Survival curves for event of recovery.
Mandalà M, Nuti D, Broman AT, Zee DS. Effectiveness of Careful Bedside Examination in Assessment, Diagnosis, and Prognosis of Vestibular Neuritis. Arch Otolaryngol Head Neck Surg. 2008;134(2):164-169. doi:10.1001/archoto.2007.35
To determine whether the use of 4 bedside tests (head-impulse, head-heave, head-shake, and vibration tests) can be as effective as the caloric test, a widely accepted standard, in the diagnosis and prediction of the time to recovery from vestibular neuritis.
Inception cohort (1-year follow-up), criterion standard study.
Primary referral center.
All patients had acute vertigo, and those having a diagnosis of vestibular neuritis were eligible for inclusion in the study. Sixty-eight patients (43 men and 25 women; mean age, 54.9 years) met this criterion, and 53 of them (77.9%) completed the study.
Main Outcome Measures
Spontaneous head-shaking and vibration-induced nystagmus elicited with a battery-powered device were tested wearing Frenzel goggles. The head-impulse and head-heave tests were performed manually. Caloric irrigation was administered with hot, cold, and ice water.
At baseline, more than half of the patients exhibited positive signs with all 4 tests and all had caloric paralysis or paresis. Signs with the head-impulse and head-heave tests correlated highly (odds ratio, 24.9; P < .001), as did those with the head-shake and vibration tests (odds ratio, 22.8; P < .001). Patients with a positive sign with the head-impulse or vibration test were 70% less likely to recover than were those with a negative sign. Head-impulse (hazard ratio, 0.08; P = .002) and head-shake (hazard ratio, 0.23; P = .01) test results were associated with the outcome of the caloric test.
Careful bedside examination of patients with vestibular neuritis has both diagnostic value in the short term and prognostic value in the long term.
Acute onset of severe vertigo with no other signs or symptoms is usually attributed to vestibular neuritis. Patients with vestibular neuritis also usually have disequilibrium, nausea, and vomiting but no auditory symptoms. Symptoms usually resolve in weeks, but there may be a more protracted course with persistent disequilibrium. The etiology is thought to be viral, though in rare cases, the cause may be labyrinthine ischemia.1 Caloric testing has been the traditional gold standard for detecting a peripheral vestibular deficit, but some recently developed bedside tests can also provide important information for the diagnosis and prognosis of vestibular neuritis.
In 2005, we reported use of the head-impulse test (a measure of function of the lateral semicircular canal) and the head-heave test (a measure of function of the utricle) as prognosticators of recovery from vestibular neuritis.2 We had the advantage of examining a large number of patients in the acute phase of the illness (within 3 days of onset of symptoms). Results were compared with those from the caloric test. Herein, we expand our previous study to determine whether the results of 4 bedside tests (head-impulse, head-heave, head-shake, and vibration tests)
can better predict the time to recovery from vestibular neuritis.
The cohort described is the same patient-based clinic population seen between January 1, 2002, and January 31, 2004, and whose results were reported previously.2 Patients were seen in the acute stage of the disease, 1 to 3 days after the onset of symptoms. All patients had vertigo, and those having a diagnosis of vestibular neuritis were eligible for inclusion in the study. Inclusion criteria were as follows: (1) acute vertigo for at least 24 hours;
(2) horizontal unidirectional spontaneous nystagmus for at least 24
hours; (3) no hearing loss; (4) no additional neurologic signs or symptoms and, when obtained, normal brain images; and (5) abnormal caloric test results (canal paralysis or paresis). Sixty-eight patients met these criteria. Patients were asked to return for follow-up evaluation after 1, 3, 6, and 12 months. Patients were considered to have recovered when results of both caloric testing and bedside examination were normal, after which they were not considered further for this study.
Patients underwent a complete bedside clinical examination at each visit by one of us (D.N.). Spontaneous nystagmus, head-shaking nystagmus, and vibration-induced nystagmus were tested for with fixation removed wearing Frenzel goggles (Gordon N. Stowe & Associates, Inc, Wheeling, Illinois). Vibration-induced nystagmus was elicited with a battery-powered device; the technical details are reported elsewhere.3 Vibration was applied to the mastoid region for 10 seconds, first on one side and then on the other. Test results were considered positive if persistent horizontal or horizontal-torsional nystagmus was evoked for the duration of the stimulus and if the phenomenon was reproducible. In patients with spontaneous nystagmus, test results were considered positive when nystagmus clearly increased in frequency. The head-shaking test was performed according to Kamei et al4; that is, test results were positive either if there were at least 3 consecutive beats of nystagmus in patients without spontaneous nystagmus or if there was a clear increase in frequency of eyelid movements in patients with spontaneous nystagmus.
The head-impulse test was performed by rapidly rotating the patient's head (abrupt, high-acceleration rotations of about 20°
amplitude) to the right and to the left.5 The head-heave test was performed by heaving the head of the patient rapidly (abrupt, high-acceleration interaural translations [heaves]
of about 5-10 cm in excursion).6 For both tests, the examiner stood in front of the patient, who was instructed to fix on the examiner's nose. Results of the head-impulse and head-heave tests were considered abnormal if there was an obvious corrective saccade supplementing an inadequate slow phase with acceleration toward one (affected) side.
Vestibular function was determined within 7 days after onset of symptoms using caloric irrigation with hot, cold, and ice water. Maximum slow-phase velocity of nystagmus evoked by irrigating each ear was analyzed for unilateral weakness and directional preponderance according to the Jongkees formulas.7 Caloric paresis was diagnosed when there was some response on both sides but the difference between the 2 ears was 30% or more. Caloric paralysis was diagnosed when there was no response to ice water irrigation on one side.
After the baseline evaluation, all patients were offered treatment with corticosteroids (oral methylprednisolone sodium succinate, 60
mg/d, tapered during 2½ weeks) and acyclovir (800 mg 4 times daily for 7 days). Twenty-two patients received no treatment, either because they refused it or because there was a contraindication (eg, peptic ulcer, hypertension, or diabetes mellitus). Information about treatment was missing for 3 patients; all had canal paralysis at the baseline examination.
The strategy for analysis of these data has been described previously.2 The strength of correlation between bedside tests was determined with pairwise odds ratios. The Kaplan-Meier method was used to estimate time to recovery. Prediction factors were analyzed using multiple regression models to estimate relative risk of an event (recovery).8 Recovery was assumed to have occurred at some point between the visit when the patient had a less than 30% caloric deficit and the previous visit, when the patient had a 30% caloric deficit or higher. Data were analyzed with a logistic model with a complementary log-log link. The interpretation of the parameter estimates is similar to that of a proportional hazards model: estimates are raised to the exponential power and represent the percent increase in the chance of recovery per unit increase of the parameter.
Although relative risk analyses consider dropout of participants during the study, these analyses assume that the dropout rates are not influenced by the outcome measures. Dropout in this study, however, was likely owing to the patient feeling better. Therefore, we performed sensitivity analyses to determine the robustness of the predictor estimates. The first sensitivity analysis assumed that patients who dropped out of the study recovered by the time of their first nonattendance at a follow-up visit. The second analysis assumed that those who dropped out did not recover. Estimates and standard errors from this analysis were compared with the analysis using the original data.
The bedside tests were correlated with the caloric test using the same analysis, with time-dependent covariates. This model predicts recovery, and the bedside tests are allowed to be changed across time.
Sixty-eight patients participated in the study. Two patients were excluded from further analyses because caloric irrigation could not be performed owing to tympanic perforation. Baseline testing in the remaining 66 patients revealed deficits of 30% to 99% (paresis)
in 10 and 100% in 54; baseline testing was not performed in 2 patients.
There were no differences between age and sex insofar as the severity of deficit at baseline (Table 1). Patients with positive head-heave or positive head-impulse signs were more likely to have a severe caloric deficit, but those with positive head-shake or positive vibration signs were not. Although all patients were offered treatment, those with less severe deficits were slightly more likely to accept it, but this difference was not significant, probably because of the small number of those with less severe deficits.
We previously reported the life tables estimating chance for recovery.2 In brief, the chance of not recovering by the end of 12 months was 49.3%, given that the subject remained in the study until recovery. Six patients (9.2%) did not return to the clinic for follow-up at 3 months, 5 (11.4%) did not return at 6 months, and 2 (7.1%) did not return at 12 months.
More than half of all patients had positive signs at baseline with the head-impulse, head-heave, head-shake, and vibration tests (Table 2). There were 3 patients (4.6%) with a negative head-shake sign at baseline compared with 12
(18.5%) with a negative head-impulse sign, 12 (18.5%) with a negative vibration sign, and 22 (34%) with a negative head-heave sign. All had caloric paralysis or paresis at baseline. When compared across time, head-impulse and head-heave signs correlated highly (odds ratio, 24.9; P < .001), as did head-shake and vibration test results (odds ratio, 22.8; P < .001) (Table 3). Results of the head-heave test did not correlate so highly as those with the head-shake or vibration tests.
Using information from the initial examination, bedside tests with negative signs independently predicted a higher chance for recovery (Figure). In a multiple regression model in which signs at all 4 bedside tests at baseline predicted recovery, patients with a positive sign with the head-impulse test were 70% less likely to recover than those with a negative sign (Table 4). In addition, patients with a positive sign with the vibration test were 70% less likely to recover than those with a negative sign. Positive signs with the head-heave and head-shake tests, however, were not associated with the chance for recovery, given information from the other 2 bedside tests. Accepting treatment seemed to have an almost 3-fold effect on the chance for recovery.
Dropout during the study affected the head-heave and head-shake estimates more than estimates for the head-impulse and vibration predictors, as shown in the sensitivity analysis (Table 5). When we assumed that dropout was owing to recovery (the likely reason), a positive sign with the head-heave test became an important predictor of lack of recovery. This was also the case for the head-impulse and vibration tests; that is, if signs were negative, recovery was more likely. Given information from the other bedside tests, results of the head-shaking test were not predictive of recovery, whether we did or did not assume that those who dropped out recovered. When we made the assumption that those who dropped out never recovered, the estimates and standard errors remained similar to those in the original model.
Analyzing the bedside test results across time, we could not estimate the association of the head-heave sign and the caloric test because there were no patients with a positive sign with the head-heave test and a negative sign with the caloric test. All other bedside tests had at least 1 instance when the bedside test yielded a positive sign and the caloric test yielded a negative sign. In a multiple regression model, adjusting for month and treatment at baseline, across time, the head-impulse (hazard ratio, 0.08; P = .002)
and head-shake (hazard ratio, 0.23; P = .01)
tests were associated with the outcome of the caloric test (Table 6) across time. In the presence of these 2 tests, the vibration test did not show any association with the results of the caloric test (hazard ratio, 0.36; P = .14).
An abnormal caloric response (paresis) with normal results of bedside examination was noted in 2 patients at 1-month follow-up, 3 patients at 3 months, and 2 patients at 6 months. However, 6 patients had a normal caloric response and at least 1 positive sign with a bedside test, primarily with the head-shake or vibration tests. At follow-up, the head-shaking test in 3 patients and vibration test in the other patients showed a response in the opposite direction to that predicted on the basis of loss of function, that is, nystagmus beating toward the side of the lesion.
Four patients developed benign paroxysmal positional vertigo during follow-up. In all 4 patients, this disorder was of the posterior semicircular canal type on the same side as the vestibular neuritis.
Our primary goal was to determine whether a complete and careful bedside clinical examination could be as effective as the caloric test (the gold standard) in the diagnosis of vestibular neuritis. Our results confirm that, with careful bedside clinical examination, one can almost always diagnose a unilateral peripheral vestibular deficit. This is especially important in those patients evaluated in the emergency department because quantitative caloric testing is rarely immediately available. Careful anamnesis for vascular risk factors and a general neurologic examination are required to rule out the common finding of a vertebrobasilar infarction as an acute peripheral vestibular deficit.9,10 Acute spontaneous prolonged vertigo without any other accompanying neurologic or audiologic signs and symptoms may also occur in patients with cerebellar stroke (pseudovestibular neuritis). Magnetic resonance imaging should be performed to rule out a central lesion in the presence of (1) nystagmus with central features, such as a directional changing gaze-evoked nystagmus, or (2) unidirectional spontaneous nystagmus with normal head-impulse test and normal caloric test results.
As is usually the case, at about 1 month after treatment, most patients showed no spontaneous nystagmus, implying either peripheral recovery or central compensation. We found that during the first few months of recovery, the sensitivity of the bedside examination is almost the same as with the caloric test. Only 7 patients had normal results at the bedside examination and an abnormal response to the caloric test, and only 6 patients had abnormal results at the bedside examination and a normal response to the caloric test.
In the acute phase of vestibular neuritis, within 1 to 3 days after onset of symptoms, the overall pattern of results from the 4
bedside tests varied considerably among patients, though the vibration test and head-shake test were the most sensitive. Results of the head-impulse and head-heave tests correlated highly across time, though the head-heave test was less sensitive in the acute phase of disease.
In trying to interpret these patterns, one must consider what each of these bedside maneuvers is testing for, the sensitivity of the test to partial loss of function, how easily an abnormal response can be discerned, and how the responses change as central adaptive mechanisms help to restore function. Both the head-impulse and head-heave responses test the high-acceleration, high-frequency response of the lateral semicircular canal (rotational vestibulo-ocular reflex [VOR])
and utricle (translational VOR), respectively. The head-heave test is probably less sensitive for several reasons. The translational VOR depends on the ability of the patient to converge on a near target to elicit the most robust response. The response is inherently undercompensatory in subjects with normal findings, who typically exhibit small corrective saccades in both directions. Thus, one must appreciate an asymmetry in corrective saccades to conclude that results of the head-heave test are abnormal, in contrast to abnormal head-impulse test results, in which there is usually striking asymmetry in corrective saccades between the response on the normal and the paretic side. There is also evidence that translational VOR may be more easily compensated than rotational VOR,11,12 though this has not been studied with high-frequency, high-acceleration head heaves.
Abnormal results with the head-shake test depend on (1) asymmetry in peripheral vestibular input, which is best elicited with high-speed head shaking owing to Ewald's second law; (2) storage of this asymmetric activity in central structures during head shaking; and (3) decay of the stored activity after the head stops moving. In the acute phase, central velocity storage may be impaired; thus, head-shaking nystagmus may be absent or even in the wrong direction.13,14 Vibration-induced nystagmus, however, probably depends on direct stimulation of the vestibular end organ on both sides because the stimulus is transmitted through the bone to both labyrinths; thus, if one side is less responsive, the other will predominate.15 Vibration-induced nystagmus also depends less on central velocity storage because the nystagmus can be easily appreciated during stimulation, unlike head-shaking nystagmus, which is seen only after the head stops shaking; also, vibration-induced nystagmus does not usually outlast the period of vibration. Previous studies that compared vibration- and head-shaking–
induced nystagmus have sometimes shown differences in their sensitivity and the axis of eye rotation.3,16- 18 Other factors may also account for differences between the 2 tests. The ability to store activity in the central velocity storage mechanism may differ among patients and in a single patient, depending on when in the course of the illness it is elicited. Differences among patients in physical characteristics of the skull and bony labyrinth may affect the ease of eliciting vibration-induced nystagmus.
The head-shake and vibration tests both depend on relatively small asymmetries in peripheral vestibular function and, thus, would be expected to be sensitive and also to correlate with each other. In contrast, the head-impulse and head-shake tests are better associated with the outcome of the caloric test. It is not surprising that vibration-induced nystagmus does not correlate with the caloric test across time. The vibration test is probably more sensitive than the other bedside tests, especially across time.17 When lesions are chronic, adaptive mechanisms can restore tonic activity and modulate dynamic sensitivity centrally. These compensatory changes likely improve responses to more natural motions of the head such as induced by head shaking, head impulses, and head heaves, each of which induces excitation in one labyrinth and inhibition in the other; that is, they work in a push-pull manner. In contrast, vibration-induced nystagmus is considerably more unnatural, primarily exciting afferents on both sides simultaneously.
Previous data suggest that abnormal head-heave and head-impulse responses are often associated with a severe caloric deficit.19 Milder degrees of paresis, as reflected in milder caloric response abnormalities, are still revealed by the relatively sensitive head-shake and vibration tests. Especially in the chronic phases, however, head-shaking or vibration-induced nystagmus may be in the wrong direction, perhaps reflecting mechanisms similar to those that underlie spontaneous nystagmus that is in the wrong direction during recovery (so-called recovery nystagmus). Thus, these tests alone cannot reliably be used to identify the side of the lesion.
As described in our previous study,8 both the head-impulse and head-heave tests have good prognostic value if these signs were absent. We found that negative results of the vibration test in the acute phase of disease is a strong predictor of a high chance for recovery. The absence of head-shaking nystagmus does not seem to be of prognostic value, possibly because velocity storage may be severely depressed, with severe acute unilateral loss. Accepting treatment seemed to have an almost 3-fold effect on chance for recovery. We emphasize, however, that this was not a primary outcome measure of this study, and other factors may have determined which patients opted for treatment. Thus, no specific treatment recommendations can be made on the basis of our data.
We have shown that careful bedside examination of patients with acute vertigo has both diagnostic value in the short term and prognostic value in the long term. Results correlate well with findings from the caloric irrigation test and, in many patients, obviate the need for repeated caloric tests. Still to be determined are which of these tests correlate with patients' symptoms and functional capabilities, especially in the later stages of the disease, months after the acute episode.
Correspondence: Marco Mandalà, MD, Department of Orthopedics, Radiology, and Otolaryngology, University of Siena School of Medicine, Viale Bracci 16, Siena 53100, Italy (email@example.com).
Submitted for Publication: March 29, 2007; final revision received June 6, 2007; accepted June 10, 2007.
Author Contributions: Drs Mandalà
and Nuti had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Mandalà and Nuti. Acquisition of data: Mandalà and Nuti. Analysis and interpretation of data: Mandalà, Nuti, Broman, and Zee. Drafting of the manuscript: Mandalà, Broman, and Zee. Critical revision of the manuscript for important intellectual content: Mandalà, Nuti, and Zee. Statistical analysis: Broman. Obtained funding: Nuti. Administrative, technical, and material support: Nuti. Study supervision: Nuti.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant Par 2004 from the University of Siena.