Sharpened Rhomberg vs. Sway Magnetometry in Assessing Postural Instability Before and After Driving Simulation Participation

Nicole Fowler
University of Maryland at College Park.
Vaughan Inman
Senior Psychologist to FHWA Office of Research Development and Technology.

Abstract

The Sharpened Rhomberg Test is intended to detect simulator induced ataxia in HCS driving simulation participants. Sway magnetometry has been identified as a more reliable and sensitive measure of ataxia. Sway magnetometry does not require participants to maintain an awkward posture, and thus can be used to assess any person who can stand or sit upright. For these reasons, sway magnetometry is proposed as a replacement for the SRT in the simulator sickness management procedure.

Keywords: sway magnetometry, simulator sickness, ataxia, sharpened rhomberg test, postural sway

Introduction

The Highway Driving Simulator (HDS) is used by the Federal Highway Administration (FHWA) Office of Research, Development and Technology (RD&T) to conduct human factors research at the Turner Fairbank Highway Research Center. This research is conducted by the Human Centered Systems (HCS) team. The HDS is comprised of a 1998 Saturn SL1 body attached to a 3-Degree of Freedom motion base; a cluster of networked computers to generate visual scenes and vehicle dynamics; 5 Barco 909 projectors; and a cylindrical projection screen. With 5 projectors the HDS is capable of displaying a 240 degree horizontal by 50 degree vertical virtual world. However, as currently configured the visual display is limited to 180 degrees in the horizontal dimension.

Simulator sickness, which is similar to motion sickness and is possibly related to motion sickness, is a common problem in man-in-the-loop simulators of all types, e.g., flight simulators, ship simulators, simulated static virtual environments, and ground vehicle simulators (Kolasinski, Goldberg, & Hiller 1995). Ataxia is a symptom that is sometimes associated with simulator sickness and has also been reported to occur in the absence of other simulator induced symptoms (Smart, Stoffregen, & Bardy, 2002).

HCS uses two tests to identify and quantify symptoms of simulator sickness. These tests are: the Simulator Sickness Questionnaire (SSQ) (Kennedy, Lane, Berbaum, & Lilienthal, 1993) and the Sharpened Rhomberg Test (SRT).

The SSQ is a self-report of symptom (Figure 1). This questionnaire consists of sixteen symptoms with a severity range from None, Slight, Moderate to Severe. The questionnaire is given to participants to complete before they enter the simulator and again after the simulation is finished. Before-after changes are used to assess the degree of simulator induced symptoms. The SSQ produces four scores: (1) Total Severity; (2) Oculomotor; (3) Nausea, and (4) Disorientation. The total severity score provides an overall measure of severity. The other three scores are computed from the same 16 measures as the total according to formulas derived by Kennedy et al. (1993) through factor analysis. The oculomotor score derives primarily from ratings of eyestrain, difficulty focusing, blurred vision, and headache. The nausea score is based primarily on ratings of nausea, stomach awareness, increased salivation, and burping. The disorientation score is based on ratings of dizziness and vertigo. Theoretically, the disorientation score should correlate with ataxia measures, and thus with SRT. Results for this have been inconclusive with some studies yielding a significant disorientation-SRT correlation (Kennedy, Berbaum, & Lilienthal, 1997) and others failing to show a correlation (e.g., Cobb, 1999; Cobb & Nichols, 1998).

Ataxia is a loss of fine motor control that results in movements taking on a more ballistic (open loop) character. Although ataxia can result from damage to the cerebellum and/or neural pathways associated with the cerebellum, the ataxia associated with simulator exposure is transient and has never been noted to last more than a few days. More typically, it lasts only minutes. Tests for the ataxia associated with simulator exposure assess postural stability – the efficiency with which the individual maintains an upright posture of the head or body.

Currently, HCS uses the Sharpened Rhomberg Test (SRT) procedure to assess postural stability. The SRT was adopted because there is a documented history of its use in testing military pilots who use military training and research flight simulators (Kolasinski, Goldberg, & Hiller, 1995). The users of military flight simulators tend, on average, to be younger, fitter, and subjected to more rigorous screening criteria than is the average for the HCS driving simulator participant population. As is documented below, a large proportion of HCS participants are unable to perform the SRT when they are well, and the proportion that cannot perform it increases with age. Thus, a test that offers detection of postural instability to a greater proportion of participants is desirable. Furthermore, some research, cited below, suggests that the SRT is not a reliable indicator of postural stability.

This paper suggests that another test, using sway magnetometry, would offer protection to all participants, regardless of age, would be more reliable and would reduce risk to participants.

Theories of Simulator Sickness

Simulator sickness is similar to motion sickness but does not require motion to induce symptoms. The most common symptoms of simulator sickness are general discomfort, apathy, drowsiness, headache, disorientation, fatigue, pallor, sweating, salivation, stomach awareness, nausea, vomiting, retching, postural instability, and flashbacks (Kolasinski, Goldberg, & Hiller, 1995). The leading explanations of simulator sickness are the sensory conflict theory (Reason & Brand, 1975) and the postural instability theory (Riccio & Stoffregen, 1991).

The sensory conflict theory posits that there is a conflict between motion information obtained from various sensory systems (visual, vestibular, and proprioceptive systems) and learned representations of what those inputs should be in a given environment. The mismatch generated by the conflicting sensory patterns is theorized to result in both motion sickness and, with time, adaptation (Reason & Brand, 1975).

The postural instability theory hypothesizes motion sickness is caused by a breakdown in postural stability. According to the postural instability theory, motion sickness occurs when the organism encounters an environment for which it has not learned the requisite strategy for maintaining postural stability (Riccio & Stoffregen, 1991). Postural stability is defined as the coordinated stabilization of all body segments.

Neither theory fully accounts for all motion and simulator sickness phenomena. However, because only the postural instability theory specifically addresses ataxic symptoms, it is used as the theoretical basis of the discussion of ataxia tests that follows.

Postural Instability

Riccio and Stoffregen (1991) define postural instability as a degradation of stability. The postural control system is maintained by inputs from the visual, somatosensory, and vestibular systems. Body sway or steady posture is the output from the postural control system (Murata, 2004).

According to the postural instability theory, instability is a necessary precursor of motion sickness (Riccio & Stoffregen, 1991). This hypothesis has been supported in several studies (Stoffregen, 1998; Stoffregen, Hettinger, Haas, Roe, & Smart, 2000; Smart, Stoffregen, & Bardy, 2002) that used various techniques, but all involved very small sample sizes. Several studies have reported increases in postural instability following exposure to simulators (e.g., Kennedy, Fowlkes, & Lilienthal, 1993; Kennedy, Berbaum, & Lilienthal, 1997; Cobb & Nichols, 1998;Murata, 2004; Fushiki, Kobayashi, Asai, & Watanabe, 2005). However, the latter studies do not serve to distinguish between the sensory conflict and postural instability hypotheses, because they do not demonstrate that postural instability preceded simulator sickness. Both sensory conflict and postural instability theories are consistent with postural instability accompanying other motion sickness symptoms; only the postural instability theory holds that postural degradation must precede other sickness symptoms.

Regardless of theory, it is important to measure postural stability because large degradation of postural stability can pose a threat to simulator research participants if they operate a vehicle following simulator exposure (Crosby & Kennedy, 1982; Kennedy, Berbaum, & Lilienthal, 1997).

If the postural instability theory is correct in positing that postural instability causes simulator sickness, then a potential use of postural stability measures is to predict simulator sickness before the onset of systems ( cit_bfStoffregen & Smart, 1998; Stoffregen, Hettinger, Haas, Roe, & Smart, 2000; Smart et al., 2002). Those investigators purported to detect degradations in postural stability during simulator sessions using a sway magnetometry device called Flock of Birds®, which is distributed by Ascension Technologies, Inc.

Sharpened Rhomberg Test

The Sharpened Rhomberg Test (SRT) is the ataxia test currently used by FHWA RD&T. Participants administered the SRT are instructed to stand erect on a firm, level surface with the feet aligned in a tandem (heel-to-toe) position. Arms are folded across the chest. Once stable in this position, the participant is instructed to close the eyes and to maintain this position for sixty seconds. The SRT posture is depicted in Figure 2. The measure of SRT performance is the length of time that the subject maintains this stance. Time is recorded to the nearest second. Timing stops when the subject opens his or her eyes, or moves his or her feet or arms. If the posture is maintained for 60 s on the first trial, that trial is terminated and a score of 60 is recorded. If the participant falters before 60 s elapses, then two additional trials are administered and the performance score is the mean of the three trials. Trials on which the posture is maintained for less than 5 s are not recorded and are retried. After three consecutive failures to maintain the posture for at least 5 s, the SRT is discontinued and it is noted that the participant cannot perform the test.

Figure 2. The proper stance for the Sharpened Rhomberg test (Graybiel and Fregly, 1961).

Other static posture tests are variations of the SRT. These include the standing on preferred leg test, standing on non-preferred leg test, and the stand on one-leg eyes closed test. The SRT is sometimes referred to as the Tandem Romberg Test. The SRT is an inexpensive test to administer in that it does not require the use of a special apparatus. Kennedy, Fowlkes, & Lilienthal (1993) and Kennedy, Berbaum, & Lilienthal (1997) compared the SRT to other static posture tests and concluded that the SRT was the most sensitive of the tests that they evaluated.

Investigators have identified several problems with the SRT:

• Lack of reliability.
• Subjectivity in scoring.
• High inter-trial variability.
• Substantial practice effects over trials.
• Inability of large proportions of the population to perform the test.

Reliability

In introducing the SRT, Graybiel and Fregly (1966) reported high test-retest reliability scores. However, reliabilities were only computed for 12 males between 18 and 49 years of age. The test-retest correlation between day one and day two was relatively low, r = 0.40, which accounts for only 16 percent of the between test variability. Test-retest reliability was relatively high, r = 0.91, between days 9 and 10; a finding consistent with a practice effect, as discussed below.

Cobb and Nichols (1998) compared SRT scores of 40 participants in a virtual reality experiment to scores on three tests: (a) a subjective measure of ataxia, (b) a sway magnetometry measure, and (c) the SSQ. They found that although the SRT before-after scores showed the same trends as those from the subjective measure and sway magnetometry, only the latter measures showed statistically reliable changes in postural stability. Within and between subject variability in SRT performance precluded the apparent trends on that test from reaching the 0.05 level of significance.

Subjectivity

Cobb (1999) has suggested that the scoring of the SRT is subjective, because the failure to maintain the SRT’s prescribed posture, illustrated in Figure 2, is ill-defined. This criticism is based on the observation that descriptions of failures in posture are not particularly precise, e.g., “moves foot”, or “unfolds arms”, “opens eyes”, or “loses balance”. Cobb points out that Kennedy and Lilienthal ( cit_bf1994) proposed a video-based scoring method to provide a more refined quantitative scoring system. In practice, the SRT scoring procedure may not be as subjective as Cobb implies, because most failures to maintain the posture are generally rapid and catastrophic – foot and arm movements are nearly always large and precipitous, and are generally associated with rapid lifting of the eyelids.

Variability

Although subjectivity and poor reliability may lead to excessive variability, HCS data tend to support the assertion that the SRT suffers from high variability. For 149 participants who were able to perform the SRT in a recent study (Davis & Inman, 2004), the mean difference between pre- and post-SRT scores was only -2.3 s. However the standard deviation in the mean difference was 10.8 s. Table 1 shows the frequency of symptoms. The rows in the table show the number of participants in each of four categories of SSQ symptom severity. SSQ scores greater than 10 generally indicate moderate discomfort and scores of 20 or more indicate relatively severe discomfort. The columns in the table indicate Sharpened Rhomberg performance. The column label UTP includes counts of participants who were unable to perform the SRT, i.e., were unable to maintain the SRT posture for 5 s. The column labeled improved includes persons who were able to maintain the posture longer after driving the simulator than before. Note that 10 moderately sick and 6 severe sick individuals showed an improvement in SRT after they reported simulator sickness. If the postural instability theory is correct (Stoffregen & Smart, 1998), and the SRT valid, then there should be no persons in these cells. There were only 5 participants who showed both a marked increase in symptoms by SSQ and a marked degradation is stability by SRT.

Table 1 . The number of participants who experienced simulator sickness symptoms by SSQ measure as a function of Sharpened Rhomberg Score.

¹ Higher numbers indicate more symptoms and/or more severe symptoms.
² Negative numbers indicate degradation of postural stability, i.e., the reduction in the average number of seconds that the SRT posture was maintained.

The data of Graybiel and Fregly (1966) clearly suggest a large practice effect when the SRT is administered over a series of days. Data from the HDS do not show clear evidence of a practice effect, perhaps because six trials that are administered on the same day is insufficient exposure to show a practice effect. The mean decrease in performance of 2.3 s from pre- to post-test, is not statistically reliable, and is in the opposite direction from that to be expected if practice were a factor.

Inability to Perform the SRT

Earlier HDS experiments clearly demonstrate that the SRT is not an appropriate measure for a large proportion of the FHWA participant population. In the Intersection Collision Avoidance Study reported by Davis and Inman (2004) and Inman, Davis, El-Shawarby, and Rakha (2006), 34 percent of 214 participants who completed the pre- and post- SSQ questionnaires were unable to perform the SRT. The mean age of the participants who could not perform the SRT was 69, and the mean age of those who were able perform the SRT was 53. Thus, not only is the SRT not available to a large proportion of the participant population, it is disproportionately not available to older adults.

Sway Magnetometry

Measurement of postural sway has been proposed as an alternative to use of the SRT and other measures that rely on dichotomous measures of success or failure in balance maintenance. Sway measures assess the size, speed, and/or frequency of corrections to maintain a particular posture rather than assessing how long a posture can be maintained. For instance, Kennedy, Lanham, Drexler, and Lilienthal (1995) used video to record the amount of hip movement while subjects maintained the Sharpened Rhomberg stance. Clinical facilities often use force platforms that measure shifts in mass while subjects stand upright (e.g., FitzGerald, Murray, Elliott, & Birchall, 1993; Fitzgerald, Murray, Elliott, & Birchall, 1994; Raymakers, Samson, & Verhaar, 2005). Subjects assessed on force platforms are typically instructed to stand still with bare feet apart for 30 to 60 s. The length and velocity of shifts in mass are then measured using strain gauges at the corners of the platform. These measures can be compared against clinical norms or, in the case of simulator sickness assessments, pre- and post- measures.

More recently, sway magnetometry has been used to assess postural stability. Otolaryngologists have used sway magnetometry instead of force platforms (e.g., Fitzgerald, Murray, Elliott, & Birchall, 1994; FitzGerald, Birchall, & Murray, 1997; Elliott & Murray, 1998). Sway magnetometry has also been used it to assess motion and simulator sickness (e.g., Cobb & Nichols, 1998; Stoffregen & Smart, 1998; Cobb, 1999; Stoffregen, Hettinger, Haas, Roe, & Smart, 2000; Smart, Stoffregen, & Bardy, 2002).

In sway magnetometry, a small transceiver (less than 1 oz.) is attached to the participant, and a second transceiver is located a few feet away. Changes in magnetic flux in three orthogonal planes are then measured. Some researchers attach the transceiver to the participant’s waist (e.g., Cobb & Nichols, 1998; Cobb, 1999) whereas others mount it to the head (e.g., Stoffregen & Smart, 1998; (Stoffregen, Hettinger, Haas, Roe, & Smart, 2000; Smart, Stoffregen, & Bardy, 2002). Both methods seem to result in reliable measures that are sensitive to exposure to sickness inducing visual stimulation. Smart, Stofferegen, and Bardy (2002) used a head mounted transceiver on participants seated in a flight simulator and reported that changes in sway preceded the onset of reported simulator sickness symptoms. That study had only 13 participants, of which 6 reported symptoms. To our knowledge the results have not been replicated and prospective identification of participants who later report symptoms has not been demonstrated. Nonetheless, the study raises the tantalizing possibility that, in the future, the antecedent of simulator sickness may be detected and mitigated before symptoms reach a perceivable level.

Cobb and Nichols (1998) and Cobb (1999) had participants use a static standing posture and then measured sway at the hips. This variation on the sway magnetometry technique is more common than head mounted measurement, and some normative data across adult ages is available (Elliott & Murray, 1998).

Sway magnetometry measures that have been reported in the literature include:

• Area of sway.
• Anterior-posterior distance traveled.
• Anterior-posterior distance/unit time.
• Lateral distance.
• Lateral distance/unit time.
• Any of the above measured with eyes open divided by eyes closed (referred to as the Rhomberg Coefficient).

Because all of these measures are computed after the data are obtained, selection of a particular measure is not critical. The Rhomberg Coefficient has not been reported as sensitive to simulator induced changes, whereas all the other measures have been reported to be sensitive to simulator induced changes.

Conclusion

The FHWA RD&T Human Centered Systems group has been using two tests to assess the severity of driving simulator induced sickness: the SSQ and the SRT. The SSQ appears to be a sensitive instrument for detecting and measuring participant discomfort. However it is not sensitive to postural instability, which sometimes accompanies simulator sickness and can occur in the absence of other symptoms. The SRT is used to detect disequilibrium. Detection of disequilibrium is important to avoid exposure of participants to increased risk when driving or in other situations where loss of postural stability could result in injury.

Unfortunately, a substantial proportion of our participants cannot perform the SRT, so for them it offers no protection from simulator induced degradation in stability. Furthermore, our experience to date does not suggest that the SRT offers protection to participants who can perform the test. Means before and after testing have been nearly identical, and do not differ between subjects who report sickness symptoms and those who do not. Either the HDS does not induce degraded stability, or the SRT is insensitive to degradation. We cannot know which of these alternatives the case is, but the literature suggests that it could well be the latter.

Therefore, we propose to replace the SRT with sway magnetometry. Sway magnetometry has been shown to detect changes in immediately following exposure to virtual environments (Cobb & Nichols, 1998). These changes in postural stability were not correlated with SSQ scores or with a separate postural stability questionnaire. Thus, sway magnetometry may detect symptoms of which participants would otherwise be unaware. In the Cobb and Nichols study, changes in sway dissipated within 10 minutes of the end of the simulation.

It is unclear whether the HDS may cause dangerous changes in postural stability. It is clear that the SSQ is not sensitive to changes in postural stability, so that if such changes are induced, another test is needed to measure them. At this time, sway magnetometry appears to be the best available alternative to assess HCS participants for such changes. The equipment needed is affordable (about $2500). The test can be done in about 2 minutes, and all participants should be able to perform it.

Although norms are available for sway magnetometry, criteria for screening participants for clinically significant pre-post changes are not. Furthermore, norms have not been identified for individuals over 65 years of age; a population that for many HCS studies is the source of half of the participants. Thus, initial use of the sway magnetometry will involve some calibration and “clinical judgment” to determine when participants should be encouraged to wait while symptoms of disequilibrium subside. The available norms do not suggest when pre-test measures are clinically outside the norm. That is, particular results can be roughly estimated to correspond to that of a particular percentile of the population, but there is no guidance as to what percentile score represents a threshold of increased risk.

Stoffregen et al. (2000) reported that a pre-exposure measure of sway in the pitch and roll axes of head movement was predictive of which participants became sick in a flight simulation. Of 14 participants in that study, 6 became sick. The pre-exposure sway measures accounted for 32 percent of the variability in sick/well classification. Because this study was not prospective in prediction and was small, sway cannot be said to be a proven predictor of who might become sick in a simulator. However, this study does suggest that with experience, a predictive algorithm may be possible.

Recommendations

The SRT should be dropped from the HCS simulator sickness protocol, and Sway magnetometry should be added. A system that attaches a sensor about the waist at the small of the back is recommended over head mounted measurement because norms are available for waist-based measures. Testing should be done before and after exposure to the HDS. Both before and after readings should be obtained first with eyes open, feet shoulder width apart, arms folded, and head oriented level and straight ahead. Participant should be instructed to stand as still as possible in this posture. After 10 s in this posture, measurement should commence and continue for 30 s. This procedure should then be repeated with eyes closed.

Although eyes open divided by eyes closed scores have not been shown to be suggestive of postural instability following simulation exposure, the two scores averaged together can be expected to be more reliable than either alone. Furthermore, the eyes open divided by eyes closed (Rhomberg Coefficient) may be useful in the future, in evaluating risk, especially for older adults. It has been useful in clinical evaluations of balance disorders, so there is some reason to hope that it may – after sufficient data are available – be useful in prospective identification of participants who may be at elevated risk of simulator sickness.

A small (e.g., n ≈10) pilot study should be conducted to establish a normal range of scores for the HCS magnetometer system and for comparisons to published norms. Following this normative calibration, the test should become an integral part of the simulator sickness protocol. Initially the mean of the eyes open and closed sway scores from the before measures should be divided by the after means. Whenever this score is less than 0.5, the participant should be considered to have transient instability, and should be treated in the same way as participants with significant SSQ symptoms. The exception should be when the post-test stability score is not elevated beyond the 75 th percentile norm. This could occur if the participant had an exceptionally low pre-test stability score.

Initially, the results from the HCS magnetometry system will be compared to the norms reported by Elliott and Murray (1998). These investigators reported path length in lateral and anterior-posterior planes and area. Other measures have been reported in other studies, e.g., mean path-length/unit time, roll, or pitch; and one these may ultimately prove more sensitive. Thus during the first year of use, extensive post-experiment analyses of the magnetometry data should be conducted to identify which measures might provide the best correlation with observed and subjective symptoms and provide the most stabile and repeatable measures of stability.