance in the emergency department: a population-based study. Stroke. 2006;37: [14] Savitz SI, Caplan LR, Edlow JA. Pitfalls in the diagnosis of cerebellar infarction. Acad Emerg Med. 2007;14:63-8. [15] Newman-Toker DE, Robinson KA, Edlow JA. Frontline misdiagnosis of cerebrovascular events in the era of modern neuroimaging: a systematic review [abstract]. Abstracts of the 133rd Annual Meeting of the American Neurological Association. Ann Neurol. 2008;64(Suppl 12):S17-S8. [16] Newman-Toker DE, Camargo CA, Jr., Hsieh YH, et al. Disconnect between charted vestibular diagnoses and emergency department management decisions: a cross-sectional analysis from a nationally representative sample. Acad Emerg Med. 2009;16: [17] Kim AS, Fullerton HJ, Johnston SC. Risk of Vascular Events in Emergency Department Patients Discharged Home With Diagnosis of Dizziness or Vertigo. Ann Emerg Med [18] Stanton VA, Hsieh YH, Camargo CA, Jr., et al. Overreliance on symptom quality in diagnosing dizziness: results of a multicenter survey of emergency physicians. Mayo Clin Proc. 2007;82:

35 [19] Eagles D, Stiell IG, Clement CM, et al. International survey of emergency physicians' priorities for clinical decision rules. Acad Emerg Med. 2008;15: [20] Kerber KA, Fendrick AM. The evidence base for the evaluation and management of dizziness. J Eval Clin Pract. 2010;16: [21] Edlow JA, Newman-Toker DE, Savitz SI. Diagnosis and initial management of cerebellar infarction. Lancet Neurol. 2008;7: [22] Lee H. Neuro-otological aspects of cerebellar stroke syndrome. J Clin Neurol. 2009;5: [23] Newman-Toker DE. Diagnosing Dizziness in the Emergency Department Why "What do you mean by 'dizzy'?" Should Not Be the First Question You Ask [Doctoral Dissertation, Clinical Investigation, Bloomberg School of Public Health]. Baltimore, MD: The Johns Hopkins University; In: ProQuest Digital Dissertations [database on Internet, publication number: AAT Available at: Accessed November 19, [24] Neuhauser HK, Lempert T. Vertigo: epidemiologic aspects. Semin Neurol. 2009;29: [25] Newman-Toker DE, Cannon LM, Stofferahn ME, et al. Imprecision in patient reports of dizziness symptom quality: a cross-sectional study conducted in an acute care setting. Mayo Clin Proc. 2007;82: [26] Lee H, Baloh RW. Sudden deafness in vertebrobasilar ischemia: clinical features, vascular topographical patterns and long-term outcome. J Neurol Sci. 2005;228: [27] Lee H, Kim JS, Chung EJ, et al. Infarction in the territory of anterior inferior cerebellar artery: spectrum of audiovestibular loss. Stroke. 2009;40: [28] Lee H, Sohn SI, Jung DK, et al. Sudden deafness and anterior inferior cerebellar artery infarction. Stroke. 2002;33: [29] Moon IS, Kim JS, Choi KD, et al. Isolated nodular infarction. Stroke. 2009;40: [30] Kim GW, Heo JH. Vertigo of cerebrovascular origin proven by CT scan or MRI: pitfalls in clinical differentiation from vertigo of aural origin. Yonsei Med J. 1996;37: [31] Rubenstein RL, Norman DM, Schindler RA, et al. Cerebellar infarction--a presentation of vertigo. Laryngoscope. 1980;90: [32] Chen CH, Young YH. Vestibular evoked myogenic potentials in brainstem stroke. Laryngoscope. 2003;113: [33] Choi KD, Oh SY, Kim HJ, et al. The vestibulo-ocular reflexes during head impulse in Wernicke's encephalopathy [research letter]. J Neurol Neurosurg Psychiatry. 2007;78: [34] Baloh RW. Clinical practice. Vestibular neuritis. N Engl J Med. 2003;348: [35] Lloyd-Jones D, Adams RJ, Brown TM, et al. Heart Disease and Stroke Statistics Update. A Report From the American Heart Association. Circulation. 2009;121:e46-e215. [36] Baptista MV, van Melle G, Bogousslavsky J. Prediction of in-hospital mortality after first-ever stroke: the Lausanne Stroke Registry. J Neurol Sci. 1999;166: [37] Fisher CM. Vertigo in cerebrovascular disease. Arch Otolaryngol. 1967;85: [38] Gomez CR, Cruz-Flores S, Malkoff MD, et al. Isolated vertigo as a manifestation of vertebrobasilar ischemia. Neurology. 1996;47:94-7. [39] Grad A, Baloh RW. Vertigo of vascular origin. Clinical and electronystagmographic features in 84 cases. Arch Neurol. 1989;46:281-4.

36 [40] Lee H, Kim BK, Park HJ, et al. Prodromal dizziness in vestibular neuritis: frequency and clinical implication. J Neurol Neurosurg Psychiatry. 2009;80: [41] Silvoniemi P. Vestibular neuronitis. An otoneurological evaluation. Acta Otolaryngol Suppl. 1988;453:1-72. [42] Kase CS, Norrving B, Levine SR, et al. Cerebellar infarction. Clinical and anatomic observations in 66 cases. Stroke. 1993;24: [43] Neuhauser H, Leopold M, von Brevern M, et al. The interrelations of migraine, vertigo, and migrainous vertigo. Neurology. 2001;56: [44] von Brevern M, Zeise D, Neuhauser H, et al. Acute migrainous vertigo: clinical and oculographic findings. Brain. 2005;128: [45] Newman-Toker DE, Stanton VA, Hsieh YH, et al. Frontline providers harbor misconceptions about the bedside evaluation of dizzy patients [research letter]. Acta Otolaryngol. 2008;128: [46] Gottesman RF, Sharma P, Robinson KA, et al. Imaging Characteristics of Vertebral Artery Dissection: A Systematic Review. 135th Annual Meeting of the American Neurological Association; 2010; San Francisco, USA; [47] de Bray JM, Penisson-Besnier I, Dubas F, et al. Extracranial and intracranial vertebrobasilar dissections: diagnosis and prognosis. J Neurol Neurosurg Psychiatry. 1997;63: [48] Saeed AB, Shuaib A, Al-Sulaiti G, et al. Vertebral artery dissection: warning symptoms, clinical features and prognosis in 26 patients. Can J Neurol Sci. 2000;27: [49] Silbert PL, Mokri B, Schievink WI. Headache and neck pain in spontaneous internal carotid and vertebral artery dissections. Neurology. 1995;45: [50] Sturzenegger M. Headache and neck pain: the warning symptoms of vertebral artery dissection. Headache. 1994;34: [51] Sturzenegger M, Mattle HP, Rivoir A, et al. Ultrasound findings in spontaneous extracranial vertebral artery dissection. Stroke. 1993;24: [52] Bartels E. Dissection of the extracranial vertebral artery: clinical findings and early noninvasive diagnosis in 24 patients. J Neuroimaging. 2006;16: [53] Schlaeger R, Naegelin Y, Welge-Lussen A, et al. Acute Vertigo with Double Vision - Brainstem Stroke or Stroke Mimic? Cerebrovasc Dis. 2010;30: [54] Kuruvilla A, Bhattacharya P, Rajamani K, et al. Factors Associated With Misdiagnosis of Acute Stroke in Young Adults. J Stroke Cerebrovasc Dis [55] Rubinstein SM, Peerdeman SM, van Tulder MW, et al. A systematic review of the risk factors for cervical artery dissection. Stroke. 2005;36: [56] Levels of Evidence and Grades of Recommendation. Oxford Centre for Evidence Based Medicine [cited November 16, 2010.]; Available from: [57] Wayman DM, Pham HN, Byl FM, et al. Audiological manifestations of Ramsay Hunt syndrome. J Laryngol Otol. 1990;104: [58] Newman-Toker DE, Reich SG. Wrong-Way nystagmus in the AICA syndrome [letter to the editor]. The Laryngoscope. 2008;118: [59] Schunemann HJ, Oxman AD, Brozek J, et al. Grading quality of evidence and strength of recommendations for diagnostic tests and strategies. BMJ (Clinical research ed. 2008;336:

37 [60] Atkins D, Best D, Briss PA, et al. Grading quality of evidence and strength of recommendations. BMJ (Clinical research ed. 2004;328:1490. [61] Guyatt GH, Oxman AD, Kunz R, et al. Incorporating considerations of resources use into grading recommendations. BMJ. 2008;336: [62] Drachman DA, Hart CW. An approach to the dizzy patient. Neurology. 1972;22: [63] Delaney KA. Bedside diagnosis of vertigo: value of the history and neurological examination. Acad Emerg Med. 2003;10: [64] Drachman DA. A 69-year-old man with chronic dizziness. JAMA. 1998;280: [65] Bisdorff A, Von Brevern M, Lempert T, et al. Classification of vestibular symptoms: towards an international classification of vestibular disorders. J Vestib Res. 2009;19:1-13. [66] Newman-Toker DE. Charted records of dizzy patients suggest emergency physicians emphasize symptom quality in diagnostic assessment [research letter]. Ann Emerg Med. 2007;50: [67] Matsuo T, Sekitani T. Vestibular neuronitis: neurotological findings and progress. ORL J Otorhinolaryngol Relat Spec. 1985;47: [68] Fisher CM, Picard EH, Polak A, et al. Acute Hypertensive Cerebellar Hemorrhage: Diagnosis and Surgical Treatment. J Nerv Ment Dis. 1965;140: [69] Rothwell PM, Buchan A, Johnston SC. Recent advances in management of transient ischaemic attacks and minor ischaemic strokes. Lancet Neurol. 2006;5: [70] Kubo T, Sakata Y, Sakai S, et al. Clinical observations in the acute phase of cerebellar hemorrhage and infarction. Acta Otolaryngol Suppl. 1988;447:81-7. [71] Rosenberg ML, Gizzi M. Neuro-otologic history. Otolaryngol Clin North Am. 2000;33: [72] Derebery MJ. The diagnosis and treatment of dizziness. Med Clin North Am. 1999;83:163-77, x. [73] Drachman DA, Diamond ER, Hart CW. Posturally-evoked vomiting; Association with posterior fossa lesions. Ann Otol Rhinol Laryngol. 1977;86: [74] Fisher CM. Vomiting out of proportion to dizziness in ischemic brainstem strokes. Neurology. 1996;46:267. [75] Lin HC, Chao PZ, Lee HC. Sudden sensorineural hearing loss increases the risk of stroke: a 5-year follow-up study. Stroke. 2008;39: [76] Lempert T, Neuhauser H, Daroff RB. Vertigo as a symptom of migraine. Ann N Y Acad Sci. 2009;1164: [77] Kelman L. Pain characteristics of the acute migraine attack. Headache. 2006;46: [78] Dieterich M, Brandt T. Episodic vertigo related to migraine (90 cases): vestibular migraine? J Neurol. 1999;246: [79] Simel DL, Samsa GP, Matchar DB. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. J Clin Epidemiol. 1991;44: [80] Mandelzweig L, Goldbourt U, Boyko V, et al. Perceptual, social, and behavioral factors associated with delays in seeking medical care in patients with symptoms of acute stroke. Stroke. 2006;37: [81] Martin-Schild S, Albright KC, Tanksley J, et al. Zero on the NIHSS Does Not Equal the Absence of Stroke. Annals of emergency medicine [82] Baloh RW. Dizziness: neurological emergencies. Neurol Clin. 1998;16:

38 [83] Tohgi H, Takahashi S, Chiba K, et al. Cerebellar infarction. Clinical and neuroimaging analysis in 293 patients. The Tohoku Cerebellar Infarction Study Group. Stroke. 1993;24: [84] Halmagyi GM, Curthoys IS. A clinical sign of canal paresis. Arch Neurol. 1988;45: [85] Leigh RJ, Zee DS. The Neurology of Eye Movements, 4th Edition. New York: Oxford University Press [86] Brodsky MC, Donahue SP, Vaphiades M, et al. Skew deviation revisited. Surv Ophthalmol. 2006;51: [87] Newman-Toker DE, Sharma P, Chowdhury M, et al. Penlight-cover test: a new bedside method to unmask nystagmus. J Neurol Neurosurg Psychiatry. 2009;80: [88] Zee DS, Fletcher WA. Bedside examination. In: Baloh RW, Halmagyi GM, eds. Disorders of the vestibular system. New York: Oxford University Press 1996: [89] Moulin T, Sablot D, Vidry E, et al. Impact of emergency room neurologists on patient management and outcome. Eur Neurol. 2003;50: [90] Chalela JA, Kidwell CS, Nentwich LM, et al. Magnetic resonance imaging and computed tomography in emergency assessment of patients with suspected acute stroke: a prospective comparison. Lancet. 2007;369: [91] Simmons Z, Biller J, Adams HP, Jr., et al. Cerebellar infarction: comparison of computed tomography and magnetic resonance imaging. Ann Neurol. 1986;19: [92] Wasay M, Dubey N, Bakshi R. Dizziness and yield of emergency head CT scan: is it cost effective? [research letter]. Emerg Med J. 2005;22:312. [93] Kerber KA. Vertigo and dizziness in the emergency department. Emergency medicine clinics of North America. 2009;27:39-50, viii. [94] Oppenheim C, Stanescu R, Dormont D, et al. False-negative diffusion-weighted MR findings in acute ischemic stroke. AJNR Am J Neuroradiol. 2000;21: [95] Marx JJ, Thoemke F, Mika-Gruettner A, et al. [Diffusion-weighted MRT in vertebrobasilar ischemia. Application, sensitivity, and prognostic value] [German]. Der Nervenarzt. 2004;75: [96] Brandt T, Zwergal A, Strupp M. Medical treatment of vestibular disorders. Expert Opin Pharmacother. 2009;10: [97] Hornig CR, Rust DS, Busse O, et al. Space-occupying cerebellar infarction. Clinical course and prognosis. Stroke. 1994;25: [98] Heros RC. Surgical treatment of cerebellar infarction. Stroke. 1992;23: [99] Laupacis A, Sekar N, Stiell IG. Clinical prediction rules. A review and suggested modifications of methodological standards. JAMA. 1997;277: [100] MacDougall HG, Weber KP, McGarvie LA, et al. The video head impulse test: diagnostic accuracy in peripheral vestibulopathy. Neurology. 2009;73: [101] Ullman E, Edlow JA. Complete heart block complicating the head impulse test. Archives of neurology. 2010;67: [102] Anagnostou E, Spengos K, Vassilopoulou S, et al. Incidence of rotational vertigo in supratentorial stroke: A prospective analysis of 112 consecutive patients. J Neurol Sci. 2010;290:33-6. [103] Naganuma M, Inatomi Y, Yonehara T, et al. Rotational vertigo associated with parietal cortical infarction. J Neurol Sci. 2006;246:

39 [104] Brandt T, Botzel K, Yousry T, et al. Rotational vertigo in embolic stroke of the vestibular and auditory cortices. Neurology. 1995;45:42-4.

40 New Acute Dizziness Transient Continuous (acute vestibular syndrome) Benign paroxysmal positional vertigo Menière / Migraine Transient ischemic attack Other (Arrhthmia, vasovagal syncope, etc ) Acute peripheral vestibulopathy Stroke Other (intoxication with antiepileptic drugs, multiple sclerosis, Wernicke, etc )

41