Examination of Proprioceptive Reliance During Backward Walking in Individuals With Multiple Sclerosis

Abstract

Background:

Slowed somatosensory conduction in multiple sclerosis (MS) increases postural instability and decreases proprioception. Despite these delays, individuals with MS rely more on proprioception for balance compared to controls. This heightened reliance, combined with slowed signal transduction, increases fall risk. Backward walking (BW) increases proprioceptive reliance by reducing visual cues. However, no study has conclusively linked proprioception to BW. This study aims to assess proprioception’s role in BW compared to forward walking (FW) in MS and to compare differences in proprioception between MS fallers and non-fallers.

Methods:

Fifty participants (average age: 50.34 ± 11.84, median Patient Determined Disease Steps [PDDS]: 2) completed the study. Participants completed BW and FW at comfortable and fast speeds. We have previously established vibration sensation as a proxy measure for proprioception. Vibration thresholds were quantified at the great toe bilaterally using a 2-alternative forced-choice procedure.

Results:

Significant correlations were seen for vibration sensation and FW comfortable (ρ = 0.35), FW fast (ρ = 0.34), BW comfortable (ρ = 0.46), and BW fast (ρ = 0.46). After controlling for age, sex, and PDDS, vibration sensation significantly predicted performance during all walking tasks, with larger beta coefficients seen during BW (comfortable β = 0.57; fast β = 0.58) compared to FW (comfortable β = 0.41; fast β = 0.45). Fallers performed significantly worse than non-fallers for vibration sensation (P = 0.04).

Discussion and Conclusions:

Considering the notable decrease in proprioception in participants with MS and the clear distinction between fallers and non-fallers, it is crucial to conduct fall risk assessments and interventions focusing on proprioception. With its heightened reliance on proprioception, BW offers a promising method for assessing fall risk and could be an effective exercise intervention.

INTRODUCTION

Multiple sclerosis (MS) is an autoimmune neurological disorder characterized by extensive demyelination occurring within the central nervous system, leading to progressive functional decline and impaired mobility.¹ In fact, almost 90% of individuals with MS experience difficulties with walking, including those with mild levels of disability.^2,3 Moreover, individuals with MS report the loss of walking ability as the most consequential factor limiting societal participation and functional independence, reducing physical activity, and decreasing overall quality of life.^4,5

The integration of sensory feedback, including vestibular, visual, and proprioceptive inputs, constitutes essential components of postural control and balance.⁶ This integration is fundamental in maintaining stability during the constant interaction between a shifting base of support and the center of mass when walking.⁷ Approximately 84% of individuals with MS exhibit some form of sensory impairment,⁸ which may further contribute to mobility impairments in MS. Previous research in individuals with MS has indicated an elevated reliance on proprioception for balance maintenance.⁹ Previous studies have identified slowed somatosensory conduction in individuals with MS,¹⁰ which contributes to increased postural instability when compared to healthy controls.¹¹ Further, research has established a significant association between proprioception and ambulation, with decreased proprioception significantly correlating with worse performance on clinical tests, such as the timed up-and-go test, the 6-minute walk test, the Berg Balance Scale, and the timed 25-foot walk (T25FW).^12,13 Furthermore, difficulties with proprioception have been noted to result in delayed muscle onset during automatic postural responses within this population.¹⁴ Therefore, targeted mobility assessments that manipulate sensory processing may be critical for understanding the functioning of the sensorimotor system in individuals with MS.

While the majority of literature investigating the influence of sensory functioning in individuals with MS has focused on standing balance and forward walking (FW), there is a growing interest in research regarding backward walking (BW). BW, a sensory-conflicting mobility assessment due to the reduction of peripheral visual information, has been shown to entail heightened cognitive demands and is better at discriminating fall status in individuals with MS.^15-17 Furthermore, BW is recognized as a necessary walking adaptability task that holds ecological validity and is a significant component of daily life.¹⁸ For instance, everyday activities such as sweeping, vacuuming, or opening a door require the ability to walk backward. The manipulation of visual information has led to the proposal of an increased reliance on proprioception, primarily due to the inability to visually perceive the ground.¹⁹ Furthermore, previous literature has shown a relationship between proprioception and BW.²⁰ However, no study to date has objectively established this relationship in persons with MS.

Therefore, the primary objective of this study is to investigate the association between BW performance and proprioception (measured via vibration sensation) and compare it to FW. Furthermore, we aimed to examine whether differences in proprioception were present between fallers and non-fallers with MS. Additionally, we sought to investigate the contributions of proprioception to both BW and FW after adjusting for age, sex, and ambulation disability. Given that individuals with MS may rely more heavily on proprioceptive sensory feedback and that BW may induce increased reliance on the proprioceptive system,^10,21 we hypothesized that individuals with higher vibration sensation thresholds would exhibit poorer BW performance measures. Moreover, considering the absence of visual cues, we hypothesized that vibration sensation thresholds would demonstrate a stronger association with BW compared to FW, and that vibration sensation would contribute more to BW compared to FW after adjusting for covariates. Lastly, we hypothesized that non-fallers would exhibit significantly lower vibration thresholds compared to fallers with MS.

METHODS

All study procedures were approved by the Wayne State University Institutional Review Board, and all study participants signed an informed consent form prior to participation. Participants were eligible for the study if: (1) they were ≥18 years of age, (2) they had physician-diagnosed MS (all subtypes), (3) self-reported none to moderate ambulation disability as determined by the Patient Determined Disease Steps (PDDS ≤ 6),²² and (4) had no exacerbation of symptoms in the past 30 days. Individuals were excluded from the study if (1) they had an exacerbation of symptoms in the past month, (2) they had a neurological or orthopedic condition other than MS that might contribute to significant walking or balance impairment, or (3) they were unable to follow the study directions. Participation in the study involved a single cross-sectional testing session involving walking and sensory assessments.

Walking Assessments

Participants completed 2 trials of the T25FW test in each of the following trials: (1) FW-T25FW comfortable, (2) FW-T25FW fast-paced, (3) BW-T25FW comfortable, and (4) BW-T25FW fast-paced. For all comfortable walking trials, participants were instructed to walk at their self-selected pace. For all fast-paced trials, participants were instructed to walk as quickly and safely as possible. Consistent with previous literature, all trials were performed with participants beginning with their feet behind a starting line, with time starting when their first foot crossed the line, and the time stopped when their first foot crossed the finish line. Time to completion was the primary outcome for all walking trials, and the 2 trials for each walking condition were averaged together. Both the FW- and BW-T25FW assessments have been shown to be reliable and valid measures in the MS population.^23,24 For all trials, participants wore a gait belt and were accompanied by a research team member to ensure safety.

Vibration Sensation

Vibration thresholds, measured as vibration units (vu), were quantified at the great toe of each foot using the Vibratron II device (Physitemp). A higher vibration threshold, reflected by a larger vu, is reflective of decrements in plantar sensation. Using a 2-alternative forced-choice procedure,²⁵ subjects were instructed to identify which of 2 rods were vibrating. The vibration threshold measured from the less sensitive (worst) great toe was used for analyses. The Vibratron is a reliable and objective measure of proprioceptive sensation in individuals with MS.²⁶ Notably, both proprioception and vibration sensation are carried by the dorsal columns and are known to be impaired in individuals with MS.^26,27 Vibration sensation is an indicator of dorsal column health, and by extension, proprioception. Utilizing vibration thresholds as a proxy measure of proprioception is a reliable and objective measure of sensation in MS²¹ and can detect declines in sensation not currently measured by clinical rating scales like the Expanded Disability Status Scale.²⁸ In fact, prior research has recognized an association between vibration sensation and both balance and walking speed.^13,29 Consequently, the assessment of vibration sensation provides a means to quantify proprioception, offering unique insights into a deeper understanding of proprioceptive impairment and its impacts on balance, mobility, and sensory reweighting in individuals with MS.

Analysis

To investigate the relationship between walking trials (FW-T25FW comfortable, FW-T25FW fast-paced, BW-T25FW comfortable, BW-T25FW fast-paced) and vibration thresholds, Spearman’s rho (ρ) was used as data were non-normally distributed. The strength of association for the correlations was interpreted as small (0.1), medium (0.3), or large (0.5) association.³⁰ To compare the contributions of vibration sensation to walking performance, linear regressions were conducted with each walking trial (FW-T25FW comfortable, FW-T25FW fast-paced, BW-T25FW comfortable, BW-T25FW fast-paced) as a dependent variable and vibration threshold as the independent variable. Beta coefficients were then compared across models and age, sex, and PDDS were included as covariates. To investigate potential differences in vibration sensation between fallers and non-fallers, independent samples t tests or Mann-Whitney U test were conducted as appropriate. Following the criteria established in previous literature, participants were categorized as fallers if they self-reported experiencing 2 or more falls in the past year, while those reporting 1 or no falls were classified as non-fallers.^31,32 The analyses were conducted using SPSS version 29.0 (SPSS Inc), with a significance level set at α = 0.05. Given the sample size (50), number of predictors (4), power (0.80), and alpha level (0.05), we were able to detect an effect size of R² = 0.21 for all linear regression models.³³

RESULTS

Fifty participants were included in the study. Descriptive statistics can be seen in Table 1. Participants were 50.34 ± 11.84 years of age, and 82% of those included in the study were female. The majority of individuals in the study reported having relapsing-remitting MS (n = 44), with 3 reporting primary-progressive, 2 reporting secondary progressive, and 1 unknown. Walking impairment, as interpreted by PDDS, ranged from 0 to 6, with a median of 2, indicating minimal to moderate walking impairment in our sample.

Table 1.

Descriptive Statistics

Descriptive Statistics	Whole Group N = 50	Fallers N = 22	Non-Fallers N = 28	P Value
Age, y	50.34 ± 11.84	50.05 ± 13.28	50.57 ± 10.82	0.88
MS subtype, % RRMS	88%	77%	96%	0.13
Symptom duration, y	17.80 ± 13.16	20.45 ± 15.66	15.71 ± 10.64	0.23
Sex (% female)	82%	82%	82%	0.63
Race (% non-white)	58%	54%	64%	0.44
PDDS [median(range)]	2 (0-6)	3.36 ± 2.06	1.54 ± 1.67	<0.01
FW-T25FW comfortable, s	11.25 ± 9.38	14.72 ± 11.83	8.53 ± 5.78	0.02
FW-T25FW fast-paced, s	6.54 ± 7.11	11.15 ± 9.37	6.53 ± 3.69	0.04
BW-T25FW comfortable, s	20.82 ± 23.60	30.43 ± 31.40	13.26 ± 10.46	0.02
BW-T25FW fast-paced, s	16.92 ± 20.77	25.49 ± 27.99	10.18 ± 8.18	0.01
Vibration threshold (vu)	4.78 ± 3.37	5.90 ± 4.09	3.90 ± 2.40	0.04

Abbreviations: BW-T25FW, backward walking timed 25-foot walk test; FW-T25FW, forward walking timed 25-foot walk test; PDDS, Patient Determined Disease Steps; RRMS, relapsing-remitting MS.

After stratifying by retrospective falls, a total of 28 individuals with MS were identified as non-fallers, whereas 22 had experienced 2 or more falls in the past year. No differences were observed between groups for age (P = 0.88), MS subtype (P = 0.13) symptom duration (P = 0.23), sex (P = 0.63), or race (P = 0.44). However, significant differences were noted between groups for PDDS (P < 0.01), with the fallers group self-reporting significantly more ambulation disability compared to non-fallers. Significant differences were also observed between all walking trials in both the forward (comfortable, P = 0.02; fast-paced, P = 0.04) and backward (comfortable, P = 0.02; fast-paced, P = 0.01) directions, with non-fallers walking significantly faster under all conditions. A significant difference was observed between groups for vibration threshold (P = 0.04), with the fallers having a threshold of 5.90 ± 4.09 and the non-fallers 3.90 ± 2.40, indicating enhanced proprioception in the non-fallers.

Table 2 presents Spearman’s rho for vibration thresholds to walking assessments in both the FW and BW direction at both the comfortable and fast-paced walking velocities. Significant positive correlations were observed for FW-T25FW comfortable (ρ = 0.35, P = 0.01) and FW-T25FW fast-paced (ρ = 0.34, P = 0.01) trials, denoting that a higher vibration threshold, indicative of poorer proprioception, correlated with slower time to complete walking trials. Similar results were observed for BW trials, with stronger associations observed for BW-T25FW comfortable (ρ = 0.46, P< 0.01) and BW-T25FW fast-paced (ρ = 0.46, P < 0.01) compared to similar walking speed categories in the FW direction.

Table 2.

Correlation Between Vibration Sensation Threshold and Walking Trials

Walking Test	Vibration Threshold
	Spearman’s Rho	Confidence Intervals	P Value
FW-T25FW comfortable	0.35	0.07-0.58	0.01
FW-T25FW fast-paced	0.34	0.06-0.57	0.01
BW-T25FW comfortable	0.46	0.21-0.66	<0.01
BW-T25FW fast-paced	0.46	0.20-0.66	<0.01

Abbreviations:BW-T25FW, backward walking timed 25-foot walk test; FW-T25FW, forward walking timed 25-foot walk test.

Results of the linear regression analysis after correcting for age, sex, and PDDS for each walking trial can be seen in Table 3. For FW trials, the vibration threshold significantly contributed to the model under both FW-T25FW comfortable (β = 0.41, T = 3.62, P ≤ 0.01) and FW-T25FW fast-paced (β = 0.45, T = 3.95, P ≤ 0.01) conditions. In the BW direction, vibration threshold was a significant contributor to BW trial time to completion, with larger standardized beta coefficients seen under both the BW-T25FW comfortable (β = 0.57, T = 5.15, P = < 0.01) and BW-T25FW fast-paced (β = 0.58, T = 5.30, P = < 0.01) walking condition compared to similar speed categories in the FW direction.

Table 3.

Linear Regression Analysis for All Walking Trials

Forward Walking
		β	T	P Value
Comfortable	Vibration threshold (vu)	0.41	3.62	<0.01
	Age	−0.27	−2.75	0.01
	Sex	0.09	0.90	0.37
	PDDS	0.49	4.45	<0.01
	R 2	0.59
	P value	<0.01
	F-statistic (4,45)	16.31
		β	T	P Value
Fast	Vibration threshold (vu)	0.45	3.95	<0.01
	Age	−0.28	−2.79	0.01
	Sex	0.13	1.31	0.20
	PDDS	0.45	4.04	<0.01
	R 2	0.59
	P value	<0.01
	F-statistic (4,45)	16.06
Backward Walking
		β	T	P Value
Comfortable	Vibration threshold (vu)	0.57	5.15	<0.01
	Age	−0.22	−2.29	0.03
	Sex	0.13	1.35	0.19
	PDDS	0.35	3.26	<0.01
	R 2	0.61
	P value	<0.01
	F-statistic (4,45)	17.82
		β	T	P Value
Fast	Vibration threshold (vu)	0.58	5.30	<0.01
	Age	−0.2	−2.13	0.12
	Sex	0.15	1.60	0.04
	PDDS	0.35	3.34	<0.01
	R 2	0.63
	P value	<0.01
	F-statistic (4,45)	18.87

Abbreviations: β, standardized beta coefficient; PDDS, Patient Determined Disease Steps.

DISCUSSION

The impact of decrements in proprioception on walking and balance performance in individuals with MS are well documented.^14,34,35 Specifically, delays in signal transduction have been shown to lead to increased sway during static balance,^34,35 worse performance on clinical assessments of balance,^12,13 and increased muscle onset latencies.¹⁴ All of these factors are associated with an increased fall risk for individuals with MS.^36-38 While previous research has focused on the contributions of proprioception to static balance and FW, no study to date has examined the contributions of proprioception to BW. Given that BW has been shown to be increasingly sensitive to falls compared to FW,^16,39 and considering the reduced visual cues present during the task, understanding the proprioceptive underpinnings involved is of utmost importance. It may provide additional insights into why BW is particularly sensitive to falls.

Results from our study highlight that decrements in proprioception, as measured by vibration sensation, were more strongly correlated with BW compared to FW. These results remained consistent at both comfortable and fast walking velocities. Furthermore, beta coefficients from linear regression analyses revealed that, while both FW and BW conditions relied on proprioception, larger coefficients were observed during BW trials. Further examination of the beta coefficients within models provides additional insights into the significance of proprioception to BW performance, as vibration threshold had the largest impact on the BW models, whereas PDDS had the largest beta coefficient in both FW models. These findings suggest that BW, at both comfortable and fast walking speeds, relies more heavily on proprioceptive information for successful execution. In addition to the novelty of the task and heightened cognitive requirements, the increased proprioceptive reliance of BW may explain why it is a more demanding and challenging task.¹⁵

The results from regression analyses identified the impact of demographic variables on both FW and BW. Age significantly contributed to FW comfortable, FW fast, and BW comfortable, whereas sex significantly contributed to BW fast, with increasing age and female sex leading to slower gait speed. These findings align with previous literature identifying the significant impact of age and sex on gait speed.^40,41 While this is true, their contributions to the models were smaller than those of vibration sensation and PDDS, with vibration sensation remaining the largest contributor to BW, even after correcting for age, sex, and PDDS. These results reveal that while demographics and PDDS are important to correct for given their contributions to gait speed, they do not fully explain gait performance, especially in the backward direction, which relies more heavily on proprioception.

The heightened sensitivity observed during BW may be attributed, at least in part, to the increased reliance on proprioception demanded by the task. With the absence of visual cues during BW, sensory reweighting—the capability to adapt and reweight sensory inputs in response to changes in the environment—becomes necessary, thereby increasing the dependence on the proprioceptive and vestibular systems. As our results demonstrate, significant differences were observed between fallers and non-fallers in vibration sensation, with non-fallers exhibiting a lower threshold compared to fallers. This finding aligns with previous research indicating that difficulties with sensory reweighting contribute to an elevated risk of falls in individuals with MS.⁴² Additionally, Craig et al discovered that individuals prone to falls may exhibit a greater dependence on visual feedback during walking.⁴³ Collectively, the findings of our study build upon previous literature, suggesting that individuals with MS who are at a heightened risk for falls tend to rely more heavily on visual cues during walking. However, when confronted with a task that places greater emphasis on proprioception, such as BW, challenges with sensory reweighting arise, resulting in increased sensitivity while walking backward.

Results from this study may offer further insights into the neural mechanisms linking both BW and reactive stepping, particularly in the backward direction. Previous research by Monaghan et al found that vibration sensation was impaired in individuals with MS compared to healthy controls, with this impairment accompanied by increased muscle onset latencies.¹⁴ Furthermore, while this prior study noted differences in muscle onset latencies in both forward and backward directions, larger effect sizes, indicative of increased muscle latency, were observed in the backward direction.¹⁴ Considering that a significant proportion of falls occur in the backward direction,⁴⁴ interventions targeting the enhancement of reactive stepping and movements in the backward direction may be clinically impactful. As previous studies have identified improvements in proprioception through intervention,⁴⁵ our study, along with the findings of Monaghan et al, suggests that an improvement in proprioception is necessary to execute a reactive step in the posterior direction, which could potentially be enhanced through BW training.

While our study had several notable strengths, including the assessment of both FW and BW at a comfortable and fast pace, as well as a diverse population of individuals with MS, there are also limitations to consider. Specifically, although the Vibratron II is a valid instrument that is clinically assessable and has been previously utilized in this population,²¹ sensation was only tested at the great toe, rather than at multiple locations on the foot. Additionally, while previous studies have established a connection between vibration sensation thresholds and cortical brain areas in persons with MS,⁴⁶ the extent of biological differences in these areas remains unknown. Therefore, future studies should investigate whether structural changes in areas of interest in both the brain and spinal tract are correlated with BW. Furthermore, our study stratified individuals by retrospective fall risk, which is subject to recall bias.⁴⁷ While spasticity and ataxia are important factors associated with mobility performance in persons with MS,^48,49 our study did not assess these 2 factors, warranting future studies. Moreover, the study did not examine vestibular function, which has both been shown to be impaired in persons with MS and important for postural control and balance.⁵⁰ Lastly, while our study contributes to establishing a plausible explanation for the relationship between BW and backward reactive stepping, reactive stepping was not specifically examined in the current study. Given the clinical significance of both BW and reactive stepping, further research is warranted to establish a direct relationship between the two.

CONCLUSIONS

Our current study builds upon prior research that has demonstrated the increased utility of BW as a mobility assessment,^15,16,51 given its heightened proprioceptive reliance compared to FW. Moreover, considering that individuals with MS who are at a heightened risk for falls often encounter difficulties with sensory reweighting and are more likely to rely on vision for walking balance,^42,43 assessments and interventions targeting proprioception could potentially mitigate the risk of future falls. This study lays the groundwork for future research to explore and assess the relationship between BW and backward reactive stepping, as well as the effects of BW interventions on clinical assessment outcomes. Furthermore, in line with previous studies,^16,51 the findings underscore the importance of incorporating BW into fall risk assessments, thereby enhancing fall risk detection and enabling earlier, more targeted interventions for individuals with MS.

Supplementary Material

Video Abstract available for more insights from the authors (see the Supplemental Digital Content available at: http://links.lww.com/JNPT/A490).

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.jnpt.org).