Age associated differences in postural equilibrium control: A comparison between EQscore and minimum time to contact (TTCmin)

1. Introduction 

Human sensory-motor systems have evolved to optimize body movements and posture control in the terrestrial gravitational field. The body maintains its balance using sensory information from visual, vestibular, proprioceptive systems, and exteroreceptors. This afference provides the central nervous system (CNS) with near-real-time information that is used to assess the biomechanical state of the body and determine the appropriate direction and scaling for motor commands to correct biomechanical state errors [1], [2], [3].

Aging degrades these corrective responses and interactions between sensory-motor systems and is associated with greater susceptibility to physical disability and frailty. Indeed, many older adults exhibit poor balance control, which is an important contributor to falls [4]. Elderly individuals show increased body sway, less secure base-of-support, and greater dependence on sensory cues from vision [5], as vestibular and somatosensory systems decline in performance. Motor control also declines with age, as manifested by slower response times, lower accuracy of movement, and loss of muscle strength [6]. Assessment of balance control performance degradation with aging is critical for successful prevention of falls in the elderly.

The most commonly used balance control model suggests that under nominal conditions, the CNS selects a quasi-steady equilibrium point that places the center-of-gravity near the center of the base-of-support. Subject to threshold limitations of the receptors, and the individual's vigilance, the CNS continually analyzes the afferent sensory information to detect deviations of the center-of-mass position (COM) from this equilibrium point and adjusts the motor command signals to return the COM to the equilibrium point. This theoretical construct lends itself well to linear control theory modeling. Thus, COM sway is generally modeled as being symmetric about the equilibrium point, and postural stability is represented by the deviation of sway from the equilibrium point [7], [8]. Perhaps the most popular stability estimator is peak-to-peak sway, which quantifies the maximum deviations from the equilibrium point, and is thought to vary inversely with postural stability. One of the most widely used clinical posturography systems (Equitest, NeuroCom International, Clackamas, OR) measures peak-to-peak sway over independent 20s sensory organization test (SOT) trials to compute its measure of balance control performance, the equilibrium score (EQscore) [9].

An alternative balance control model was suggested by Koozekanani et al. [10], who pointed out the objective function for posture control might be best represented by a stability margin, which is not the deviation of sway away from an equilibrium point, but rather the deviation of sway toward a stability limit. This insightful recommendation has been largely ignored, likely because when one assumes an equilibrium point centered in the base-of-support and a fixed stability limit, any sway excursions away from the equilibrium point are inversely proportional to the stability margin. However, sway excursions of the COM are rarely symmetric about the equilibrium point, so the relative stability is likely to be underestimated by peak-to-peak sway. Furthermore, impending postural instabilities cannot be predicted by position information alone, but requires velocity information. The relatively new approach of examining time to contact (TTC) with the stability boundary [11], [12] may therefore improve the accuracy of the relative stability estimate by adding a velocity term to the position estimate proposed by Koozekanani et al. [10]. Postural TTC has been computed relative to the entire composite boundary defined by the feet [13], separate anterior–posterior and medial–lateral boundaries [14], [15], and functional boundaries [13]. Owing to the velocity term, TTC may better differentiate between movements that are threatening and non-threatening to future postural stability [16], and may better reflect the method used by the CNS to control postural stability [17]. Patton et al. [16] identified the minimum TTC (TTCmin) within a trial as being an important quantitative measure for assessing postural stability and compensatory strategies. TTCmin represents the least stable posture over the trial, and from that perspective is analogous to peak-to-peak sway.

The aim of this study was to compare TTCmin against EQscore in a putatively normal subject population of widely varying age performing conventional SOTs. It was expected that this population would afford comparison of EQscore and TTC over a wide range of normal stability variation, and thereby allow a broad comparison, as well as an assessment of age-related changes in balance control performance estimates.

2. Methods 

This study was conducted with 155 subjects (69 males, 86 females) recruited from the Baltimore Longitudinal Study of Aging. All subjects gave written consent to participate in this Johns Hopkins Bayview IRB approved study after all procedures and risks had been explained to them. Table 1 provides the demographic representation for each decade. Subjects selected were generally healthy, and reported no history of persistent vestibular problems, no diagnosis of benign positional vertigo, Meniere's disease, or other medical conditions that could affect the vestibular system. Further exclusion criteria required no history of neurological diagnoses including stroke, Parkinson's disease, polyneuropathy, muscle disease, or taking medications that affect the vestibular system.

Table 1. Demographics of the sample representing each age decade

		20–29	30–39	40–49	50–59	60–69	70–79	>80 years
	Subjects	10 (4 ♂, 6 ♀)	17 (5 ♂, 12 ♀)	28 (12 ♂, 16 ♀)	33 (12 ♂, 21 ♀)	23 (15 ♂, 8 ♀)	24 (13 ♂, 11 ♀)	19 (8 ♂, 12 ♀)
	Mean height (cm)	169.2±12.2	167.8±8.0	168.4±9.0	166.5±8.4	169.7±7.1	169.2±9.3	161.5±8.2
	Mean weight (kg)	69.4±12.7	71.2±15.4	81.5±14.0	77.9±13.2	83.3±16.1	81.1±13.3	68.5±10.5
	BMI	24.1±1.8	25.3±4.6	28.8±4.8	27.4±4.7	28.8±4.6	28.4±4.2	26.2±2.8

Balance control was evaluated using a computerized dynamic posturography system (Equitest, NeuroCom International). Subjects performed six sensory organization tests with SOT 1 (Romberg, eyes open) considered the simplest task to SOT 6 (sway-referenced support and vision) considered the most challenging. The remaining SOTs are defined as SOT 2 (Romberg, eyes closed), SOT 3 (sway-referenced vision only), SOT 4 (sway-referenced support only), and SOT 5 (absent vision and sway-referenced support). Sway referencing was achieved by rotating the force platform, and/or visual surround within the sagittal plane in direct proportion to the estimated instantaneous COM sway angle. Throughout each 20s trial, subjects were instructed to maintain a stable, naturally upright posture with arms folded across the chest. External auditory orientation cues were masked by white noise supplied through headphones. Instructions to the subjects were also provided through the headphones, including test-condition specifics (e.g. eyes open or eyes closed), and identifying rest periods.

Testing took place in two separate sessions on 2 successive days. This allowed us to test the reliability of our protocol, a statistic that has so far rarely been available. At each session, the subjects performed three trials for each of the six SOT conditions. The order of these 18 trials was block-randomized. Prior to the first test session, foot measurements were taken with the Brannock Device (The Brannock Device Co. Inc., Liverpool, NY). Additional length measurements were made from the ankle to the back of the heel and the ankle to the tip of the toe to estimate the stability boundary, the A–P limits of the base of support about the rotational axis of the platform. The instantaneous COM location was then computed by low-pass filtering the center-of-pressure (COP) (second order low pass Butterworth filter with cutoff frequency 0.85Hz) waveform derived from the force transducers within the standing surface as provided by NeuroCom International. The primary dependent measures were derived from the COM. In particular, peak-to-peak COM sway amplitudes were used to calculate equilibrium scores for each condition. TTC was calculated from the COM directional velocity, a first derivative of COM positional data, and the distance between COM and the stability boundary toward which it was moving. Dividing the instantaneous distance of the COM to the boundary by its closing velocity gave us an instantaneous TTC value. The TTCmin was the minimum value obtained for instantaneous TTC within a trial. Fig. 1 illustrates the relationship of COM position to the change in instantaneous TTC for a representative subject during a single trial from SOT 5. This figure highlights the effect of the velocity component on the TTC.

View Large Image Download to PowerPoint

Fig. 1. The COM position (bold line) surrounded by the location of the stability margin (top and bottom bold lines) and the TTC (solid line) for one SOT 5 trial (TTC was truncated for all values that exceed 15s, displaying the most threatening values).

We examined the reliability of our testing paradigm for each SOT and stability measure estimating intraclass correlation coefficients (ICC) for an average of three trials. Also, zero order correlations were used to compare similarities in TTCmin and EQscore for each SOT condition and session across age groups, and repeated measures analysis was used to test differences between mean SOT values for both EQscore and TTCmin. These statistical comparisons were based on the mean of three trials in each session. Falls in any trial were assigned a value of zero for both EQ and TTCmin, thereby reducing the average for the session and indicating greater instability.

3. Results 

The aim of this study was to determine whether the EQscore and TTCmin represented different postural control qualities during a variety of SOT conditions of dynamic posturography. Both measures were moderately reliable, Table 2, and displayed a moderate correlation with each other for every SOT condition and session (0.59<Pearson's r<0.83).

Table 2. Intraclass correlation coeffients for TTCmin and EQscore for each SOT condition

		SOT 1	SOT 2	SOT 3	SOT 4	SOT 5	SOT 6
	TTCmin	0.71	0.66	0.72	0.69	0.67	0.67
	EQscore	0.72	0.65	0.51	0.86	0.72	0.71

Weight, height, BMI, and gender all failed to significantly contribute to the differences measured between SOT conditions. There was an improvement in both measures from the first to the second session for all SOTs except SOT 1 and 2 for EQscore (p<0.001) and SOT 2 for TTCmin (p<0.001) (Fig. 2). This learning effect was dependent on age for TTCmin, with greater improvement demonstrated by the younger subjects (p<0.05). Learning measured by the EQscore was unaffected by age.

View Large Image Download to PowerPoint

Fig. 2. (a) The mean TTCmin (+S.E.M.) separated into sessions and (b) the mean EQscore (+S.E.M.) separated into sessions.

As expected, a general decline in both the EQscore and the TTCmin was demonstrated with increased difficulty of the SOT. However, the TTCmin measure appears to detect greater stability threats in SOT 2 and SOT 5. Consequently, the TTCmin identifies eyes closed tests (SOT 2 and 5) as less stable conditions than the EQscore reflects (Fig. 2).

For both measures, across all SOTs there was also a general decline in performance with increasing age (p<0.001). The same pattern of response in the EQscore was reflected in SOTs 1–3; however, significantly lower EQscore with increasing age was demonstrated for SOTs 4–6 (p<0.05)(Fig. 3). The TTCmin demonstrated the reverse trend, where the least complex tests, SOT 1, 3, and 4, were variable with age, showing greater decrease in TTCmin as age increased (p<0.001) (Fig. 4).

View Large Image Download to PowerPoint

Fig. 3. The EQscore regression lines with age for each SOT condition for session A (−0.47<Pearson's r<−0.36).

View Large Image Download to PowerPoint

Fig. 4. The TTCmin regression lines with age for each SOT condition for session A (−0.53<Pearson's r<−0.31).

Falls were particularly prevalent in the more difficult test conditions, SOT 5 and 6, and were also more prevalent in older subjects (Fig. 5). Those who fell in SOT 5 and 6 after gaining experience with the conditions also scored less in SOT 1–4 for both the EQscore and TTCmin (p<0.001) (Fig. 6).

View Large Image Download to PowerPoint

Fig. 5. Number of falls in each SOT condition as a function of age.

View Large Image Download to PowerPoint

Fig. 6. (a) The mean TTCmin (+S.E.M.) for SOT 1–4 of non-fallers and fallers in SOT 5 and 6 (p<0.001) and (b) the mean EQscore (+S.E.M.) for SOT 1–4 of non-fallers and fallers in SOT 5 and 6 (p<0.001).

4. Discussion 

Data from computerized dynamic posturography tests performed by a putatively normal population of community-dwelling adults ranging in age from the third to ninth decade were used to examine similarities and differences between two different techniques for estimating postural stability. EQscore, which is based on COM position, and TTCmin, which is based on COM position and velocity, followed similar trends with increasing SOT condition and session, but differences were observed in their sensitivities to eyes closed SOTs and age-related deficits. These differences may result from an insensitivity of the EQscore to variations in postural sway velocity information, which is clearly available to the CNS and likely used in controlling postural stability.

When compared to younger individuals, older subjects showed modest declines in EQscore for SOTs 1–3, but far more prominent declines in TTCmin. As a result, discriminating stability information was gained from the easier SOT conditions (eliminating or sway referencing 0 or 1 sensory system) for TTCmin, but required harder SOT conditions (eliminating or sway referencing 2 sensory systems) for EQscore. Therefore, the use of TTCmin may allow patients, particularly frail ones, to avoid the stresses of performing more complicated SOTs and possibly experiencing a fall during testing. In addition, independence is gained from requiring the more complex equipment to perform the higher number of SOTs, facilitating a ubiquitous diagnostic tool through the use of force plates.

Furthermore, in terms of a possible diagnostic tool, the primary benefit of the TTCmin measure is an enhanced sensitivity to velocity [18] and its potential effects on postural instability. For example, TTCmin appeared to be more sensitive than EQscore to stability threats under eyes closed conditions, SOT 2 and 5, suggesting that sway position information alone is insufficient to fully characterize stability under these conditions (Fig. 3, Fig. 4). Van Wegen et al. [15] previously demonstrated decreased TTCmin and lack of sensitivity from COP position when vision was eliminated. The enhanced sensitivity of the TTCmin measure suggests it may be a better diagnostic tool than the EQscore, however further research on patients with known fall risks and pathology would be needed to truly compare the diagnostic value of these two measures.

Differences between TTCmin and EQscore were also found with increasing age. Both measures showed decreasing stability with increasing age across nearly all SOTs, but the age-related decrements varied between measures. As with previous studies [19], [20], the EQscore revealed greater age-related deficiencies for the more challenging SOT conditions (5 and 6) than for the simpler conditions. In contrast, TTCmin revealed greater age-related differences for the less challenging SOT conditions (1 and 3) than for the more complex conditions. This suggests that EQscore, by ignoring sway velocity components, may overestimate the stability of older subjects in the easier tests and the stability of younger subjects in the more difficult tests.

Across all SOT conditions older subjects had less variability in TTCmin values than younger subjects, suggesting that increased sway velocity may be more important for their daily functioning even under putatively non-threatening conditions. If older adults struggle to prevent their COM from quickly approaching a stability boundary during quiet-standing and quiet-standing with eyes closed, then they would be expected to have a greater susceptibility to falling during daily tasks. In fact, subjects who fell during testing (in SOT 5 and 6) after gaining experience with the conditions reached a lower TTCmin in SOTs 1–4 than non-fallers (Fig. 6). While the same was true for the EQscore, its range between fallers and non-fallers was such that practical differentiation of patients would prove difficult [20]. The TTCmin provided a greater practical range from which to identify potential stability deficiencies. We found TTCmin values of 7s or less indicate a potential sensory deficiency that contributes to instability (Fig. 6). A prediction of 7s to destabilization may not seem cause for concern considering response times are usually achieved in hundreds of milliseconds. However, rather than an absolute temporal measure, TTC reflects a relative scale of increasing instability danger. TTCmin values less than 7s do not necessarily reflect an instability danger of the moment, but may reflect a compromised postural control performance that would be unable to maintain stability in more difficult scenarios. Therefore, it may be possible to predict falls in more difficult scenarios from TTCmin values observed under less challenging conditions.

Although postural instability with aging is a complex issue, widely recognized as not purely being a function of age [21], but confounded with strength loss, activity levels, or lack of coordination, to name a few, TTCmin provides a useful way to examine specific postural insecurities relative to the stability-threat problem. Previously, the lower TTCmin values recorded for older adults have been attributed to an increased ‘leniency’ in the control system where corrective actions are delayed, rather than performed incorrectly [15]. This delay could be a function of decreased sensory sensitivity to TTC with increasing age [15] and/or inappropriate perception of corrective measure margins [22]. A reduction in sensory sensitivity is supported by the trend of increasing TTCmin homogeneity across SOT conditions with age, as the sensory challenging conditions will only create a less stable environment if that manipulated input is used and relied on. Further support may be found from the reduced learning from one session to another with age, also demonstrated by Camicioli et al. [23] with EQscore across trials. The indication that the ability to adapt may be compromised with age suggests a lack of sensory sensitivity may impede the learning by older adults from one session to another. Therefore, TTCmin could help clinicians identify and treat individuals who allow a greater postural threat to develop and the magnitude of danger associated with that event.

Both TTCmin and EQscore were found to be moderately reliable measures, with the EQscore reliability exceeding that previously reported [24], perhaps owing to the higher N or wider age range used in the current study. A notable variability existed in EQscore ICCs across SOTs, ranging from 0.51 to 0.86, Ford-Smith et al. [24] also exhibited a similar phenomenon across SOTs, ICC ranging from 0.26 to 0.68, although the pattern for ICC and corresponding SOT was not the same. This may reflect one of the shortcomings of the SOT protocol, namely that the posture control may not be stationary, in the mathematical sense, over the testing intervals. Another shortcoming is that SOT results using the EQscore may not be correlated with activities of daily living [25]. Whether that correlation can be improved using TTCmin, remains to be tested.

One limitation of TTCmin is in the identification of the stability margin from which it is calculated. Postural TTC has been computed relative to the entire composite boundary defined by the feet [13], separate anterior–posterior and medial–lateral boundaries [14], [15], and functional boundaries [13]. The stability margin used in this study was defined by foot measures and height but only in the A–P direction. The inclusion of the medio–lateral sway could help to provide a more complete picture. Furthermore, natural human movements are typified by substantial flexibility enabled by an adaptive control system. Consequently, the boundary that defines stability is not a fixed margin, but rather a dynamic one that varies with factors such as foot position and locomotion. Moreover, the perceived location of the stability margin may not necessarily match the actual biomechanical stability margin, and can vary with diligence and/or the perception of stability risk or consequences. The definition established in this study may not be a perfect measure when considering that the level of control possible by an average person and a ballerina would not likely be equivalent values. To better estimate the true stability margin for each individual one benefit might be the inclusion of the dynamic and perceptual aspects of stability margins, which could further explain the role of TTCmin in postural control and, ultimately, determine the appropriate diagnostics and optimal interventions.

Based on our results it seems clear that TTCmin measures can offer increased information on the integrity and operation of the postural control system, and contribute to understanding the changes in balance with increasing age. We show that TTCmin can yield information not available solely through the observation of peak-to-peak sway amplitudes, as is assumed by the EQscore. Therefore, it appears that TTCmin may provide a more sensitive diagnostic measure than the EQscore for the assessment of balance under more natural sensory conditions.