Changes in standing stability with balance‐based torso‐weighting with cerebellar ataxia: A pilot study

Abstract Objectives People with cerebellar ataxia have few options to improve the standing stability they need for function. Strategic placement of light weights on the torso using the balance‐based torso‐weighting (BBTW) method has improved stability and reduced falls in people with multiple sclerosis, but has not been tested in cerebellar ataxia. We examined whether torso‐weighting increased standing stability and/or functional movement in people with cerebellar ataxia. Methods Ten people with cerebellar ataxia and 10 matched controls participated in this single‐session quasi‐experimental pilot study. People with ataxia performed the Scale for the Assessment and Rating of Ataxia (SARA) prior to clinical testing. All participants donned inertial sensors that recorded postural sway; stopwatches recorded duration for standing and mobility tasks. All participants stood for up to 30 s on firm and foam surfaces with eyes open then eyes closed, and performed the Timed Up and Go (TUG) test. Light weights (0.57–1.25 kg) were strategically applied to a vest‐like garment. Paired t tests compared within‐group differences with and without BBTW weights. Independent t tests assessed differences from controls. All t tests were one‐tailed with alpha set at .05. Results Duration of standing for people with ataxia was significantly longer with weighting (p = .004); all controls stood for the maximum time of 120 s with and without weights. More severe ataxia according to SARA was moderately correlated with greater improvement in standing duration with BBTW (Pearson r = .54). Tasks with more sensory challenges (eyes closed, standing on firm surface) showed less body sway with weighting. Duration for the TUG was unchanged by torso‐weighting in people with ataxia. Conclusion Strategic weighting improved standing stability but not movement speed in people with ataxia. BBTW has potential for improving stability and response to challenging sensory conditions in this population. Future studies should further examine gait stability measures along with movement speed.


| INTRODUCTION
In people with cerebellar ataxia (CA), dyscoordinated movement can result in postural instability, falls, and associated injuries (Rüb et al., 2008). Controlled movement depends on a person's ability to perceive self-motion and coordinate muscle activity within and between body parts to perform the desired task efficiently (Marsden, 2018). The normally functioning cerebellum plays a key role in motor control by comparing an efferent copy of the motor command with sensory information resulting from truncal, limb, and ocular movement (Ioffe, Chernikova, & Ustinova, 2007;Marsden, 2018). When the cerebellum degenerates, as in CA, damaged neurons and their connections with the brainstem areas, contralateral motor cortex, or vestibular nuclei can impair the cerebellum's ability to integrate sensory input sufficiently to monitor and correct motor commands and their consequences (Marsden, 2018). The result can include increases in postural sway, slowed or hypermetric responses to balance perturbations, and irregular stepping during gait (Marsden, 2018;Stephen et al., 2019). Few treatment options have been identified that address instability and fall risk in this population. No drugs or medications have been approved by the U. S. Food and Drug Administration for the treatment of CA or the motor features associated with ataxia (Ilg et al., 2014;Sarva & Shanker, 2014;Stephen et al., 2019). Evidence of the effectiveness of rehabilitation is not conclusive, with partial response to physical therapy for some patients (Fonteyn et al., 2014;Milne, Corben, Georgiou-Karistianis, Delatycki, & Yiu, 2017). Researchers state that evidence-based guidelines for the rehabilitation of people with CA need to be developed (Ilg et al., 2014). The purpose of this pilot study was to determine if a single session of strategic torso-weighting (Gibson-Horn, 2008) changed immediate standing stability and/or functional movement in people with CA.
BBTW involves assessment of a standing individual's directional instability while a trained clinician manually applies perturbations and resisted rotations at the shoulders and pelvis. The clinician strategically places light weights (objects with designated mass) on the trunk to counter instability. The mechanism for the effectiveness of BBTW is not specifically known. Prior work has disproven a strictly biomechanical mechanism (Crittendon et al., 2014) and argued against alternative hypotheses such as joint compression or increased conscious awareness proposed when heavier weights are used (Hunt, Widener, & Allen, 2014).
The mechanism likely involves enhanced sensory input to improve perception of self-motion and position (Crittendon et al., 2014). The theoretical rationale for proposing BBTW for people with CA is that a stronger sensory signal may help compensate for sensory integration deficits resulting in ataxia and postural control issues in this population.
Our study focused on scientific feasibility according to Thabane et al.'s (2010) recommendations for initial pilot testing (Thabane et al., 2010), examining the possibility of response to the intervention in people with CA. We hypothesized that participants would show improvement in standing stability and/or functional movement compared with a no-weight condition. We also tested age-and sex-matched controls to validate management feasibility, that is, the ability of the protocol to result in usable data (Thabane et al., 2010).

| METHODS
This was a quasi-experimental pilot study examining postural and mobility responses with and without BBTW in a single session. This study was approved by the appropriate university ethics committees.
All participants provided written informed consent.  Using standardized terminology and recommendations for pilot studies (Thabane et al., 2010), we focused on scientific and management feasibility of BBTW in people with CA. Scientific feasibility was determined by a significant response of participants to weighting.
Clinical testing was performed with no weight first (in prior studies, some participants maintained stability changes for over an hour after weights were removed). Management feasibility (Thabane et al., 2010) examined measures new to this research (inertial sensors to assess body sway) by testing age-and sex-matched controls and confirming differences in performance between those with and without movement disorders observed in prior studies (Crittendon et al., 2014;Gorgas et al., 2015).
The outcome measures used in this study were the modified Clinical Test of Sensory Interaction on Balance (mCTSIB; Cohen, Blatchly, & Gombash, 1993;Wrisley & Whitney, 2004) and the Timed Up and Go (TUG; Podsiadlo & Richardson, 1991) We chose these measures to test participants' performance in both static and dynamic balance tasks. For standing stability (mCTSIB), participants stood on firm and foam surfaces, with eyes open and eyes closed for 30 s each (or for shorter time if the person moved their feet out of position or needed assistance to prevent falling). Each of the four standing stability tasks was performed once. For dynamic stability (TUG test), participants stood up from a chair, walked 3 m, turned around, walked back, and sat down in the chair again (Podsiadlo & Richardson, 1991). The TUG was performed twice. The first trial was for practice, and the second trial was recorded. Rest breaks were provided as needed between trials. The participants performed all testing tasks without weights before undergoing the BBTW procedure. After receiving BBTW weights, participants repeated the standing stability and TUG tasks.

| BBTW procedure
To control for possible inter-rater differences, one physical therapist (CGH) with over 30 years of clinical experience in balance and gait rehabilitation performed all BBTW procedures for this study. The therapist started balance assessment by observing the relative amount and direction of sway while the participant stood quietly (Gibson-Horn, 2008). The therapist then applied anterior, posterior, and lateral perturbations (nudges) at the shoulders and pelvis to observe the participant's response and direction of balance loss. Rotational forces were applied manually through the shoulders and pelvis to determine asymmetries in the participant's ability to maximally resist while maintaining upright balance. In prior studies, this therapist used a handheld dynamometer to standardize and confirm the reliability of perturbations and rotational forces applied during the BBTW procedure (Crittendon et al., 2014). Loss of balance during perturbations and rotational forces was scored on a 0-4 scale developed to facilitate application of weights in the BBTW procedure (Allen et al., 2018).
Responses were scored: (0) No balance loss, brisk response to perturbation; (1) Minimal balance loss, delayed onset of return to upright, (2) Moderate balance loss, large trunk movement or parachute reaction with no foot movement; (3) Moderate-severe balance loss, large trunk movement with foot movement or takes a small step; (4) Severe balance loss, manual contact by the researcher required to prevent a fall. The scale has shown good to excellent inter-rater agreement (Allen et al., 2018) indicating that other therapists can observe balance loss similarly.
Weights (0.06, 0.11, or 0.23 kg) were placed using Velcro on a size-adjustable vest-like garment (BalanceWear, Motion Therapeutics, Oxnard, CA). Weight location was customized to counter the individual's direction of balance loss, asymmetry of resistance, and latency of response to perturbations. For example, if a participant showed a 3 on the balance loss scale when attempting to resist pelvic rotation to the right, a 0.23 kg weight affixed to the left low back region would assist the participant to maintain balance and resist this force. Balance was reassessed with weights in place to confirm that greater stability and/or quicker response were demonstrated. Better response was indicated by reduction of balance loss scores to 0 or 1 with reassessment. Weight amount and location were recorded. In addition, we recorded participants' tolerance of the intervention and any adverse events that occurred.

| Data analysis
Primary outcomes for standing stability were "duration of standing" recorded by stopwatch (starting once participants were in position and ready) and "body sway" as recorded by body-worn inertial sensors (APDM Wearable Technologies, Portland, OR), technology used effectively in other populations (Deshmukh, Russell, Lucarino, & Robinovitch, 2012;Spain et al., 2012). According to manufacturer suggestions, the sensors were placed at the lumbar spine, anterior sternum, bilateral ankles, and bilateral wrists (Horak, King, & Mancini, 2015), and recorded movement with an accelerometer, gyroscope, and magnetometer. Software combined information from individual sensors to record body sway as 95% of the ellipse sway area, defined as "the area of an ellipse covering 95% of the sway angle in both the coronal and sagittal planes" (Whitepaper, 2015). A lower number for sway area indicates greater stability.
The primary outcome for the TUG was time. The participant was seated with back touching the chairback: Time started when the participant was told to "go"; time stopped when the participant's back touched the chairback again after walking. Secondary TUG outcomes included turn duration, sit to stand duration, and peak turn velocity recorded by the inertial sensors (Salarian et al., 2010).
In addition to the primary outcome measures, participants completed the Activities-specific Balance Confidence (ABC) scale (Myers, Fletcher, Myers, & Sherk, 1998). A score of 80% or more suggests a high level of physical function or balance confidence (Myers et al., 1998). A score of less than 67% in elderly individuals is predictive of a future fall (Lajoie & Gallagher, 2004). All participants self-reported the number of falls they had experienced in the previous 6 months. Participants with ataxia also underwent testing using the Scale for the To describe the sample, we recorded mean and standard deviation for age, ABC scale, and performance on the SARA. We calculated associations between measures using Pearson r correlations. To compare performance results, we used paired t tests within groups (no weight compared with BBTW weighted condition) and independent t tests between groups. Alpha was set at .05, using one-tailed tests because the research hypotheses were unidirectional. Comparisons and correlations were performed using Excel 2013; effect sizes were calculated using equations for effect size d for unpaired and paired t tests (Portney & Watkins, 2009) and comprehensive meta-analysis software (Borenstein, Hedges, Higgins, & Rothstein, 2005).

| RESULTS
Eleven people with ataxia participated. One participant (A05) did not meet resting blood pressure criteria, and further testing for this participant was discontinued. Included participants (Table 1: 6 female, 4 male) had mean (SD [standard deviation]) age of 47.2 years (6.6), SARA score of 12.5 (4.16), and ABC score of 54.6 (16.8). The correlation (r) between age and ABC score was −0.75: Younger people rated themselves as more confident. The correlation between age and total SARA scores was −0.23: Age did not significantly correlate with dyscoordinated movement. In the previous 6 months, participants with ataxia reported a mean (SD) of 15.1 (27.4) falls per person, a number that did not correlate highly with age (r = .18). For the weighting condition, participants were weighted with a total ranging from 0.68 to 1.23 kg (equivalent to 0.7% to 1.4% of body mass).
Ten gender-and age-matched control participants were recruited.
The mean (SD) age was 47.8 (8.8), and ABC score was 97.5 (1.54). Participants without ataxia had significantly higher ABC scores when compared with participants with ataxia (p < .001). In the previous 6 months, controls reported a mean (SD) of 0.33 (0.7) falls per person. For the weighting condition, controls were weighted with a total ranging from 0.11 to.86 kg (equivalent to.1% to 1% of body mass). All participants in both groups tolerated the weighting intervention with no adverse events.

| Standing stability
All control participants stood for the full 30 s during each task and condition with and without weighting (ceiling effect). • Eyes open on firm surface had "no" sensory modalities constrained: All participants with ataxia stood for the full 30 s without and with weights; • Eyes open on foam surface (affecting the somatosensory system) and eyes closed on firm surface (affecting the visual system) each had "one" modality constrained: Participants with ataxia increased standing duration with weights, within group p = .03;

T A B L E 1 Participants with ataxia
• Eyes closed on foam surface constrained "two" modalities, including both the somatosensory and visual systems: Participants with ataxia improved with weights, within group p = .02.
Controls showed no significant difference in sway area between no weight and BBTW weighted conditions for any standing tasks ( • eyes closed on foam surfaces, no (p = .48).
Technical difficulties with recordings from the inertial sensors prohibited retrieval of sway area data for some trials for the participants with

| Timed Up and Go
In controls, time decreased from no weight to BBTW weighted trials in overall TUG duration (within group p < .001, Figure 3, effect size , but not in turn duration (p = .14, Figure 3), sit to stand duration (p = .47), or peak turn velocity (p = .32; Table 2). Participants with ataxia had significantly longer TUG durations compared with the controls (between group p < .001; effect size d = 2.5, 95% CI [1.33, 3.67]). Participants with ataxia did not change with weighting in TUG duration (within group p = .43, Figure 4), turn duration (p = .27, Figure 4), sit to stand duration (p = .33), or peak turn velocity (p = .42). Technical difficulties with inertial sensors prohibited retrieval of TUG turn and sit to stand data for control participant H01.

| DISCUSSION
Ten participants with ataxia and 10 age-and sex-matched controls participated in this study. All but two participants with ataxia had ABC scores below 67% indicating increased fall risk (Lajoie & Gallagher, 2004), whereas controls averaged 97% indicating high balance confidence. Reported falls experienced in the last 6 months also differed: averaging .33 per person for controls versus 15.1 for those with ataxia. Participants with ataxia scored an average of 12.5 on the SARA indicating they had mild to moderate levels of ataxia.
F I G U R E 2 95% Ellipse away area for participants with ataxia during a) eyes closed on firm surface, n = 8, within-group differences in unweighted and BBTW conditions, p = .020; and b) eyes open on foam, n = 9, within-group differences in unweighted and BBTW conditions, p = .02. Sway area generally decreased with torso weighting People with ataxia showed within-group standing stability improvements with BBTW, as hypothesized. However, TUG duration did not change when this cohort wore BBTW weights. The measures used in this study confirmed between-group differences: Controls showed better standing stability and TUG times compared with people with ataxia. Also, all controls scored above 80% on the ABC indicating a low risk of falls (Myers et al., 1998).  Figure 1). With two modalities constrained (eyes closed on foam surface), participants with ataxia also improved with BBTW, but standing duration did not reach 30 s. BBTW during this single session seems to suffice as compensation for a single-modality constraint (vision or somatosensation), but was insufficient to completely compensate when two sensory modalities were constrained in this small sample.
Improvements in sway area for people with ataxia also varied based on sensory constraints provided in different standing tasks (see Table 2). Weighting resulted in a significantly smaller sway area when one sensory modality was constrained (p = .03) although the resultant sway area did not match that of controls. Participants with ataxia showed no improvements with zero constraints (eyes open on a firm surface) or two constraints (eyes closed on a foam surface).
The magnitude of the sway area was small with both no weights and suggested that decreased gait speed may in itself be a compensatory strategy for the high variability of stepping in CA (Marsden, 2018).
Other researchers applying proprioceptive input have noted changes in postural control and stepping variables in people with CA but no significant changes in gait speed (Leonardi et al., 2017). Despite lack of speed improvements, other changes may have occurred for our participants: One participant with ataxia with slower TUG performance volunteered that with BBTW, "It was much easier to stand up from the chair." Further research investigating the kinetic and kinematics of movement may reveal additional changes in coordination while weighted.
Unlike the participants with ataxia, 10 out of 10 controls decreased TUG duration with weighting, likely a reflection of increased straight-course gait speed because duration of turns and transfers did not change. These results correspond to improved gait velocity with BBTW in people with MS noted in a previous study (Gorgas et al., 2015). Controls did not change standing stability with weighting because pretest values were already at the ceiling of these measures. The sample sizes were small and technical difficulties with the inertial sensors restricted sample sizes further. In addition, large standard deviations indicating great amounts of variability among participants also contributed to large confidence intervals. Future studies with larger samples will likely increase the precision of the estimated effect sizes. Despite these challenges, the power was 79% to over 90% for the measures that were significant in this study. Modifications to the protocol to ensure proper functioning of the equipment (e.g., inertial sensors) are recommended prior to determining sample size needed for a larger study.

| Study limitations
In our single-session design, we had no data regarding possible effects of BBTW on fall incidence. Future studies with longer term use of BBTW could prospectively record differences in fall rate.
Blinding of assessors could help minimize possible bias. Generalization of these results to clinical use should take into account that using and applying the BBTW method requires a training course consisting of 2 days in person and 2-4 hours on-line instruction.

| Clinical implications and future research
The results from this study provide evidence of the scientific feasibility of BBTW in this population. Sway magnitude for people with ataxia was considerably more than the sway of controls even when weighted. We contend that reduction in sway was clinically meaningful because participants with ataxia were able to stand statically for longer durations with BBTW (p = .004). One specific concern in CA is the potential for a worse ataxia or tremor when the weights are removed (Marsden, 2018). No signs of worsening or other adverse events were noted when these light weights were removed in this study; clinical experience has shown that other populations typically retain some benefit from BBTW for minutes to hours following removal.
Evidence of management feasibility was mixed. Technical difficulties with recording from the inertial sensors hindered extraction of usable data especially during the performance of the TUG. However, these measures might show improved management feasibility with appropriate protocol modifications. For example, in order for the inertial sensors to collect gait velocity during the TUG, the walk distance must be extended. However, despite the technical difficulties, the controls responded to BBTW as in prior studies, (Crittendon et al., 2014;Gorgas et al., 2015) supporting management feasibility of the protocol. Despite hindrances, this study validated hypothesized between-group differences.
Because standing stability improved with weighting but movement speed did not in people with ataxia, additional variables should be considered for greater scientific feasibility. People with ataxia may use slower movement as a coping strategy to accomplish tasks (Bastian, 1997), with the speed-accuracy trade-off formalized by Fitts' law (Fitts, 1954). Testing accuracy (such as step length, step width, and percent of gait cycle in single-and double-limb support during gait) could supplement tests of speed. Further, additional practice time with BBTW applied for long-term use may allow for participants to adjust strategies and may better reflect the effect BBTW has on both speed and accuracy in people with ataxia.

| CONCLUSIONS
This pilot study provides evidence of scientific feasibility of BBTW for improving standing stability in participants with ataxia. BBTW may be particularly useful in increasing stability in this population when one sensory modality is constrained. Between-group differences provide evidence of management feasibility. Modifications for future research in people with ataxia might examine measures of accuracy along with speed of functional movement.