Static and Dynamic Posturography in Patients With Vestibular and Cerebellar Lesions

Objective  To assess the diagnostic usefulness of posturography in 2 well-defined patient groups with impaired balance.

Patients  Ten control subjects, 10 patients with bilateral vestibular loss, and 10 patients with cerebellar atrophy.

Outcome Measures  Amplitude, velocity, and frequency of sway in the anteroposterior and medial-lateral directions on a static platform, on foam, and on a moving platform.

Results  Both patient groups consistently had increased sway compared with controls, particularly when standing on foam or on a moving platform with eyes closed. Sway amplitude and velocity were increased about the same amount. The Romberg ratio (sway with eyes closed/sway with eyes open) did not reliably differentiate patients from controls or the 2 patient groups from each other. Some patients with cerebellar atrophy exhibited a characteristic body tremor at about 3 Hz in the anteroposterior direction.

Conclusions  Although sway amplitude and velocity were consistently increased in patients with bilateral vestibular loss and patients with cerebellar atrophy, none of the posturography measurements reliably distinguished the 2 patient groups. The finding of increased frequency of sway in the anteroposterior direction in patients with cerebellar atrophy was of limited value since the tremor was visible at the bedside.

BODY SWAY is a normal phenomenon that occurs to some degree in everyone. Although sway can be estimated by several different methods, the most commonly used technique is to record displacement of the center of pressure on a force-measuring platform (posturography). Numerous studies in patients with a variety of neurologic disorders suggest that posturography might be a useful clinical tool for evaluating balance problems in patients.^1-8 Since sway tends to be small when subjects stand on a stable platform, moving platforms (dynamic posturography) have been developed in an attempt to increase test sensitivity. The platform can be either tilted or linearly displaced and sway can be measured immediately after the movement or during the movement. Furthermore, in an effort to dissect the different sensory contributions to the maintenance of balance, systems have been developed to selectively manipulate somatosensation and vision.⁹ With these devices, the angle of sway is fed back to a dynamic posture platform or to a movable visual surround so that movement about the ankle joint or movement of the visual surround is "sway referenced." Sway measured during sway-referencing conditions with eyes closed increases in patients with vestibular lesions because such patients are reliant on proprioceptive and visual input to compensate for the vestibular loss. Such an abnormality on posturography has been called a vestibular pattern, suggesting that it is specific for vestibular system disease.^2,9

A wide range of stimulus-response parameters have been measured with both static and dynamic posturography without any consensus having developed as to which measurements are most useful. Measurements of sway area or sway path are popular but such measurements combine sway in the anteroposterior (AP) and medial-lateral (ML) directions and may, therefore, miss important directional information. For example, Maki et al¹⁰ reported that sway in the ML but not AP direction is a good indicator for the propensity to fall in older people. Mauritz et al¹¹ found that patients with cerebellar lesions, particularly those involving the anterior lobe, often have an increased frequency of sway in the AP direction. Although these and other reports^1,2 suggest the possibility that posturographic data might provide specific diagnostic information, most investigators have concluded that posturography is not useful for localizing lesions or for making specific diagnoses.¹² To further address these issues, we have performed static and dynamic posturography in 2 well-defined patient groups—one with bilateral peripheral vestibular loss and the other with cerebellar atrophy—and compared the results with those of an age-matched control group. We measured the amplitude, velocity, and frequency of sway in both the AP and ML directions, with eyes open and eyes closed, in each test condition. We also measured sway with subjects standing on foam to distort proprioceptive input and mimic a "sway-referenced" condition. Our goal was to define which measurements were most sensitive in separating patients from controls and to see whether any measurement or test condition could reliably separate the 2 patient groups.

Subjects consisted of 10 control subjects, 10 patients with bilateral peripheral vestibular loss, and 10 patients with cerebellar atrophy. All underwent quantitative visual-vestibular testing as previously described.¹³ The control subjects (5 women and 5 men) had normal vestibulo-ocular reflex gain and time constant measurements, normal visual tracking, normal optokinetic responses, and normal visual-vestibular interaction. Their mean age (±1 SD) was 46.1 ± 11.2 years (range, 31-63 years). Both patient groups complained of imbalance when walking. The patients with bilateral vestibular loss (5 women and 5 men) all had markedly decreased vestibulo-ocular reflex gain to sinusoidal rotation over a broad frequency range (>3 SDs below the normal mean at frequencies from 0.05 to 0.8 Hz).¹⁴ They had normal visual tracking, normal optokinetic responses, and normal visual-vestibular interaction.¹⁵ The mean age (±1 SD) was 45.6 ± 10.3 years (range, 25-59 years). In all cases, the vestibular loss had occurred in adulthood. The cause of the bilateral peripheral vestibular loss was ototoxicity in 3 patients, bilateral autoimmune inner ear disease in 1, and idiopathic in the remaining 6. Two of these latter patients had a family history of similar bilateral peripheral vestibular loss. The patients with cerebellar atrophy (7 women and 3 men) all exhibited normal vestibulo-ocular reflex gain but impaired visual tracking, optokinetic responses, and visual-vestibular interaction. The mean age (±1 SD) was 49.1 ± 12.1 years (range, 21-65 years). All were ambulatory but each had obvious gait and extremity ataxia on clinical examination. Clinical diagnoses in these 10 patients were olivopontocerebellar atrophy in 6 and isolated cerebellar atrophy in the remaining 4.¹⁵ One of the former and 2 of the latter patients had a family history of similar syndromes.

The Chattecx balance system (Chattecx Corp, Chattanooga, Tenn) uses vertical force transducers to determine instantaneous fluctuations in the center of pressure (COP).¹⁶ Grabiner et al¹⁷ have shown that COP calculated in this fashion with the balance system is a good estimate of COP measured with traditional biomechanical force plates. There are 2 pairs of independent force transducers, 1 pair for the forefoot and heel of each foot. The distance between the 2 foot plates was maintained at 4 cm for all individuals tested. The foot plates sat on a motor-driven platform that tilted up and down about a central axis at a frequency of 0.1 Hz and a peak amplitude of 4°. Flat-soled shoes were worn for testing. Subjects stood on the platform with their feet centered on the foot plates (in parallel) while wearing a security harness to prevent them from falling. They were instructed to look straight ahead at the surrounding room with arms at the sides and were allowed to stand on the platform until they felt secure. The standard test battery included measurements of sway for 10 seconds with eyes open and eyes closed under each of 4 conditions: (1) platform still (static); (2) foam rubber (thickness, 7.6 cm; density, 30.3 kg/m³; Specialty Composites Corp, Indianapolis, Ind) on a still platform; (3) platform tilting in the AP direction; and (4) platform tilting in the ML direction.

Complete details of the analysis have been reported elsewhere.¹⁶ In brief, the balance system measures the relative vertical loading or distribution of weight (vertical force) beneath the heel and forefoot of each leg. The COP was calculated from these vertical forces. To obtain a measurement of average amplitude (A) of sway in each direction (A_AP and A_ML), we calculated the root-mean-square about the mean COP for each 10-second test. We then differentiated the instantaneous COP units using a 2-point difference formula (25 Hz low-pass filter). To obtain a measure of the average velocity (V) of sway, we calculated the root-mean-square of V_AP and V_ML for each 10-second test. Finally, we performed a frequency analysis (fast Fourier transform) of A_AP and V_AP and generated histograms of the power in 0.5-Hz bins from 0 to 5 Hz. We only assessed the frequency content of sway in the AP direction because prior reports found this measurement to be most useful for differentiating patients with cerebellar lesions from controls.¹¹ To summarize the frequency data in a single value, we calculated a frequency quotient, defined as the power of frequencies between 2 and 5 Hz divided by the power of frequencies between 0 and 2 Hz. The cutoff between low- and high-frequency sway of 2 Hz was empirically chosen based on previously reported pilot data.¹⁶ Because several patients with cerebellar atrophy were unable to stand on foam, testing on foam was only studied in the controls and patients with bilateral vestibular loss.

As we¹⁶ and others^10,11 have previously reported, sway in the AP direction was consistently greater than sway in the ML direction in the control subjects (both for amplitude and velocity measurements) (compare top and bottom of Table 1). Sway amplitude and velocity increased when subjects stood on foam and when the platform moved, with the greatest increase in sway occurring in the direction of platform movement. The percentage increases in amplitude and velocity of sway during platform movement were about the same in the AP and ML directions. The Romberg ratio (sway with eyes closed/sway with eyes open) was on average about 1.5 in the static condition, increasing to near 2.0 in the dynamic tests, and reaching a high of 3.0 for V_AP measurements on foam (Table 2). There was large variability in these measurements, however, with the SDs being about equal to the mean values, showing that some subjects had much greater increase in sway with eye closure than did others. One control subject had a Romberg ratio less than 1.0 for measurements on the static test (ie, amplitude and velocity of sway were greater with eyes open than with eyes closed). All subjects had greater amplitude and velocity of sway with eyes closed compared with eyes open on the other tests, with the highest ratios occurring on the foam test.

As expected, the frequency content of sway in the AP direction was greater for velocity than amplitude measurements (Table 3). None of the measurements had much energy in frequencies above 2 Hz for any of the tests in the controls. Unlike the amplitude and velocity measurements, there was relatively little change in the frequency of sway with eye closure (ie, the Romberg ratio was about 1.0 for all tests).

Overall, the static test condition was least sensitive for distinguishing patients from controls. Sway amplitude and velocity in both the AP and ML directions were increased about the same amount on average, but several patients had values within the normal range (Table 1). The best separation between patients and controls occurred on the foam and dynamic tests with eyes closed. The differences between patients and controls on these tests were highly significant (P<.001) and nearly all patients were identified as having abnormal measurements for both amplitude and velocity (Table 1). The Romberg ratio was less sensitive than absolute measurements with eyes closed in differentiating patients and controls (Table 2).

Measurements of the amplitude and velocity of sway in both the ML and AP directions overall increased about the same amount in the patients with cerebellar atrophy and in the patients with bilateral vestibular loss. There were no differential features identified on any of these measurements. On the other hand, measurements of the frequency of sway in the AP direction did discriminate between the 2 patient groups and between the patients with cerebellar atrophy and the controls (Table 3). The best discrimination occurred during the static testing. Seven of 10 patients with cerebellar atrophy had a significantly increased frequency of sway in the AP direction when standing on a static platform with eyes open. None of the patients with bilateral vestibular loss had a significant increase in frequency of sway under similar conditions. This increase in the frequency of sway was even more pronounced with eye closure in the static condition and during the dynamic tests, but under these conditions, 3 of the patients with bilateral vestibular loss also had a significant increase in the frequency of sway compared with controls. The explanation for this finding is apparent when one looks at the frequency content of sway in the static and a dynamic condition for the 2 patient groups (Figure 1). In the static test condition, patients with bilateral vestibular loss have relatively little frequency content above 2 Hz, whereas patients with cerebellar atrophy have a second peak in frequency content between 2 and 3 Hz. During the dynamic tests, patients with bilateral vestibular loss continue to have the main peak in the frequency content around 1 Hz but the curve is shifted upward overall so that there is much more frequency content above 2 Hz and the difference in curves between the 2 patient groups becomes less distinct.

Overall, we found that patients with bilateral vestibular loss and patients with cerebellar atrophy had increased sway compared with controls, particularly when standing on foam or on a moving platform with eyes closed. Traditionally, it has been thought that increased sway with eye closure is characteristic of vestibular lesions and not cerebellar lesions.¹⁸ This is the basis of the clinical Romberg test. However, we found that sway increased with eye closure to about the same degree in patients with cerebellar lesions and with bilateral vestibular loss. Others have also found that changes in sway with eye closure does not discriminate between these 2 patient groups.¹¹ The Romberg ratio (sway with eyes closed/sway with eyes open) was less effective than absolute measurements of sway with eye closure in differentiating patients from controls mainly because of the wide range of normal Romberg ratios. For example, using the amplitude of sway in the AP direction on the static test, the Romberg ratios in control subjects ranged from 0.95 to 4.0. Others have found a similar range for Romberg ratios in control subjects.¹⁹ Because standing on foam distorts proprioceptive input from the feet, we expected that standing on foam with eyes closed would be the best subtest for discriminating between patients with bilateral vestibular loss and controls. However, the percentage increase in sway with eye closure was about the same when patients stood on a tilting platform as when they stood on foam. Angular tilt of the platform and the associated stretch reflexes apparently destabilized the body as much as when standing on foam.

Sway amplitude and sway velocity were consistently greater in the AP direction than in the ML direction in controls and patients, presumably owing to the increased mechanical stability of the ankle in the ML direction. Although sway (both amplitude and velocity) tended to increase slightly more in the AP direction in both patient groups, measurements of sway in either plane separated normal from abnormal results. Although we suspected that measurements of the velocity of sway might be a better indicator of the effort required to maintain balance during platform perturbations than the amplitude of sway,²⁰ amplitude and velocity measurements increased about the same amount in both patient groups on the dynamic tests.

The only response measurement that did show differential effects in the 2 patient groups was measurement of the frequency of sway in the AP direction. The frequency quotient (power above 2 Hz/power below 2 Hz) was significantly higher in the patients with cerebellar atrophy compared with the controls and patients with bilateral vestibular loss. The best distinction was achieved with measurement of the frequency of velocity of sway in the AP direction with either eyes open or eyes closed on a static platform. The frequency of sway increased further during platform tilting in the cerebellar group but it also increased in the bilateral vestibular group so that the separation between the 2 groups became less clear. Mauritz et al¹¹ suggested that this increase in the frequency of postural sway was specific for lesions of the anterior lobe of the cerebellum, particularly of the vermal and paravermal parts of the anterior lobe. They reached this conclusion because the increased frequency of sway was typically seen in patients with cerebellar atrophy secondary to chronic alcoholism, a disorder with remarkably localized abnormality in the anterior lobe of the cerebellum. Our patients had generalized cerebellar atrophy so that our material was not suitable for localizing the source of the increased frequency of sway within the cerebellum.

The mechanism for the increase in postural sway seen in patients with vestibular and cerebellar lesions is poorly understood. Maintenance of balance when standing is a complex process that involves multiple peripheral sensory inputs, central integrating pathways, and efferent outputs.^21,22 Postural sway presumably reflects noise and regulatory activity within these afferent−efferent control loops. The amplitude and velocity of sway seems to increase in a nonspecific fashion with altered sensory input (vestibular, somatosensory, or visual) or with brainstem and cerebellar lesions. Vestibular signals may be critical for scaling the magnitude of the postural responses.^23,24 The approximate 3-Hz postural tremor in the AP direction recorded in patients with cerebellar lesions probably results from delays within the long loop cerebellar postural reflexes.²⁵

Does posturography have a role in the clinical diagnosis of patients with balance disorders? As currently conducted, the answer is probably no, although more studies in well-documented patient groups are needed. There is a large intraindividual and interindividual variance in normal sway measurements, on both static and dynamic posturography.¹⁶ Normal and abnormal results often overlap even when assessing patients with obvious deficits such as those studied in this report. The characteristic AP postural tremor recorded in patients with cerebellar lesions is an exception, but the tremor in our patients could be seen at the bedside without the need for recording equipment. We do not know if posturography can identify subclinical AP tremor in patients with more subtle cerebellar deficits. Distorting somatosensory and visual signals with sway referencing seems a logical method for separating the different sensory inputs but does not reliably separate peripheral from central lesions. Although certain patterns are common in patients with vestibular lesions, patients with well-documented vestibular lesions often have normal posturography results.²⁶ Sensitivity and specificity probably could be improved by combining posturography with other tests such as electromyography from postural muscles² and by development of more informative analysis techniques.²⁷

Accepted for publication August 25, 1997.

This work was supported by grants AG9063 and PO1 DC02952 from the National Institutes of Health, Bethesda, Md.

Reprints: Robert W. Baloh, MD, Department of Neurology, University of California–Los Angeles, Box 951769, Los Angeles, CA 90095-1769 (e-mail: rwbaloh@ucla.edu).