Risk of acute brain lesions in dizzy patients presenting to the emergency room: who needs imaging and who does not?

Dizziness is one of the most common presenting complaints in emergency departments (EDs) [17, 30]. Diagnosing patients with dizziness is challenging and ED physicians often request neurological consultation and brain imaging studies to differentiate non-vestibular medical causes (e.g., orthostatic dizziness) from peripheral (e.g., benign paroxysmal positional vertigo) or central vestibular disorders (e.g., brainstem/cerebellar stroke) [11, 27]. The likelihood of detecting an acute brain lesion (ABL), for instance a stroke, inflammatory lesion or tumor, varies greatly depending on the clinical preselection process of patients and the modality and timing of brain imaging studies. Thus, numbers range between 3–4% in unselected cohorts of dizzy patients in the ED [15, 30], 27% in patients with an acute transient vestibular syndrome [3], and up to 75% in clinically preselected cohorts with an acute (persistent) vestibular syndrome and a high-risk vascular profile [14]. Notwithstanding, a stroke is still missed in about 35% of patients presenting with acute dizziness to the ED [15].

Brain imaging studies can help to differentiate peripheral vestibular and central causes of dizziness [13]. However, the sensitivity for detecting an ABL, particularly ischemic stroke, depends on the imaging modality and its timing [33, 34]. Computed tomography (CT) is usually performed in the emergency setting, due to its around-the-clock and on-site availability in most of the hospitals. However, the diagnostic yield of CT brain imaging in the evaluation of non-preselected dizzy patients in the ED is low (~ 2%) [21]. One reason is the CT’s low sensitivity (~ 10%) for acute ischemic strokes [2], especially if they occur in the posterior cranial fossa [33]. In line with that, more CT imaging in the ED does not lead to an improvement in stroke diagnosis [20]. A negative CT result may even falsely reassure the ED physician of a peripheral cause instead of the actual central dizziness [10]. Although magnetic resonance imaging (MRI) has a higher sensitivity for acute ischemic strokes than CT [2], within the first 48 h diffusion-weighted MRI still can be false negative in about 50% of small ischemic strokes in the posterior fossa [34]. Furthermore, MRI is not as available (and feasible) as CT for dizzy patients in the ED and conducting MRIs in all dizzy patients presenting to the ED would unnecessarily burden many patients as well as the health-care system, as it exceeds the acceptable cost limits of most national economies [28, 33].

Hence, there is a clear need of a clinical preselection to identify those patients with a high pre-test probability for an ABL on imaging. However, the sensitivity and specificity of clinical tests that allow differentiation between peripheral and central causes of dizziness, such as the head impulse test [5, 31] or the HINTS triad (head impulse, nystagmus, test of skew) [14, 16], largely depend on the experience of the examiner [18]. On the other hand, ‘rater-independent’ risk scores like the ABCD²-score [25] or single risk factors such as high age [26] are less sensitive/specific [32] and they can only provide a rough estimation on the overall probability of a cerebrovascular cause, which may not be helpful in the individual case.

To specify, ED physicians and neurologists, who may not be neuro-otology experts, are confronted on a daily basis with the following questions [6]: Does this dizzy patient needs brain imaging or not? Is a CT sufficient, or is an MRI necessary? Is an immediate imaging study required? How do we deal with a negative result?

We aimed to develop a practical guide based on empirical evidence. By reviewing medical records of more than 600 patients presenting with dizziness, vertigo or imbalance to our university hospitals’ ED, we analyzed the frequency of brain imaging studies and the occurrence of ABLs. Taking established diagnostic algorithms for dizzy ED patients into account [8, 37], we assessed the association of ABLs with specific clinical signs and ‘dizziness syndromes’. Based on the results, we propose a clinical pathway that stratifies dizzy patients into subgroups of those who (i) do not need any brain imaging, (ii) do not need early CT but delayed MRI or (iii) require urgent CT and additional, delayed MRI when necessary.

We analyzed the ‘real-world’ frequency and diagnostic yield of brain imaging studies (CT, MRI) in patients presenting with dizziness, vertigo or imbalance to the ED, excluding those with a clear medical cause or uncomplicated BPPV. Early CT imaging was conducted in over 80% of the patients and almost every second patient later received an MRI (median 4 days). Dizzy patients who did not receive brain imaging were mostly female, typically reported transient/triggerable symptoms and seldom exhibited central oculomotor signs or other focal abnormalities on clinical examination. This most probably reflects a preexisting selection strategy in our ED to especially prevent young women with transient symptoms and normal neurological examination from unnecessary CT imaging.

Every fourth patient (24%) revealed an ABL (usually ischemic stroke) on adequately sensitive brain imaging. This could be taken as an argument to perform brain imaging in every dizzy patient presenting to the ED. However, in light of the high economic burden and overexposure of patients to imaging studies, we should make use of clinical markers that allow to preselect those dizzy patients who have a high pre-test probability for a central lesion.

In our cohort of dizzy ED patients, we found two clinical markers to be independent predictors for an ABL: any central oculomotor sign or a focal abnormality detected on neurological examination. In contrast, vascular risk factors such as arterial hypertension or an ABCD²-score ≥ 4 were only associated with ABLs in the univariate, but not in the multivariate logistic regression analysis. These results are in line with the previously proposed markers for central causes of acute dizziness: Navi and colleagues identified ‘focal examination abnormality’, ‘higher age (≥ 60 years)’ and ‘imbalance as presenting complaint’ as independent predictors [26]. Kerber et al. found the ‘continuous ABCD² score’, a ‘central pattern on the oculomotor assessment’, ‘any other CNS feature’ on clinical examination and ‘prior stroke’ to be related with stroke in patients with ‘new and continuous dizziness’ [16]. Taken together, brain imaging is strongly suggested in dizzy patients who exhibit any central oculomotor sign or focal abnormality on neurological examination. In contrast, vascular risk factors such as higher age or an ABCD²-score ≥ 4 are not consistently found to be reliable and independent predictors for ABLs in dizzy patients and should therefore not solely trigger imaging studies.

But how shall we manage dizzy patients in the ED who do not reveal any central oculomotor sign or focal abnormality? Conducting brain imaging in those with at least one vascular risk factor would lead to an excessive use of imaging studies in many dizzy patients with an actually benign peripheral-vestibular or non-vestibular cause. On the other hand, applying risk factors as ‘conditio sine qua non’ may prevent necessary imaging studies in juvenile stroke patients with a dissection or thromboembolic occlusion of the vertebral artery [19].

We rather suggest a clinical stratification of patients into specific ‘dizziness syndromes’, based on symptoms (timing, triggers) and signs from a targeted clinical examination [8, 37]. We could show that these previously proposed (and slightly modified) algorithms can be successfully applied to clinical routine in the ED when dizzy patients are assessed by non-experts in neuro-otology. Using their clinical assessment allowed to distinguish dizzy patients with a zero, moderate or high pre-test probability for ABLs on brain imaging. Based on that, we developed a clinical pathway (Fig. 2) that can be used practically by clinicians working in EDs to decide whether an individual dizzy patient should receive brain imaging or not.

To specify, patients with an episodic vestibular syndrome triggered by positional change should only receive brain imaging when there is no BPPV-typical nystagmus on Dix–Hallpike maneuver and CPV is suspected. In AVS patients, brain imaging is dispensable if the HINTS-triad is negative. While the HINTS’ high sensitivity for detecting strokes in dizzy patients is known to neuro-oto-/ophthalmological experts [14, 32], our study impressively confirms their usefulness when applied by neurology residents under ‘real-world’ conditions in the ED. When the bHIT is used as a single test to differentiate peripheral and central vestibular causes in AVS, neurology residents even revealed a higher sensitivity for central pathologies than experts in previous studies: only 11% of patients with an abnormal bHIT (indicating peripheral-vestibular pathology) later disclosed strokes on MRI versus 30% of patients in studies using experts’ assessments [5, 31]. We like to emphasize, however, that we definitely recommend using the complete HINTS triad, or even the HINTS+ assessment (including examination of a sudden hearing loss) [32], because the bHIT as a single test may be abnormal (and thereby falsely indicate an isolated peripheral vestibulopathy) in some patients with circumscribed ponto-cerebellar lesions or a combined peripheral–central vestibulopathy as in AICA infarctions that include the labyrinthine artery [4, 22, 31]. AVS patients with a normal bHIT or any other positive HINTS sign (skew deviation, nystagmus’ fast phase alternating, sudden hearing loss) should always receive brain imaging as previously suggested [14, 32] and confirmed in our cohort (21% ABLs).

Patients with acute imbalance exhibited the highest risk for ABLs among all dizziness syndromes (ABLs in 50% of vascular high-risk AIS patients). Even AIS patients with a ‘low risk’ ABCD²-score < 4 had a considerable risk of ABLs (27%). Therefore, we would extend the recommendation by Zwergal et al. [37] that brain imaging should be performed in all AIS patients independent of their individual vascular risk profile.

Patients with spontaneous transient vestibular syndrome (sTVS) remain the most challenging subgroup. Differential diagnoses encompass vertebrobasilar transient ischemic attacks (TIA), (first) manifestation of a ‘benign’ episodic vestibular disease (Meniere’s disease, vestibular migraine) and also non-vestibular causes such as transient disturbances of blood glucose/electrolytes, cardiovascular disorders (arrhythmia) and attacks of phobic/functional vertigo [29]. Clinical examination can hardly contribute to making the diagnosis, as it is usually normal when the asymptomatic patient is assessed in the ED, and transient (vestibular) symptoms are a proven risk factor for missing a stroke in the ED [36]. Brain imaging may be helpful, since diffusion-weighted MRI has been shown to detect acute ischemic lesions in 34% of TIA patients [1] and in 15% of sTVS patients [3]. In our cohort, only 8% of sTVS patients revealed an ABL on MRI. However, absence of an MRI-DWI lesion does not rule out a transient ischemia [3]. Although we could not relevantly increase the probability of detecting ABLs when including the ABCD²-score and CNS symptoms during the attack into our model, as a practical approach we suggest to consider brain imaging in those sTVS patients with an ABCD²-score ≥ 4 or accessory CNS symptoms. If any of these features is present, early brain imaging is recommended and, in case of sudden head or neck pain potentially indicating vascular dissection, an additional angiogram [9]. If markers are negative and this was the first vestibular episode, we would refrain from early CT but recommend a delayed MRI during the stay in the hospital. In case of recurrent episodes (most probable differential diagnoses: vestibular migraine, Meniere’s disease), a non-urgent MRI should be performed once to exclude rare structural pathologies such as vestibular schwannoma [23] or a pathological contact between the eighth nerve and a cerebellar artery as in vestibular paroxysmia [35].

In contrast to a prospective controlled study, this study has some limitations that are implicitly due to its retrospective design and the data acquisition in a real-world setting. First, there may have been a selection bias during the initial assessment in the ED when deciding which patient needs to be admitted to the neurological service and which patient should receive brain imaging. Indeed, we cannot fully exclude that patients who did not receive any brain imaging or only a CT scan would have shown an acute brain lesion on proper MR imaging. Furthermore, patients were examined by different neurology residents with different levels of experience and they all followed a certain clinical routine but they did not fill out a structured study protocol, e.g., when assessing the HINTS and other clinical signs. Hence, not all items/variables were available for all the individual patients and the final assessment of the signs depended on the individual level of experience and skills of the examiner.

Furthermore, we cannot rule out that the main outcome measure of this study—an ABL usually defined by a DWI lesion on delayed MRI—may not have missed a few ischemic strokes that were small enough to be MRI negative [7]. However, due to the retrospective design of the study under real life but uncontrolled circumstances, we regarded a DWI lesion on delayed MR imaging to be a more reliable marker for the final diagnosis of a stroke than the clinical diagnosis that may have been solely based on the clinical assessment (e.g., the HINTS examination) performed by different clinicians with varying levels of experience.

Patients presenting with dizziness, vertigo or imbalance to the ED should first be screened for a general medical cause [8, 30] as well as BPPV as a common and easily treatable disorder. Next, thorough history taking should include timing and triggers of dizziness symptoms and clinical examination should focus on central oculomotor signs and focal abnormalities indicating CNS involvement. Patients with any of the latter signs should undergo brain imaging, usually an early CT in the ED and a delayed MRI after admission. In the remaining patients, decision-making on the need, modality and timing of imaging studies can by guided by our proposed clinical pathway which integrates the empiric risk for ABLs into established clinical algorithms for dizzy patients. Using this practical approach could prevent many unnecessary brain imaging studies in dizzy patients without a central cause and assure imaging in those who require it.