Effect of Backward and Forward Walking on Lower Limb Strength, Balance, and Gait in Multiple Sclerosis: A Randomized Feasibility Trial

Abstract

BACKGROUND

Backward walking (BW) interventions have improved gait and balance in persons with stroke, cerebral palsy, and Parkinson disease but have not been studied in persons with multiple sclerosis (MS). We examined the feasibility of a BW intervention and how it affected strength, balance, and gait vs forward walking (FW) in persons with MS.

METHODS

Sixteen persons with MS with a Patient-Determined Disease Steps (PDDS) scale score of 3 to 5 (gait impairment-late cane) were randomized to the FW (n = 8) or BW (n = 8) group. Participants did 30 minutes of FW or BW on a treadmill 3 times per week for 8 weeks (24 visits). Enrollment, adherence rate, and safety were tracked. The Timed Up and Go test, Six-Spot Step Test, single-leg stance, and abbreviated Activities-specific Balance Confidence scale were used to measure balance. Hip and knee flexion and extension strength (isometric peak torque), gait speed, and spatiotemporal gait parameters were measured. A 2×2 factorial multivariate analysis of covariance was used to examine changes in strength, balance, and gait, with the PDDS scale score as the covariate.

RESULTS

Treatment adherence rate was 99.7%, with no safety concerns. After controlling for baseline differences in disability (PDDS scale score; P = .041), the BW group improved dominant hip flexion strength preintervention to postintervention compared with the FW group (F_1,13 = 9.03; P = .010). No other significant differences were seen between groups.

CONCLUSIONS

This was the first study to look at BW as an intervention in persons with MS. Based on its feasibility, safety, and significant finding, BW should be studied in a larger, definitive trial in the future.

Gait impairment occurs in approximately 85% of persons with multiple sclerosis (MS) and greatly affects participation and independence.1 Impairment in gait stems from multiple factors, including muscle weakness2 and decreased balance.3 Compared with controls, persons with MS exhibit decreased gait velocity, shorter step and stride lengths, and lower percentage of time in the swing phase.4 In addition, persons with MS spend increased time in double limb support,4 which is thought to be a compensation for postural instability.5 The medication dalfampridine sustained-release has been shown to increase gait velocity, but only in individuals who are responders to the drug.6 Other typical interventions to address gait impairments include combined exercise training, such as aerobic and resistance training; physical therapy activities, such as balance training and neurofacilitation; and gait training via a treadmill, robot-assisted training, or conventional overground training.6 Regardless of the mode of gait training, the most common direction is forward walking (FW), which has been shown to improve balance in persons with MS7 and improve walking velocity by treadmill training,8 conventional walking,9 and robot-assisted walking.9 Although FW is often used to improve gait and balance in persons with MS, to our knowledge, its effect on lower limb strength9 has not been studied.

Alternatively, backward walking (BW) has been incorporated into rehabilitation approaches,10 resulting in improvements in objectively measured balance,11 self-perceived balance,12 and gait characteristics such as velocity,10,13–15 stride length,10,13,15 step length,14,16 cadence,14,15 and double limb support time13 for persons with stroke,10,12,13,16 cerebral palsy,11,14 and Parkinson disease.15 When specifically compared with FW, BW has resulted in greater improvements in step length,14 stride length,15 velocity,14 cadence,14,15 stance phase,14 and swing phase percentage14 in these populations. Although motor learning theory requires repetition of the task that is being learned, FW and BW share the same central pattern generator,17 muscle synergy modules,18 and common control strategies,18 allowing adaptations made from BW to translate to FW. Training in BW may address balance and gait impairments by improving postural stability,19 increasing neural and muscular activity,20,21 and increasing metabolic response19,20 due to the novelty, increased complexity, and longer duration of muscle activity21 during backward movement; however, limited studies have reviewed BW in persons with MS.

Given the promising early evidence to support BW as an intervention in other populations and the importance of understanding the effect of walking interventions on strength in persons with MS, understanding the role of BW as an intervention for persons with MS is imperative. The purpose of this study was to establish the feasibility of BW as a therapeutic intervention and to explore the effects of a BW intervention vs a FW intervention on strength, balance, and gait in persons with MS. Because previous studies found that BW resulted in greater improvements in balance and gait parameters over FW,14,15 in addition to the effects from the novelty of the movement, we hypothesized that a BW intervention would result in greater improvements in balance, walking velocity, step length, and stride length in persons with MS.

METHODS

Study Design

This was a randomized, single-blind pilot study that examined the effects of BW vs FW training on lower limb strength, balance, and gait in persons with MS. The study was approved by the Trinity Health Of New England institutional review board and was performed in accordance with the Declaration of Helsinki. Recruitment occurred between April 2016 and September 2017. Participants completed a preintervention visit to determine eligibility and collect baseline assessments. Demographic and disease characteristics collected included self-reported disability (Patient-Determined Disease Steps [PDDS] scale score), age, sex, disease duration, pain (visual analog scale [VAS]–pain), fatigue (VAS-fatigue and Daily Fatigue Impact Scale), and percentage use of assistive devices (no assistance, wall or furniture, unilateral, bilateral, or wheelchair/scooter). After confirmation of eligibility and baseline assessment, participants were randomly assigned to the FW group (n = 8) or the BW group (n = 8) in a 1:1 ratio. Depending on randomization, participants completed 30 minutes of FW or BW on a treadmill during a 1-hour session 3 times per week for 8 weeks (24 visits).

Participants

Eligible participants were English-speaking adults aged 18 to 65 years with a clinical diagnosis of MS; a score of 22 or greater on the Mini-Mental State Examination; gait impairment, defined as a PDDS scale score of 3 to 5 (in the past 12 months); and were able to complete the 6-Minute Walk Test. Individuals were excluded if they were unwilling or unable to complete the assessments and intervention; were currently participating in physical therapy; had major changes in exercise habits in the past 3 months (ie, starting or stopping a new program, or a change in frequency, intensity, or duration); had an MS relapse within the past 2 months; required constant bilateral support for ambulation and/or could not walk at least 100 m without resting; had a self-reported history of any absolute contraindications for exercise22; and/or could not safely adhere to the protocol. A convenience sample of participants was recruited from a comprehensive MS clinic, the Mandell Center for Multiple Sclerosis, at Mount Sinai Rehabilitation Hospital, Trinity Health Of New England.

Given that BW has not been studied as an intervention for persons with MS or determined to be a feasible and tolerable intervention for this population, the following study was designed as a pilot study to inform the sample size required for a larger definitive trial. A sufficient sample size to calculate the same size of a 90% powered main trial with a large, standardized difference on the primary treatment outcome23 was deemed to be 10 participants per trial arm.

Intervention

Each 1-hour treatment session consisted of a 5- to 10-minute, low-intensity warm-up (including the participant’s unsupervised walk from the parking lot [2–5 minutes] and marching in place), stretching of the lower limb, and then 3 to 5 minutes of slow walking on the treadmill. The warm-up was followed by 30 minutes of walking at 0% incline (FW or BW) for the intervention. Finally, the session ended with a 5- to 10-minute cooldown, walking at a slow speed on the treadmill for 5 minutes, followed by stretching and taking deep, slow breaths.

For the intervention, each participant was placed in a harness mounted over a treadmill to prevent falls (no weight support was provided). All participants started at a comfortable walking speed, determined by slowly increasing the speed throughout the first session to where the participant was able to maintain the exercise and reported that they were working at a Borg rating of perceived exertion (RPE) between very light (9) and somewhat hard (13). Due to the heterogeneity of the population, each participant was progressed according to their abilities and response to the training with the goal of completing 30 minutes of walking. To accommodate this heterogeneity, the treatment was progressed by first decreasing rest breaks and then by increasing speed and was customized for each participant at the most challenging speed they could tolerate to complete the full 30 minutes.

Heart rate, oxygen saturation, and RPE were collected every 3 minutes to monitor intensity. The research therapist controlled the speed and provided verbal cues to promote a normal gait pattern. Measures were collected before the session (blood pressure [BP], VAS-fatigue, and Daily Fatigue Impact Scale), during rest breaks (BP), and after the session (BP and VAS-fatigue) for safety monitoring. Rest breaks were incorporated as needed to ensure completion of the intervention and did not count toward the 30 minutes of walking. Reasons for rest included shortness of breath; lightheadedness; feelings of fainting, fatigue, or cramping; inability to safely perform the walking protocol; unsafe drop in oxygen saturation; and at the therapist’s discretion to maximize the potential for the completion of 30 minutes of walking.

Enrollment was tracked throughout the recruitment and study procedures, including how many people were approached or expressed interest; reasons why people declined or did not meet the inclusion criteria; and reasons why participants discontinued the trial. Adherence to the treatment program was noted, including the number of sessions completed by each participant and the overall adherence rate by treatment group. Safety and adverse effects were logged. Baseline and posttreatment strength, balance, and gait functioning were assessed by a blinded evaluator unaware of group assignment using the measures detailed in TABLE S1 (references in TABLE S2), available online at IJMSC.org. All participants were educated on the importance of not informing the evaluator of any details of the intervention they received, especially walking direction, at the posttreatment assessment, which was completed within 1 week of the final treatment session.

Statistical Analysis

Statistical analyses were performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp). Descriptive statistics were used to report enrollment, adherence, and safety data. Baseline demographics were analyzed using the Kolmogorov-Smirnov test to determine normality. Age, fatigue in the past month, and percentage of time not using an aid were normally distributed (P > .05); however, all other measures significantly deviated from normal (P < .05). An independent t test (normal) and a Mann-Whitney U test (nonnormal) were used to compare the groups’ baseline demographics to check for significant differences. A Fisher exact test compared sex, and a median test compared disability level (PDDS scale score). Two-tailed P < .05 was considered significant for the bivariate analysis. Two-tailed P < .01 was considered significant for the multivariate analyses due to the multiple comparisons. Due to some of the outcome measures deviating from normal, all the measures were individually ranked to meet assumptions. After ranking, baseline outcome measures were compared using a multivariate analysis of covariance (MANCOVA) between the groups, with PDDS scale score as the covariate. A 2 (BW group and FW group) × 2 (pretreatment and posttreatment) factorial MANCOVA examined changes in the strength, balance, and gait outcome measures, with PDDS scale score as the covariate.

RESULTS

Enrollment and Participants’ Demographics

A total of 48 persons with MS were contacted or inquired about the study (FIGURE S1). Seven had time conflicts (eg, work schedules), 8 were not eligible (ie, currently in physical therapy, use of bilateral support, or not within the age parameters), 3 were not interested, and 6 did not respond to follow-up contact. The remaining 24 persons with MS provided written consent; however, 6 of these individuals were excluded during the baseline assessment for not meeting the inclusion criteria (for having an absolute contraindication to exercise or a PDDS scale score outside of the 3–5 range). Two participants assigned to the FW group withdrew after randomization: 1 due to a schedule conflict and the other after reporting fatigue after treatments. The remaining 16 individuals completed the study.

Participant demographics are displayed in TABLE 1. A baseline difference in disability level (PDDS scale score) was observed between groups (χ²₁ = 6.35; P = .041), but no differences were observed in any of the other demographic and baseline characteristics. There was no baseline difference (P > .01) between the outcome measures (strength, gait, balance) as measured by the MANCOVA after controlling for PDDS scale score.

TABLE 1.

Demographic and Baseline Characteristics

Characteristic	FW group (n = 8)	BW group (n = 8)	P value
Age, mean ± SD, y	49.6 ± 12.4	49.1 ± 8.3	.926
Sex, M/F, No.	3/5	3/5	>.99
Disease duration, mean ± SD, y	12.2 ± 9.8	9.8 ± 8.4	.798
PDDS scale score, median	3.0	4.0	.041a
Fatigue (% of 100), mean ± SD
At baseline	2.4 ± 2.1	3.9 ± 3.8	.382
Past month	9.2 ± 5.2	4.3 ± 2.8	.461
Pain (% of 100), mean ± SD
At baseline	2.0 ± 2.3	0.9 ± 1.7	.646
Past month	3.3 ± 4.1	1.8 ± 2.2	.721
Time not using aid, mean ± SD, %	70.6 ± 29.6	52.8 ± 32.7	.272
Time using aid, mean ± SD, %
Wall or furniture	13.8 ± 14.6	17.5 ± 18.5	.574
Unilateral support	8.1 ± 11.9	21.6 ± 17.1	.105
Bilateral support	7.5 ± 21.2	3.1 ± 8.8	.959
Wheelchair or scooter	0	6.3 ± 14.0	.442

BW, backward walking; FW, forward walking; PDDS, Patient-Determined Disease Steps.

^aStatistically significant.

Treatment Adherence and Safety

There was an overall treatment adherence rate of 99.7% (n = 16). The BW group had a 100% adherence rate, and the FW group had a 99.5% adherence rate due to 1 participant completing 23 of 24 sessions (scheduling conflict). In the BW group, 62.5% of participants (n = 5) completed the 24 sessions within 8 weeks ± 5 days; 12.5% (n = 1) completed the 24 sessions over 9 to 10 weeks; and 25% (n = 2) completed the sessions over 10 to 11 weeks. In the FW group, 75% (n = 6) completed the 24 sessions within 8 weeks ± 5 days and 25% (n = 2) completed the 24 sessions over 9 to 10 weeks. Median (interquartile range) of the metrics (heart rate, RPE, fatigue, oxygen saturation, rest, BP, distance) collected during the first and last sessions for each group are presented in TABLE S3. No adverse effects or safety concerns were reported.

Strength, Balance, and Gait

Greater improvement in dominant hip flexion strength was observed in the BW group compared with the FW group from preintervention to postintervention (F_1,13 = 9.03; P = .010). No other significant differences were observed in any of the other strength, balance, or gait measures between groups (TABLE 2).

TABLE 2.

Preintervention (Pre) and Postintervention (Post) Between-Group Comparison of Strength, Balance, and Gait Measures (2×2 Factorial MANCOVA)

		FW group	BW group	P value
Strength (peak torque)
Dominant hip flexion, ft-lb	Pre	18.4 (26.6)	7.4 (12.6)	.010a
	Post	17.1 (27.7)	17.1 (17.7)
Nondominant hip flexion, ft-lb	Pre	14.5 (27.4)	9.5 (13.9)	.842
	Post	16.9 (18.2)	15.0 (26.6)
Dominant hip extension, ft-lb	Pre	60.3 (39.7)	50.0 (22.1)	.303
	Post	71.5 (61.9)	52.7 (41.4)
Nondominant hip extension, ft-lb	Pre	61.7 (38.4)	50.1 (46.4)	.783
	Post	71.5 (52.6)	51.8 (54.4)
Dominant knee flexion, ft-lb	Pre	46.9 (38.0)	31.2 (28.3)	.160
	Post	35.4 (39.4)	33.1 (27.2)
Nondominant knee flexion, ft-lb	Pre	42.2 (40.7)	33.5 (25.1)	.962
	Post	39.2 (29.1)	30.0 (27.8)
Dominant knee extension, ft-lb	Pre	71.9 (84.4)	55.5 (33.1)	.475
	Post	67.6 (73.4)	57.0 (58.5)
Nondominant knee extension, ft-lb	Pre	82.8 (72.1)	37.6 (64.8)	.462
	Post	75.7 (48.6)	53.9 (60.3)
Balance
TUG, sb	Pre	8.8 (3.1)	12.6 (6.4)	.150
	Post	8.4 (4.0)	11.4 (8.4)
SSST, sb	Pre	13.2 (5.0)	17.7 (21.7)	.109
	Post	12.2 (4.1)	15.4 (26.3)
SLS, sc	Pre	30.2 (40.2)	8.6 (16.0)	.965
	Post	30.2 (43.7)	6.3 (26.0)
ABC-6, %d	Pre	54.2 (22.7)	52.5 (33.8)	.910
	Post	55.0 (34.4)	39.2 (33.8)
Gait
T25FW test, se	Pre	5.5 (2.0)	7.0 (2.9)	.068
	Post	5.3 (2.3)	7.1 (4.2)
Velocity, cm/sf	Pre	141.2 (38.1)	113.3 (39.2)	.135
	Post	145.8 (53.1)	109.0 (48.8)
Left step length, cmf	Pre	67.6 (9.2)	59.4 (23.1)	.246
	Post	68.8 (12.1)	59.7 (32.9)
Right step length, cmf	Pre	67.0 (13.9)	61.0 (31.0)	.926
	Post	68.5 (18.5)	61.1 (24.9)
Left stride length, cmf	Pre	133.9 (22.9)	119.9 (52.3)	.584
	Post	137.2 (30.7)	120.7 (60.6)
Right stride length, cmf	Pre	133.8 (22.6)	120.3 (52.5)	.584
	Post	137.5 (30.3)	121.6 (60.6)
Left DS%, %e	Pre	26.7 (9.7)	31.4 (7.5)	.649
	Post	25.4 (9.5)	30.3 (9.3)
Right DS%, %e	Pre	26.6 (9.9)	32.1 (8.0)	.481
	Post	24.2 (8.8)	29.9 (9.2)

Data are given as median (interquartile range). ABC-6, abbreviated Activities-specific Balance Confidence scale; BW, backward walking; DS%, double limb support; FW, forward walking; MANCOVA, multivariate analysis of covariance; SLS, single-leg stance; SSST, Six-Spot Step Test; T25FW, Timed 25-Foot Walk; TUG, Timed Up and Go.

^aSignificant finding in the groups preintervention vs postintervention.

^bTUG and SSST: faster time = positive improvement.

^cSLS: longer duration = positive improvement.

^dABC-6: higher score = positive improvement.

^eT25FW test and DS%: decrease = positive improvement.

^fVelocity, step length, stride length: increase = positive improvement.

DISCUSSION

This pilot study evaluated the feasibility of BW as a therapeutic intervention and compared the effect of BW and FW as an intervention on strength, balance, and gait in persons with MS. Overall, BW was a well-tolerated intervention with no safety issues. Although 2 participants randomized to the FW group withdrew from the study, those who completed the study had 99.7% adherence to the treatment program. No difference was found between the 2 groups for any of the balance measures, spatiotemporal gait characteristics, or Timed 25-Foot Walk test. Therefore, the data did not support the hypothesis that BW would result in greater improvements in balance, walking velocity, step length, and stride length.

The results of this study indicate an increased use of and resultant strengthening of hip flexors during BW that may not occur with FW training. The increased dominant hip flexion strength is likely due to the increased eccentric control required by this muscle group during BW. Comparing the intermuscular synergies between FW and BW, the main propulsive thrust for FW is from the plantar flexors through the ankle joint. Two studies reported that the main propulsive thrust for BW is not provided by the ankle24 but by the hip or knee extensors.20 Given that this study did not find increases in hip or knee extension strength, although these muscle groups are suspected to provide the main source of propulsion, monitoring these outcomes in a larger sample size is recommended.

The increased dominant hip flexion strength may have occurred in the BW group from the increase in hip flexion required to lift the leg high enough while stepping BW or during the stance phase while eccentrically contracting to improve control during the contralateral swing phase. Due to the repetitive nature of continuously flexing the swing phase hip with each step for 30 minutes, the participants completed many repetitions of this movement, allowing for the possibility for neural adaptations to occur. The BW group may have had greater room for improvement in hip flexor strength due to the difference in the baseline assessment compared with the FW group (median: 18.4 ft-lb at baseline; BW group median: 7.4 ft-lb at baseline). Currently, the literature supports the use of resistance training, among other physical therapy strategies, to address hip muscle weakness for persons with MS.25 Now, BW may be considered as an option for persons with MS to improve hip flexor strength.

No differences were found between the groups for the balance measures (Timed Up and Go test, Six-Spot Step Test, single-leg stance, and abbreviated Activities-specific Balance Confidence scale) or any of the gait measures (Timed 25-Foot Walk test, velocity, bilateral step length, bilateral stride length, and bilateral double limb support), which was inconsistent with previous findings. However, only 2 of the previous studies compared BW and FW,14,15 and others compared BW with a balance program12 or conventional physical therapy (consisting of approximately strengthening, function, and mobility activities and 30% gait training).10,13

Two factors may explain the limited differences between the 2 groups. One may be the statistical power of the present sample size. Given the lack of evidence of BW as an intervention in persons with MS, this study was designed as a small pilot study to inform a larger, definitive clinical trial. Another factor is that an objective measure of spasticity was not included. Spasticity could have been a confounding variable, affecting motor recruitment and gait, and ultimately affecting progression throughout the intervention. Omission of a spasticity measurement may have masked some of the adaptations and made it more difficult to demonstrate statistical significance in the findings.

The current literature in other special populations (ie, Parkinson disease, stroke, cerebral palsy) suggests that BW is a useful intervention to improve spatiotemporal gait parameters and balance. Backward walking is a novel task that can challenge the body differently than FW while still training the same central pattern generator,17 muscle synergy modules,18 and common control strategies.18 Weakness affects 70% of persons with MS,26 which can lead to gait changes and balance deficits, so studying this intervention in persons with MS is vital, as these deficits can lead to an increased risk of falls, which can lead to many negative consequences (ie, injury, fear of falling, decreased physical activity, decreased activities of daily living). Although this study did not demonstrate greater improvement from BW compared with FW, positive changes were observed and the intervention was well tolerated with a high adherence rate, supporting investigation of the use of BW as another treatment for persons with MS.

PRACTICE POINTS

• Backward walking has improved gait, balance, and strength in other special populations. Studying this intervention in persons with multiple sclerosis (MS) is vital because a large portion of persons with MS have gait impairments, balance deficits, and weakness, which makes them more prone to falls.

• Backward walking was found to be a safe and feasible intervention in persons with MS, with an extremely high adherence rate of 99.7%.

• Backward walking demonstrated an increase in dominant hip flexion strength compared with forward walking, indicating that backward walking may be used in persons with MS to improve hip flexion weakness, a prevalent symptom in MS.

The study design was a simple 1:1 randomization with a small sample size; therefore, it would have been difficult to control for differences at baseline. Given the small sample size, a crossover design may have been helpful in interpreting the effect of baseline differences. The intervention was performed over a treadmill in a harness; however, this setup may not be readily available in all rehabilitation settings, making clinical application a challenge.

Future research could explore targeted interventions based on individual goals and existing deficits to assess the potential benefit from combined interventions, including BW and cross-training.

Funding Statement

This work was supported by the National Multiple Sclerosis Society (G-1510-07287). Dr Gromisch is a Harry Weaver Scholar of the National Multiple Sclerosis Society.