Diagnosing Stroke in Acute Vertigo: Sensitivity and Specificity of HINTS Battery in Indian Population
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.310089
Source of Support: None, Conflict of Interest: None
Keywords: HINTS, MRI, sensitivity, specificity, stroke, vertigo
Vertigo is the third most common major medical symptom reported in general medical clinics and accounts for roughly 3%–5% of visits across clinical settings. Epidemiologic data indicate that more than 20% of adults experience dizziness within a given year.
Most patients presenting with acute vertigo are believed to suffer from an acute, self-limited, presumed viral or post-viral vestibular disorder commonly called vestibular neuritis (VN); acute peripheral vestibular disorders. About 25% of such cases can be called “pseudo VN” and are due to central causes, most often stroke.,
In the primary care setting and emergency department, in patients with first episode of acute vertigo, decision-making is both difficult and critical because there is no previous diagnosis to rely on. Central causes of acute vertigo, particularly brainstem and cerebellar ischemic stroke, can closely mimic benign peripheral causes, particularly VN and migrainous vertigo. These benign conditions have to be separated from such serious disorders.
Head impulse, nystagmus, and test of skew deviation (HINTS) battery was previously studied in American population. An attempt has been made in this research to study the sensitivity and specificity of HINTS battery for diagnosis of stroke in Indian population with acute-onset first episode of spontaneous vertigo, with final diagnosis obtained after magnetic resonance imaging (MRI) brain.
All patients of either sex who visited outpatient department or were admitted in Poona Hospital and Research Centre, Pune, Maharashtra, between August 2014 and April 2016 with first episode of acute-onset spontaneous vertigo and ready to participate were included in the study. Permission was obtained from ethics committee and scientific advisory committee of the institution. Informed written consent of patients was obtained after explaining details of the study and risk involved.
Patients who presented with acute-onset first episode of spontaneous vertigo with age >18 years.
Patients with a history of recurrent vertigo, chronic dizziness, only positional vertigo, and patients who withdrew written informed consent.
Based on previous studies,, setting an alpha error at 0.05 and power at 80%, a sample size of 75 patients was calculated using the formula by Charan and Biswas for this cross-sectional observational study.
In this study, vertigo was defined as a specific type of dizziness in which the patient has perception of spinning or tilting. Each patient was subjected to detailed clinical history, central nervous system, and vestibular examination including bedside oculomotor tests and HINTS.
Horizontal head impulse test (h-HIT) was conducted by rapid, passive head rotation from center to lateral (10–20°) position as subject fixates at central target (e.g., examiner's nose). Nystagmus was examined for its direction (unidirectional, direction-changing, or other). Skew deviation was done by alternate cover test. With patient fixating on central target, normal response to alternately occluding each eye was for eyes to remain motionless. Abnormal response was indicated by the presence of a refixation saccade after transfer (removal) of cover.
All patients underwent MRI brain, after bedside evaluation. Stroke was diagnosed by MRI brain. If initial MRI brain was normal and still clinical signs favor stroke, repeat MRI brain-diffusion-weighted imaging (DWI) study was done at 72 h after symptom onset to confirm diagnosis. Peripheral lesions were diagnosed by normal MRI and other clinical signs.
We defined HINTS examination as either benign (abnormal h-HIT plus direction-fixed horizontal nystagmus plus absent skew) or dangerous (normal/untestable h-HIT or direction-changing horizontal nystagmus present/untestable or skew deviation present/untestable) which rules in stroke and have compared sensitivity, specificity, and likelihood ratios for the presence of stroke relative to other neurological findings and early MRI with DWI.
Data on categorical variables were presented as percentages. Statistical significance of difference of categorical variables was tested using Fisher's exact test. Statistical measures of diagnostic accuracy such as sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and Youden's index were used to quantify the diagnostic efficiency of HINTS at presentation against final MRI which was considered a gold standard for the study. P values <0.05 were considered as statistically significant. The entire data were analyzed using Statistical Package for Social Sciences version 20.0, Inc. (Chicago, IL, USA) for MS Windows.
We enrolled 75 patients (49 males and 26 females) presenting as first episode of acute spontaneous vertigo. After clinical examination, clinical tests, and MRI brain, these 75 patients were grouped into three categories: stroke 35 (46.6%) patients, VN 32 (42.7%), and vestibular migraine (VM) 8 (10.7%) patients.
As depicted in [Table 1], there was statistically significant difference between age groups and final diagnosis. Most of the patients of stroke were >51 years, whereas most of the patients of VN and VM were <50 years of age.
Out of 35 patients who had stroke, 8 (22.9%) had no risk factors, 21 (60%) had one risk factor, whereas 6 (17.1%) had =2 risk factors. We found that hypertension was the most common risk factor, followed by diabetes mellitus and ischemic heart disease.
In all, 22 of 35 (62.9%) patients with stroke had neurological signs, whereas 13 of 35 (37.1%) did not have any neurological signs. Neurological signs were crossed sensory loss (68.2%), finger nose (FN) ataxia (59.1%), dysphagia (31.8%), and dysarthria (4.5%). Patients with VN and VM did not have any neurological signs. Of 13 stroke patients presented without any focal neurological signs, 7 patients (53.8%) had medial posterior inferior cerebellar artery (PICA) territory infarct and 5 patients (38.5%) had lateral PICA infarct.
As depicted in [Table 2], there was statistically significant difference between abnormal h-HIT, presence of skew deviation, direction-changing nystagmus, and final diagnosis. Of 35 stroke patients, 33 had normal h-HIT. Two anterior inferior cerebellar artery (AICA) infarct had abnormal h-HIT. All patients of VN had abnormal h-HIT, whereas all patients with VM had normal h-HIT. All patients with positive skew deviation were diagnosed as stroke. Skew deviation was absent in all patients with VN and VM. Final diagnosis differed significantly (P = 0.001) between cases who had unidirectional nystagmus. All patients with VN had unidirectional nystagmus. Six patients of stroke and one patient with VM also had unidirectional nystagmus. Significantly higher proportion of cases with direction-changing nystagmus had higher incidence of stroke (P = 0.001). Of 35 stroke patients, 19 patients (54.3%) presented with direction-changing nystagmus. One patient with direction-changing nystagmus was diagnosed as VM. Direction-changing nystagmus was not found in any patients with VN.
As depicted in [Table 3], there was statistically significant difference between dangerous HINTS and final diagnosis. Significantly higher proportion of cases with dangerous HINTS had higher incidence of stroke. All patients with VN had benign HINTS. Of 75 patients, 42 (56%) patients had dangerous HINTS in our study. Of 42 patients with dangerous HINTS, 34 (81.0%) were diagnosed as stroke, whereas 8 patients (19.0%) were diagnosed as VM. HINTS examination findings of all patients with VN were benign.
As depicted in [Table 4], normal h-HIT showed higher sensitivity, specificity, PPV, NPV, and accuracy with relatively higher value of Youden's index in predicting the incidence of stroke. Direction-changing nystagmus showed higher sensitivity, specificity, PPV, NPV, and accuracy with relatively higher value of Youden's index in predicting the incidence of stroke. The presence of skew deviation showed relatively low sensitivity, NPV, and Youden's index but high specificity, PPV, and accuracy in predicting the incidence of stroke. Dangerous HINTS showed high sensitivity, specificity, PPV, NPV, and accuracy with relatively higher value of Youden's index in predicting the incidence of stroke. Dangerous HINTS with hearing loss (HINTS Plus) increased HINTS sensitivity up to 100%, with high specificity, PPV, NPV, and accuracy with relatively higher value of Youden's index in predicting the incidence of stroke. “HINTS Plus” battery was more sensitive than only HINTS.
Of 35 stroke patients, 6 patients (17.1%) who had normal initial MRI were diagnosed as stroke on second MRI done after 72 h.
As depicted in [Table 5], first MRI investigation showed relatively low sensitivity and NPV, but with high specificity and PPV, and accuracy with relatively higher value of Youden's index in predicting the incidence of stroke. Initial MRI was less sensitive than HINTS battery to diagnose stroke (82.86% vs. 97.1%).
Kattah et al. reported that of 101 patients, 25% with ischemic stroke were under 50 years of age which is comparable to our study.
Lee et al. stated that 12% of patients with cerebellar or brainstem stroke had no known risk factors, whereas in this research 22.9% patients with stroke had no risk factors. Atanassova et al. reported that 42 (64.6%) subjects had at least one of the conventional cerebro-vascular risk factor (arterial hypertension, ischemic heart disease, diabetes mellitus, dyslipidemia, family history for vascular accidents, dementia, etc.). A hospital-based study conducted by Chih-Ying et al. stated that hypertension and DM were more frequent risk factors, whereas in this research the most common risk factors were hypertension, diabetes mellitus, and ischemic heart disease.
In this study, 37.1% of patients did not have any neurological signs at presentation, called as “pseudo VN” except subtle oculomotor signs, whereas Kattah et al. and Saber Tehrani et al. reported that 50% and 27%, respectively, of patients had no neurological signs.
Stroke causing vertigo or dizziness is mostly located in lateral brainstem and cerebellum. As a result, primary motor and sensory pathways are usually intact. Kattah et al. mentioned that lack of motor involvement or brainstem signs is one of the major reasons that these “pseudo VN” are not recognized in emergency department.
Saber Tehrani et al. reported lateral medulla (60%) as the most common stroke location for strokes causing isolated vertigo. We found similar result in our study; medial PICA territory infarct was the most common site (53.8%) followed by lateral PICA (38.5%) in patients presented as “pseudo VN.”
In our study, two patients presented with acute hearing loss with spontaneous vertigo, both of them diagnosed as AICA territory infarct. One of them had no dangerous sign on HINTS examination and could be missed if only HINTS would be considered, so adding acute hearing loss to HINTS battery—”HINTS Plus” approach increased sensitivity to 100%. Kattah et al. reported that 3% of patients presented with acute auditory symptom which would be missed with only HINTS examination; these patients were diagnosed with HINTS Plus battery. Newman-Toker et al. stated that one patient in their series missed by HINTS would have been captured by the HINTS “plus” a pproach. They concluded that acute hearing loss is a sign of AICA infarction in patients with acute vestibular syndrome (AVS). Lee and Cho studied 16 consecutive cases of AICA infarction diagnosed by brain MRI. Five patients (31%) had an acute auditory syndrome as a prodrome of AICA infarction.
Kattah et al. and Oppenheim et al. reported that h-HIT result probably outperforms MRI with DWI sensitivity for detecting posterior fossa stroke (80%), with comparable specificity. Newman-Toker et al. studied 44 patients of AVS; all patients with VN had positive h-HIT. Most patients with stroke had a negative h-HIT (91%). However, three patients with stroke (9%) demonstrated a positive h-HIT; these three patients had positive skew deviation and diagnosed as AICA territory infarct on MRI brain. Lee et al. also found similar result in their study and concluded that negative h-HIT was the most useful clinical sign for differentiating between central and peripheral causes of AVS.
Regarding false-positive h-HIT, Lee and Edlow et al. mentioned that falsely abnormal h-HIT findings were almost exclusively associated with AICA infarctions. Our study substantiated these findings. AICA stroke often results in an ischemic lesion of the labyrinth or eighth nerve root entry zone that mimics VN anatomically and physiologically.
We found direction-changing nystagmus as 76% sensitive and 97.4% specific to diagnose stroke in patients with AVS. Tarnutzer et al. reported direction-changing nystagmus as 38% sensitive and 92% specific. In our study, presence of skew deviation was found 60% sensitive and 100% specific for stroke. Kattah et al. reported that skew deviation was 30% sensitive and 98% specific for stroke.
Combining all three oculomotor signs as HINTS battery in diagnosing stroke was 97.1% sensitive and 80% specific in our study. We had eight patients who were diagnosed as first episode of VM and all had dangerous HINTS, and adding acute hearing loss to HINTS, called “HINTS Plus,” it became 100% sensitive to diagnose stroke. Kattah et al. stated that of patients with one or more risk factors for stroke, none of the patients had VM as final diagnosis. They reported HINTS battery as 100% sensitive and 96% specific for stroke, whereas Newman-Toker et al. reported that HINTS “plus” had an estimated sensitivity and specificity of 99% and 97%, respectively, for identifying central causes of AVS, whether isolated or not.
In our study [Table 6], initial MRI was 83% sensitive to detect stroke, whereas HINTS' battery sensitivity was 97%. Kattah et al. reported 88% sensitivity of early MRI with DWI for lateral medullary or pontine infarction, which was lower than that of bedside HINTS examination with comparable specificity (100% vs. 96%). Similar results were mentioned by Newman Toker et al. and Tarnutzer et al. in their studies.
Brevern et al. reported that VM may be present in myriad of ways, with examination signs mimicking an acute or central peripheral vestibular syndrome. Seemungal et al. reported that in patients presenting with vertigo for the first time, it may not be possible to clinically differentiate between acute VM and an acute cerebellar stroke because both may present with severe vertigo, nausea, vomiting, posterior headache, gait ataxia, few ocular signs, and a preserved HIT. They concluded that in such cases, immediate brain MRI should be performed and possibly repeated if the initial scan is normal and clinical suspicion is high. A history of multiple similar episodes over a prolonged period is likely to represent VM and not stroke. Kerber et al. also mentioned that early diagnosis with HINTS allows prompt and correct management, which can prevent disability or death due to misdiagnosis-related harm for missed stroke.
HINTS battery is a clinical sign which needs experience and is subject to interpersonal variation.
Typical neurological signs were absent in 37% of stroke patients; they had only subtle oculomotor signs. Presence of any one “dangerous” sign of HINTS battery, either normal h-HIT or direction-changing nystagmus or positive skew deviation, was 97.1% sensitive and 80% specific for stroke. HIN?TS battery was found more sensitive than initial MRI brain done in first 24 h of symptom onset in diagnosing stroke in patients presented with acute spontaneous vertigo (97.1% vs. 82.86). NPV of initial MRI in our study was 86.36%, while of HINTS battery was 97.0%.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6]