effect of partial weight-supported treadmill gait training ... · training on balance in patients...

Original Research

Effect of Partial WeighteSupported Treadmill GaitTraining on Balance in Patients With Parkinson DiseaseMohan Ganesan, PhD, Talakad N. Sathyaprabha, MBBS, MD,Anupam Gupta, MBBS, MD, Pramod Kumar Pal, MBBS, MD, DM

M.G. Department of Neurophysiology, Na-tional Institute of Mental Health & Neurosci-ences, Bangalore, India; currently at

Objective: To investigate the role of conventional gait training and partial weightesupported treadmill gait training (PWSTT) in improving the balance of patients withParkinson disease (PD).Design: Prospective randomized controlled design.Setting: National-level university tertiary hospital for mental health and neurosciences.Patients: Sixty patients with PD fulfilling the United Kingdom Brain Bank PD diagnosticcriteria were recruited from the neurology outpatient department and movement disorderclinic.Methodology: The patients were randomly assigned into 3 equal groups: (1) a controlgroup that only received a stable dosage of dopaminomimetic drugs; (2) a conventional gaittraining (CGT) group that received a stable dosage of dopaminomimetic drugs and con-ventional gait training; and (3) a PWSTT group that received a stable dosage of dop-aminomimetic drugs and PWSTT with unloading of 20% of body weight. The sessions forthe CGT and PWSTT groups were provided for 30 minutes per day, 4 days per week, for 4weeks (16 sessions).Outcome measures: The Unified Parkinson Disease Rating Scale (UPDRS) motorscore, dynamic posturography, Berg Balance Scale, and Tinetti performance-orientedmobility assessment (POMA) were used as main outcome measures.Results: A significant interaction effect was observed in the UPDRS motor score, medio-lateral index, Berg Balance Scale, limits of stability (LOS) total score, POMA gait score, andbalance score. Post-hoc analysis showed that in comparison with the control group, thePWSTT group had a significantly better UPDRS motor score, balance indices, LOS in 8directions, POMA gait, and balance score. The CGT group had a significantly better POMAgait score compared with control subjects. Compared with the CGT group, the PWSTTgroup had a significantly better UPDRS motor score, mediolateral index, POMA gait score,and LOS total score.Conclusion: PWSTT may be a better interventional choice than CGT for gait and balancerehabilitation in patients with PD.

PM R 2014;6:22-33
Department of Physical Therapy, University ofIllinois, Chicago, IL Disclosure: nothing to disclose
T.N.S. Department of Neurophysiology, Na-tional Institute of Mental Health & Neurosci-ences, Bangalore, IndiaDisclosure: nothing to disclose

A.G. Neurological Rehabilitation, NationalInstitute of Mental Health & Neurosciences,Bangalore, IndiaDisclosure: nothing to disclose

P.K.P. Department of Neurology, NationalInstitute of Mental Health & Neurosciences,Bangalore 560029, India. Address corre-spondence to: P.K.P.; e-mail: [email protected]: nothing to disclose

Submitted for publication August 11, 2012;accepted August 22, 2013.

INTRODUCTION

Parkinson disease (PD) is characterized by deteriorating motor function due to degenera-tion of the dopaminergic nigrostriatal pathways. Gait and balance abnormalities observedin persons with PD are important components of the disability [1]. Despite the use oflevodopa, mobility deficits related to gait and balance are very difficult to treat [2,3].Adjuvant physiotherapy has shown promise in addressing this crucial aspect of manage-ment of PD. Partial weightesupported treadmill gait training (PWSTT) is a promisingtherapeutic approach to help retrain patients with PD to walk [4-6]. The authors of a meta-analysis on treadmill training in patients with PD identified 8 trials in which investigatorsassessed the utility of variants of treadmill training and PWSTT in improving mobility inpatients with PD [7]. Only a few studies have addressed the effect of treadmill training onbalance in persons with PD [6,8,9]. The Berg Balance Scale (BBS), fall efficacy scale, andstep test were used as outcome measures for balance control.

22PM&R1934-1482/13/$36.00

Printed in U.S.A.

ª 2014 by the American Academy of Physical Medicine and RehabilitationVol. 6, 22-33, January 2014

http://dx.doi.org/10.1016/j.pmrj.2013.08.604

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.pmrj.2013.08.604

PM&R Vol. 6, Iss. 1, 2014 23

Only one study has addressed the sensory orientationtest with use of dynamic posturography as an outcomemeasure after PWSTT [6]. A knowledge gap exists withregard to the effect of PWSTT on the limits of stability(LOS) and direction-specific balance components in per-sons with PD. Visintin et al [10] reported a significantimprovement in functional balance, motor recovery, andgait after PWSTT in patients who had a stroke comparedwith patients undergoing treadmill training alone. PWSTTis considered to be a safer method of training, provides asense of security regarding falls, and facilitates free legmovements compared with treadmill training alone. Inaddition, persons with neurologic conditions were able towalk for a longer duration and with minimally increasedheart rates when partial weight support was provided [11].Therefore the current study was planned to assess theeffects of PWSTT on the dynamic balance indices, LOS inall directions, and clinical balance measures in personswith PD.

Figure 1. A flow chart of the study. CGT ¼ conventional gait trUPDRS ¼ Unified Parkinson’s diseases rating scale; OBI ¼ overall balanBBS ¼ Berg balance scale; POMA ¼ Tinetti Performance-Oriented M

METHODS

This study was a prospective randomized controlled trial.The schematic flow of the research study is depicted inFigure 1. Sixty patients with PD were randomly assignedinto 3 equal groups (n ¼ 20/group): (1) a control group thatonly received a stable dosage of dopaminomimetic drugs; (2)a conventional gait training (CGT) group that received astable dosage of dopaminomimetic drugs and conventionalgait training; and (3) a PWSTT group that received a stabledosage of dopaminomimetic drugs and PWSTT with 20%unloading of weight. Patients were recruited from theneurology outpatient department and movement disordersclinic at the National Institute of Mental Health & Neuro-sciences. The diagnosis of PD was confirmed by a movementdisorders specialist as per the United Kingdom Brain BankCriteria [12]. The ethics committee at the National Instituteof Mental Health and Neurosciences approved the study,and written informed consent was obtained from all

aining; PWSTT ¼ partial weightesupported treadmill training,ce index; API ¼ anteroposterior index; MLI ¼mediolateral index;obility Assessment.

24 Ganesan et al PWS TREADMILL TRAINING IN PATIENTS WITH PD

participants. Patients with cognitive deficits (ie, a mini-mental status examination score �24), moderate to severedepression (Beck Depression Inventory score �17), severedyskinesia (Goetz score >3), advanced PD (Hoehn and Yahr[H&Y] stage >3), unpredictable motor fluctuations, andorthopedic problems affecting gait training, as well as sub-jects who had undergone previous formal gait training orbalance training, were excluded from the study.

The severity of PD was assessed with use of the motorsection of the Unified Parkinson’s Disease Rating Scale(UPDRS-III). The hand and leg dominance of subjects wereassessed with use of the Edinburgh inventory. All partici-pants took a stable dosage of dopaminomimetic drugsthroughout the study period. The assessment and trainingwere performed during the best “on” period after the regularmedications were taken. All assessments were performed atbaseline, after 2 weeks (W2), and after 4 weeks (W4). Thebalance assessment and gait training were conducted inthe Gait and Balance Laboratory at the Department ofNeurorehabilitation.

Balance Assessment

Dynamic Posturography. Balance was tested with dy-namic posturography (Stability System [model 945-302];Biodex Medical Systems, software version 3.12, New York,NY), the details of which have been published elsewhere[13,14]. In brief, this system consists of a circular platformsupported so that it can tilt 20� in all directions from thehorizontal. The system’s microprocessor-based actuatorcontrols the extent of the surface instability of the platform.The surface instability can be adjusted from level 8 (mini-mally unstable) to level 1 (very unstable). Level 8 was usedfor examination of balance in all the subjects.

The subjects were tested for dynamic stability in alldirections (overall balance index [OBI]), front to back sta-bility (anteroposterior index [API]), and side to side stability(mediolateral index [MLI]). Each subject was asked tomaintain the center of his or her mass in the middle of aconcentric circle that appeared on the screen placed in frontof the subject. The instrument records the actual posturalsway and calculates the variance from the center, which isexpressed as a balance index. More postural sway is re-flected as a large variance that is quantified as a greater scoreof the balance index. The test consisted of 3 trials, each ofwhich were 20 seconds in duration with a 25-secondintertrial rest period. The average value of the 3 trials wastaken as the final balance index and was used for furtheranalysis.

The subjects also were tested for LOS in 8 directions,including forward (FW), backward (BW), right (RT),left (LT), forward-right (FW-RT), forward-left (FW-LT),backward-right (BW-RT), and backward-left (BW-LT). Forthe LOS test, each subject had to move his or her center ofmass, without changing foot position, into 8 targets, the

perimeter of which corresponded to 50% of the theoreticalLOS. The target is displayed on the screen by a blinkingsquare, which appeared randomly in different directionsonly once. The postural sway required to reach the targetfrom the center by the shortest vertical or horizontal path isrecorded by the instrument and given as a score. In aparticular direction, the maximum achievable perfect score is100. A lower score indicated greater sway. The maximumtime allowed to perform the movements to complete theLOS test was 300 seconds. The balance indices and the LOSscores as computed and obtained from the system were usedfor further analysis.

Clinical Measures of Balance

Clinically, balance was assessed with the BBS [15] andthe Tinetti Performance-Oriented Mobility Assessment(POMA) [16]. The clinical measures of balance performedby the BBS contain 14 items; each item is scored from 4(normal) to 0 (unable to perform task). The maximum scoreis 56. Each item in the POMA scale was scored on a 3-point ordinal scale ranging from 0 to 2 (0 ¼ maximumimpairment; 2 ¼ normal). The POMA has a gait componentand a balance component. The total balance score was 16,and the gait score was 12. The balance score was based ontesting of the following 9 items: balance while sitting, arisingfrom a chair, attempts to arise, immediate standing balance,standing balance, response to a nudge, eyes closed, turning360 degrees, and sitting down. The gait score was obtainedby testing the following 7 items: initiation of gait, step lengthand clearance, step symmetry, step continuity, walking path,trunk sway, and walking stance. The balance measures wereperformed by a qualified physiotherapist. The same raterperformed all the repeated measurements.

Training Groups

The control group received only a stable dosage of dop-aminomimetic drugs for the entire study period, whereas theCGT and PWSTT groups received gait training along withmedications. All 3 groups were encouraged to perform theirregular activities of daily living. The control group wasinstructed not to undergo any specific gait, balance, orexercise training until the end of the study period.

CGT included walking in a straight path, training toturn, and incorporating arm swinging strategies whilewalking. In addition, verbal auditory cues were given toencourage longer steps. The first 2 sessions of CGT wereperformed with the subject inside parallel bars, and thensessions were performed outside parallel bars. Each sessionof CGT consisted of 3 sets of 10-minute continuoustraining with 2 minutes of rest between the sets of training.The 10-minute walking training sessions included walkingin straight path and turning strategies such as step andpivot turnings with the help of floor markings in the

PM&R Vol. 6, Iss. 1, 2014 25

walking path. Arm swinging was taught and encouragedduring walking.

The PWSTT unit (Gait trainer [model 945-397]; BiodexMedical Systems, software version 2.01, Shirley, NY) con-sisted of a treadmill with visual biofeedback of step lengthand step speed and an unweighing support system. Thesupport system was set to unload 20% of body weight whilethe subject was training on the treadmill. The percentage ofbody weight unloaded was chosen on the basis of previousstudies [4,5]. The unweighing support system (model945-473; Biodex Medical Systems) provides continuousmonitoring of unweighed weight. In the CGT and PWSTTgroups, the progression and speed of the training wereindividualized to the subject’s own comfortable groundwalking speed and fast walking speed. In the initial 2 ses-sions, the subjects were encouraged to perform the walkingtraining at a comfortable ground walking speed. After 2sessions, the subjects were encouraged to perform thewalking at a self-selected comfortable fast walking speed.The CGT and PWSTT groups received four 30-minutetraining sessions, excluding a 5-minute warm-up and cool-down period for each session, for 4 weeks (a total of 16sessions).

Data Analysis

A 3�3 mixed factorial analysis of variance (with repeatedmeasures) was used to assess the significant changes in thestudy parameters. SPSS 15.0 statistical software was used fordata analysis. The 2 independent factors included in theanalysis were (1) group (between-subjects factor) with 3levels—control, CGT, and PWSTT; and (2) time (within-subjects factor) with 3 levels—baseline, W2, and W4. ABonferroni adjustment was performed for post-hoc com-parison. Two-sided tests were performed, the alpha was setat 5%, and the effect size was calculated by the partial etasquared (h2) value. Partial h2 interpretation for effect size isas follows: >0.2, large-effect size; >0.1, medium-effect size;and >0.05, small-effect size.

Table 1. Clinical details of the patients with Parkinson disease

CT

Age, y 59.1 � 6.8Women/men ratio 4:16Height, cm 162.4 � 5.4Weight, kg 61.3 � 10.9BMI, kg/m2 23.2 � 3.8Mean age of onset, y 53.6 � 7.5Duration of disease, y 5.5 � 3.4Hoehn and Yahr stage, 2/2.5 16:4Levodopa equivalent dosage, mg 698.3 � 227.1BDI 6.9 � 2.6MMSE 28.7 � 0.8

The Fisher exact test was used for gender distribution and Hoehn and Yahr Stage;CT ¼ control; CGT ¼conventional gait training; PWSTT ¼ partial weightesupported

MMSE ¼ mini-mental state examination.

RESULTS

Clinical details for the patients are provided in Table 1. De-mographic variables such as age, gender, height, and weightwere comparable for the 3 study groups. No significant dif-ferences were found in the mean duration of the disease, themean age of onset of symptoms, H&Y stage, UPDRS-IIIscores, Beck Depression Inventory scores, mini-mental statusexamination scores, and the levodopa equivalent dosage be-tween the 3 groups of patients. All the subjects completed thestudy, and no adverse effects were reported in any of thetreatment groups.

UPDRS-III Motor Scores

Table 2 shows the comparison of UPDRS-III (motor) scores.The UPDRS-III score showed significant group with timeinteraction (F2,114 ¼ 27.01; P < .001). Significant im-provement from baseline to W4 was noted in both the CGTand PWSTT groups (CGT group, P ¼ .014; PWSTT group,P < .001). In addition, the PWSTT group showed significantimprovement from W2 compared with the control group.No significant changes were seen in the control group(P ¼ .999). In group comparison, the PWSTT group showeda significant improvement compared with the control group(P ¼ .001) and the CGT group (P ¼ .040) at W4. However,the CGT group did not differ significantly from the controlgroup.

Clinical Measures of Balance

BBS. A significant group with time interaction was observedin the clinical balance measures of BBS (F ¼ 27.09; P <.001). The BBS score significantly increased in both the CGTand PWSTT groups after 4 weeks of treatment comparedwith baseline. Adjusted post-hoc group comparison didnot show any significant difference between the groups(Table 2). Effect sizes (partial eta value) were as follows:PWSTT, 0.666; CGT, 0.221; and control group, 0.20.

CGT PWSTT P Value

57.7 � 10.3 57.6 � 9.1 .8415:15 5:15 .999

158.3 � 7.0 159.4 � 7.8 .15658.8 � 12.4 61.05 � 8.4 .72023.4 � 4.4 24.1 � 3.5 .74352.8 � 9.1 51.9 � 10.6 .8454.9 � 3.1 5.7 � 3.9 .728

17:3 17:3 .999577.1 � 291.8 625.7 � 315.1 .394

5.8 � 1.8 6.3 � 2.3 .28728.8 � 1.0 29.3 � 0.9 .106

analysis of variance was used for other variables.treadmill training; BMI ¼ body mass index; BDI ¼ Beck Depression Inventory;

Table 2. Comparison of Unified Parkinson’s Diseases Rating Scale, balance indices, and clinical balance measures in all 3 groups

Parameters Groups BL Mean ± SD After 2 Weeks, Mean ± SD After 4 Weeks, Mean ± SD

UPDRS-III motor scoreCT 30.15 � 3.88 30.05 � 3.90 30.35 � 3.80CGT 30.70 � 5.04 30.15 � 4.43 28.70 � 4.69PWSTT 31.95 � 4.26 30.55 � 3.98 25.10 � 4.79

Balance indicesOBI CT 2.2 � 0.6 2.3 � 0.6 2.2 � 0.5

CGT 2.1 � 0.6 1.9 � 0.5 1.9 � 0.6PWSTT 2.2 � 0.4 1.8 6 � 0.5 1.6 � 0.6

API CT 1.8 � 0.4 1.7 � 0.5 1.8 � 0.3CGT 1.7 � 0.6 1.6 � 0.5 1.5 � 0.6PWSTT 1.7 � 0.4 1.5 � 0.4 1.4 � 0.5

MLI CT 1.4 � 0.3 1.5 � 0.4 1.5 � 0.3CGT 1.4 � 0.2 1.4 � 0.3 1.4 � 0.2PWSTT 1.4 � 0.4 1.2 � 0.5 1.1 � 0.4

Clinical balance measuresBBS score CT 48.9 � 2.4 49.0 � 2.5 48.9 � 2.8

CGT 49.8 � 3.0 49.9 � 3.0 51.4 � 2.0PWSTT 47.5 � 7.7 47.8 � 7.6 51.6 � 6.7

POMA-gait score CT 8.8 � 1.7 8.9 � 1.7 8.8 � 1.7CGT 9.7 � 1.3 9.8 � 1.3 10.7 � 1.1PWSTT 9.1 � 1.5 9.4 � 1.5 11.7 � 0.5

POMA ebalance score CT 13.0 � 2.0 13.0 � 2.0 13.0 � 2.0CGT 13.5 � 2.2 13.6 � 2.1 14.1 � 1.9PWSTT 12.9 � 2.3 13.3 � 2.0 15.0 � 0.9

BL ¼ baseline; W2 ¼ week 2; W4 ¼ week 4; UPDRS-III ¼ Unified Parkinson’s diseases rating scale; CT ¼ control; CGT ¼ conventionalgait training; PWSTT ¼ partial weightesupported treadmill training; OBI ¼ overall balance index; API ¼ anteroposterior index; MLI ¼mediolateral index; BBS ¼ Berg Balance Score; POMA ¼ Tinetti Performance-Oriented Mobility Assessment; RMANOVA ¼ repeated-measures analysis of variance.F and P values were expressed after RMANOVA with Bonferroni adjustment; group-specific and time-specific comparisons are

adjusted for multiple comparisons. Degree of freedom for group effects ¼ 2.57, interaction effects ¼ 4.114, and time effect ¼ 2.114.


Tinetti POMA. Table 2 provides a comparison of POMAcomponents. A significant interaction effect was observed inthe POMA gait component (F ¼ 32.813; P < .001) andbalance component (F ¼ 12.679; P < .001).

At W2, only the PWSTT group showed a significantimprovement in gait score (P ¼ .001) and balance score (P ¼.015) compared with baseline. At W4, the CGT group andPWSTT group had a significant increase in the POMA gaitscore (suggesting improvement) compared with baseline. Atthe same time, an improvement in the POMA balance score,compared with the baseline, was seen only in the PWSTTgroup. At W4, the control group did not show any signifi-cant change compared with baseline.

In the group comparison, the POMA gait score signifi-cantly improved at W4 in the CGT and PWSTT groupscompared with the control group. In addition, the gait scorewas significantly better in the PWSTT group compared withthe CGT group at W4. At W4, the POMA balance scoresignificantly improved only in the PWSTT group comparedwith the control group (P ¼ .001).

The effect size for PWSTT (partial h2 value for gaitscore ¼ 0.737 and balance score ¼ 0.505) was greater.The effect size for CGT on gait score (partial h2 ¼ 0.245)was greater, and no effect was found on the balancescore (partial h2 ¼ 0.063). The control group did notshow any effect on gait score (partial h2 ¼ 0.007) andbalance score (partial h2 ¼ 0.001). A good test-retestvalue was found for both the BBS (interclass correla-tion coefficient [ICC] ¼ 0.98) and the Tinetti POMA(ICC ¼ 0.98).

Dynamic Posturography

Balance Indices. The balance indices are given in Figure 2and Table 2. The OBI showed a significant group effect(F2,114 ¼ 3.059; P ¼ .039) and time effect (F2,114 ¼ 4.359;P ¼ .015). Similarly, API also showed significant group effect(F2,114 ¼ 4.271; P ¼ .009) and time effect (F 2,114 ¼ 3.716;P ¼ .027). MLI showed a significant group with time inter-action (F4,114 ¼ 5.557; P ¼ .002).

Group, Time, and Interaction Effect

Intergroup Comparison at Different Time Points (P Values)

Groups BL W2 W4

Group effect: F ¼ .296; P ¼ .745Time effect: F ¼ 50.013; P < .001Interaction: F ¼ 27.019; P < .001

CT-CGT .999 .999 .737CT-PWSTT .609 .999 .001CGT -PWSTT .999 .999 .040

Group effect: F ¼ 4.672; P ¼ .013Time effect: F ¼ 4.359; P ¼ .015Interaction: F ¼ 1.949; P ¼ .107

CT-CGT .999 .120 .120CT-PWSTT .999 .075 .002CGT-PWSTT .999 .999 .433



Group effect F ¼ 5.229; P ¼ .008Time effect: F ¼ 2.489; P ¼ .087Interaction: F ¼ 4.557; P ¼ .002

CT-CGT .999 .383 .227CT-PWSTT .999 .025 <.001CGT-PWSTT .999 .999 .020

Group effect: F ¼ .591; P ¼ .557Time effect: F ¼ 67.443; P < .001Interaction: F ¼ 27.098; P < .001


Group effect: F ¼ 5.385; P ¼ .007Time effect: F ¼ 81.093; P < .001Interaction: F ¼ 32.813; P < .001

CT-CGT .154 .196 <.001CT-PWSTT .999 .903 <.001CGT-PWSTT .444 .999 .026

Group effect:F ¼ .968; P ¼ .368Time effect: F ¼ 24.944; P < .001Interaction: F ¼ 12.679; P < .001


Table 2. Continued

PM&R Vol. 6, Iss. 1, 2014 27

At W2, none of the groups showed significant changesin any of the balance indices compared with the baseline.At W4, significant changes were noted only in the PWSTTgroup. In this group, only the OBI and MLI showed a sig-nificant decrease in score from baseline (suggesting im-provement). In group comparison, a significant improvementin the PWSTT group was found compared with the controlgroup at W4 (OBI, P ¼ .002; API, P ¼ .003; MLI, P < .001).In addition, the MLI was significantly better in the PWSTTgroup than in the CGT group (P ¼ .020) at W4.

The effect size for PWSTT was larger for OBI (partialh2 ¼ 0.229) and MLI (partial h2 ¼ 0.274) and medium forAPI (partial h2 ¼ 0.103). The control group (partial h2 valuefor OBI ¼ 0.001, API ¼ 0.054, and MLI ¼ 0.027) and CGTgroup (partial h2 value for OBI ¼ 0.022, API ¼ 0.029, andMLI ¼ 0.006) showed minimum or no effect on balance.

LOS (Table 3). The time to complete the LOS task showeda significant group (F ¼ 4.914, P ¼ .011) and time effect(F ¼ 6.494, P ¼ .008). A significant group with time

interaction was found in the total LOS scores (F ¼ 3.781,P ¼ .017), as well as in the following subscores: FW(F ¼ 3.064, P ¼ .023); LT (F ¼ 3.879, P ¼ .010); BW-RT(F ¼ 4.012, P ¼ .004); and BW-LT (F ¼ 2.996, P ¼ .022). Asignificant time effect was noted in the following LOS sub-scores: RT (F ¼ 5.485, P ¼ .007); FW-RT (F ¼ 4.492,P ¼ .015); and FW-LT (F ¼ 3.489, P ¼ .034). Similarly, afew subscores also showed significant group effect: RT(F ¼ 7.721, P < .001); FW-RT (F ¼ 5.721, P ¼ .005); FW-LT (F ¼ 9.009, P < .001); and BW (F ¼ 12.431, P < .001).The comparisons of LOS subscores are provided in Figure 3.

At W2, none of the groups showed any significantchanges in direction-wise analysis of LOS scores comparedwith baseline. At W4, only the PWSTT group had a signif-icant increase in all the LOS subscores (Figure 3) and asignificant reduction in the time to complete the LOS task(Table 3).

In group comparisons, the PWSTT group showed a sig-nificant improvement in the total LOS score and the LT andBW-RT subscores compared with the control group at W2.

Figure 2. The pattern of mean changes in balance indices in all the groups. W2 ¼ after 2 weeks; W4 ¼ after 4 weeks; OBI ¼ overallbalance index; API ¼ anteroposterior index; MLI ¼ mediolateral index; PWSTT ¼ partial weightesupported treadmill training. yP < .05,yyP < .01 between the groups; **P < .01; *P < .05 within PWSTT group comparison. No significant within-group changes were found incontrol subjects and the CGT group.


In addition, the PWSTT group also demonstrated significantimprovement of the total LOS scores and all the subscorescompared with control subjects at W4. The total LOS scoreand FW, BW, RT, BW-RT, and BW-LT scores were betterthan the CGT group scores at W4. The CGT group did notshow any significant change compared with the controlgroup at either W2 or W4.

The effect size for PWSTT (partial h2 value for FW ¼0 .245; BW ¼ 0.122; RT ¼ .242; LT ¼ 0.280; FW-RT ¼ 0.155, FW-LT ¼ 0.207; BW-RT ¼ 0.392, andBW-LT ¼ 0.307) was medium to large. The control group(partial h2 value for FW ¼ 0.012; BW ¼ 0.008; RT ¼ 0.021;LT ¼ 0.011; FW-RT ¼ 0.012, FW-LT ¼ 0.002; BW-RT ¼0.004; and BW-LT ¼ 0.001) and CGT group (partial h2 valuefor FW ¼ 0.051; BW ¼ 0.022; RT ¼ 0.034; LT¼ 0.067;FW-RT ¼ 0.021, FW-LT ¼ 0.022; BW-RT ¼ 0.061; andBW-LT ¼ 0.015) showed minimum or no effect on balance.

The ICC values for the balance indices OBI (0.96), API(0.86), MLI (0.86), and the direction-specific LOS scores,that is, FW (0.93), BW (0.93), RT (0.94), LT (0.95), FW-RT(0.94), FW-LT (0.95), BW-RT (0.94), and BW-LT (0.91),showed good inter-rater reliability.

The dynamic posturography measure did not have anycorrelation with the clinical severity UPDRS-III motor scoreand the Tinetti POMA gait component score. BBS showed asignificant positive correlation with FW (r ¼ 0.327, P ¼.011), BW-RT (r ¼ 0.334, P ¼ .009), and BW-LT (r ¼0.293, P ¼ .023) direction LOS scores. The Tinetti POMAbalance component had a significant positive correlationwith BW (r ¼ 0.267, P ¼ .039) and BW-LT (r ¼ 0.392, P ¼.002) direction LOS scores.

In brief, the study showed that 4 weeks of PWSTTsignificantly improved the UPDRS-III motor score, balanceindices (OBI, API, and MLI), LOS in 8 directions, TinettiPOMA gait score, and balance component scores. The CGTgroup showed significant improvement in motor score andthe gait component of POMA (but not in the balancecomponent). BBS showed significant improvement in thePWSTT and CGT groups compared with baseline; however,

no group differences were observed. No significant changeswere observed in dynamic posturography parameters incontrol subjects and the CGT group.

DISCUSSION

The results of the present study show that improvement inthe UPDRS-III motor score, MLI, POMA gait score, and LOStotal score was significantly better in the PWSTT group thanin the CGT group. Although the gait score improved in theCGT group, improvement in balance was seen only in thePWSTT group. Various techniques have been used to assessbalance control in persons with PD, such as posturography,force platforms, kinematics, electromyography techniques,and clinical scales [17,18]. Dynamic posturography is usedto evaluate 2 components of balance: dynamic balanceindices and LOS. Dynamic balance indices reflect the inte-gral control of proprioceptive refiexes required to producetimely coordinated movement to retain balance and equi-librium [19]. Poor neuromuscular control was indicated bya large variance quantified as an increased dynamic balanceindices score [13,14]. Although a number of studies haveaddressed the effects of therapeutic interventions on balancein persons with PD, limited randomized controlled trialshave been reported [6,20-22].

The PWSTT group showed significant improvement in allthe balance indices compared with the control group.However, only the MLI is significantly better comparedwith the CGT group at W4. API did not show significantchanges compared with the CGT group, which points to thedifficulty in balance training in the anteroposterior direction.Improvement of MLI at W2 compared to control explainsthe early trainability of balance in the mediolateral directionrather than the anteroposterior direction. In addition, wespeculate that this phenomenon also may contribute to theincreased frequency of falls in the anteroposterior directionin patients with PD.

Although both CGT and PWSTT groups were instructedto perform the training at a self-selected comfortable fast

PM&R Vol. 6, Iss. 1, 2014 29

walking speed, the subjects in the PWSTT group preferred arelatively faster walking speed compared with the subjectsin the CGT group. When subjects are trained with CGT,they may use a wider walking base of support (BOS) thansubjects trained with a treadmill. Using a smaller walkingBOS is a more demanding task that could have contributedto a greater balance control after PWSTT training. However,CGT could be associated with walking with a relativelylarger walking BOS, which is a less demanding task in termsof balance control, and therefore they showed improvementin gait only. It is likely that training the subjects with atreadmill (with a relatively smaller walking BOS) results inmore demand on weight transfer in the mediolateral di-rection. In addition, the anteroposterior direction trans-lation is easily obtained through treadmill motion. Thesefactors probably resulted in improved balance in themediolateral direction than in the anteroposterior directionin the PWSTT group.

Although the PWSTT reduces trunk motion, the doseof walking, the walking speed, and training in a relativelysmall walking BOS may contribute to the improvementof balance measures in the PWSTT group compared withthe CGT group. However, it is also possible that similarbenefits can be obtained through treadmill trainingwithout partial weight support. No conclusions can bedrawn about this possibility based on the current studybecause of a lack of a group undergoing treadmill trainingwithout partial weight support. However, the results favorthe beneficial effects of PWSTT on balance comparedwith CGT.

LOS reflects the interlimb coordination based ondifferent task requirements in different directions, and theincrease in LOS scores suggests a good control of balancetoward the specific directions [13,14]. A significant im-provement was found in the overall LOS scores and in the 8direction-wise LOS components in the PWSTT group after4 weeks of training, which suggests that PWSTT increasesthe LOS whereas CGT does not influence the LOScomponent after 4 weeks of training. Reduced LOS scoreshave been reported earlier in persons with PD [14,23],which might be the result of abnormal temporal features inbalancing strategy during implementation of the vision-dependent behavior in persons with PD and impairedanticipatory motor strategies [24]. The present study isprobably the first to report the influence of PWSTT on theLOS in patients with PD. Studies have reported direction-specific postural instability in patients with PD [14,17,25].By using posturography in patients with early-stage PD,Horak et al [25] reported impairment of balance in back-ward directions and different diagonal directions in patientswith PD. Our study has demonstrated significant improve-ment in LOS scores in patients in all 8 directions after4 weeks of PWSTT. In contrast, CGT does not have sig-nificant benefit in direction-specific balance improvementafter 4 weeks of training.

On the clinical measures, the UPDRS-III motor scoreshowed significant improvement in the PWSTT groupcompared with the control group and significantly betterimprovement than the CGT group at W4. Similarly, thePWSTT group gait score improvement at W4 was signifi-cantly better than that of the CGT group, which probablyindicates that PWSTT is better than CGT in improvingmotor scores and gait in patients with PD.

On the clinical balance measurement scale (POMA), asignificant improvement in the gait score and balance scorewas found in the PWSTT group compared with the controlgroup after 4 weeks of training. Although the CGT grouphad significant improvements in the gait score, the balancescore did not show any significant changes after 4 weeks oftraining, which indicates that although the gait componentcan be improved by either CGT or PWSTT, the balancecomponent requires PWSTT training.

Although BBS demonstrated a significant group with timeinteraction effect, adjusted pair-wise group comparison didnot reveal significant group differences, which suggests thatphysical trainingeinduced improvement in balance asassessed by BBS may not be sensitive enough to detect dif-ferences between PWSTT, CGT, and control subjects in a4-week period. Consistent with these findings, Toole et al[6] and Ashburn et al [20] did not find any improvement inBBS but reported improvement in the functional reach test.

The exact pathogenesis of postural instability in personswith PD is still unknown. The underlying mechanismsare complex, with involvement of different neural structures[2,3]. Blin et al [2] reported levodopa (L-DOPA)esensitiveand L-DOPAeresistant components of walking and otherpostural tasks. Bloem et al [3] concluded that balanceimpairment in persons with PD is related to a combination ofboth dopaminergic and nondopaminergic lesions. We spec-ulate that the improvement of balance observed in the presentstudy was a result of improvement of both L-DOPAesensitive and L-DOPAeresistant components of balance.However, the exact mechanism of improvement is unknown.

A better functional outcome with body weight supportthan without weight support has been reported in patientswho have had a stroke [10]. This finding supports ourobservation that partial weight support in treadmill trainingmay contribute to the improvement of balance in thePWSTT group. Among the tests of dynamic posturographymeasures, LOS produced noticeable changes as mentionedpreviously compared with the balance indices, which in-dicates that LOS may serve as an early predictor outcomemeasure in evaluating short-term rehabilitation in personswith PD. Hence among the various parameters that assessbalance, dynamic posturography may predict the smallchanges in balance. PWSTT can be an intervention of choice,especially in persons with PD who have direction-specificbalance impairment.

Using a similar methodology in patients with PD, Miyaiet al [4] reported greater improvement in activities of daily

Table 3. Direction-wise analysis of limits of stability components in all the 3 groups

LOS Group BL After 2 Weeks (W2), Mean ± SD After 4 Weeks (W4), Mean ± SD

Time to complete LOS test CT 187.7 � 46.2 180.9 � 41.2 178.1 � 36.6CGT 167.3 � 32.7 165.2 � 36.9 155.9 � 31.1PWSTT 175.9 � 42.1 152.90 � 46.4 131.3 � 45.1

Total LOS score CT 20.9 � 9.0 22.6 � 8.5 21.8 � 9CGT 25.4 � 8.8 27.0 � 8.7 29.9 � 7.8PWSTT 26.8 � 9.2 31.9 � 10.6 42.2 � 14.7

FW CT 27 � 13.2 29.5 � 15.7 26.9 � 11.1CGT 27.1 � 12.9 29.6 � 10.8 34.9 � 12.7PWSTT 28.6 � 12.4 37.4 � 12.6 48.9 � 20.6

BW CT 15.6 � 11.4 14.7 � 7.9 13.1 � 6.7CGT 18.2 � 12.2 23.1 � 17.6 20.6 � 7.5PWSTT 19.7 � 9.5 22.8 � 14.9 30.2 � 16.3

RT CT 23.9 � 10.5 28.8 � 16.8 26.4 � 12.7CGT 27.6 � 11.9 32.3 � 13.2 35 � 16.8PWSTT 32.8 � 17.3 33.1 � 13.9 49.3 � 22

LT CT 18.9 � 11.5 20.3 � 10.9 18.4 � 9.5CGT 26.1 � 12.2 23.7 � 7.1 28.7 � 10.9PWSTT 21.3 � 9.7 28.4 � 11.6 37.8 � 18.4

FW-RT CT 22.7 � 10.9 25.1 � 16.5 26.3 � 15.5CGT 26.6 � 8.4 27.6 � 10.9 31.2 � 17.8PWSTT 27.2 � 8.8 37.9 � 16.4 39.4 � 18.6

FW-LT CT 24.3 � 16.1 25.7 � 11 24.5 � 11.3CGT 31.7 � 15.1 32.3 � 15.8 35.9 � 10.8PWSTT 30.2 � 14.5 36.2 � 16.6 47.1 � 23.6

BW-RT CT 18.5 � 11.4 19.9 � 11.9 20.8 � 11.5CGT 23.9 � 9.3 26.0 � 13.4 31.0 � 12.4PWSTT 25.2 � 12.1 33.2 � 17.3 48.4 � 19.9

BW-LT CT 22.2 � 9.8 21.6 � 12.3 22.2 � 12.8CGT 22.7 � 9.3 25.4 � 8.5 26.0 � 5.7PWSTT 25.7 � 11.2 29.9 � 14.7 40.9 � 16.3

LOS ¼ limits of stability; BL ¼ baseline; W2 ¼ week 2; W4 ¼ week 4; CT ¼ control; CGT¼ conventional gait training; PWSTT ¼ partialweightesupported treadmill training; FW ¼ forward; BW ¼ backward; RT ¼ right; LT ¼ left; RMANOVA ¼ repeated-measuresanalysis of variance.F and P values were expressed after RMANOVA with Bonferroni adjustment; group-specific and time-specific comparisons are

adjusted for multiple comparisons. Degree of freedom for group effects ¼ 2.57, interaction effects ¼ 4.114, and time effect ¼ 2.114.


living, motor performance, and ambulation after PWSTT with20% unloading of weight than with conventional physio-therapy. A follow-up study by the same group showed thebeneficial effects lasting for 6 months [5]. However, unlikeus, Miyai et al [4,5] did not address any balance parameters.Toole et al [26] reported improved balance in patientswith PD after 10 weeks of a strength and balance trainingprogram compared with a group of nonexercising PD controlsubjects, which was similar to our observations. However, inour study, specific balance exercises were not included.

A meta-analysis of treadmill training in patients withPD (H&Y stages 1-3) concluded that treadmill training

improved speed of gait and walking distance [7]. However,the dynamic balance and the components of LOS that weaddressed in our study have not been addressed in previousstudies. The meta-analysis reported the absence of adverseevents in treadmill training, an observation with which weagree based on the results of our study.

The added advantage of PWSTT may also result from (1)symmetrical continuous sensory stimuli from the movingtreadmill, enhancing locomotor pattern generators [27];(2) partial weight support, providing a sense of securityregarding falling and facilitating free leg movements [4,5];and (3) getting positive feedback from every step, which

Group, Time and Interaction Effects

Intergroup Comparison (P Values)

Groups BL W2 W4



Group effect: F ¼ 16.767; P < .001Time effect: F ¼ 9.062; P ¼ .001Interaction: F ¼ 3.781; P ¼ .017




Group effect: F ¼ 12.431; P < .001Time effect: F ¼ 1.140; P ¼ .322Interaction:F ¼ 1.694; P ¼ .160




Group effect: F ¼ 9.120; P < .001Time effects: F ¼ 5.026; P ¼ .013Interaction: F ¼ 3.879; P ¼ .010


Group effect: F ¼ 5.721; P ¼ .005Time effects: F ¼ 4.492; P ¼ .015Interaction: F ¼ 1.017; P ¼ .381


Group effect: F ¼ 9.009; P < .001Time effect: F ¼ 3.489; P ¼ .034Interaction: F ¼ 1.159; P ¼ .107


Group effect: F ¼ 15.498; P < .001Time effect: F ¼ 11.615; P <.001Interaction: F ¼ 4.012; P ¼ .004


Group effect: F ¼ 10.112; P <.001Time effect: F ¼ 4.990; P ¼ .008Interaction: F ¼ 2.996; P ¼ .022


Table 3. Continued

PM&R Vol. 6, Iss. 1, 2014 31

helps in autocorrection of step length and enhances motorlearning [28]. It has been also postulated that these cueingstrategies help bypass the defective basal ganglia by usingalternative pathways unaffected by PD [29]. Attentionstrategies offer an alternative to external cues because theyrely more on cognitive mechanisms of motor control andare internally generated, which help improve motor per-formance [29]. Other possible explanations include task-specific motor learning or improvement in posturalreflexes [30], exercise-induced and activity-dependent neuralplasticity (ie, neurogenesis, synaptogenesis, and molecularadaptation) [31,32], and normalization of corticomotor

excitability/cortical reorganization, especially in the supple-mentary motor area in persons with PD [33].

The limitations of our study include the lack of a patientgroup subjected to training on a treadmill without partialweight support, lack of follow-up, and lack of blinding of theparticipants to the assessment. It is possible that the “Haw-thorne effect” can influence the subject’s performance whileperforming the outcome measures. In the current study, allthree groups received a similar level of attention. However,no significant improvement was observed in the measures ofbalance outcome in the control and CGT groups, whichimplies that the influence of attention on balance would have

Figure 3. Limits of stability scores at baseline and after 4 weeks (W4) in all the groups. CGT ¼ conventional gait training; PWSTT ¼partial weightesupported treadmill training; FW ¼ forward; BW ¼ backward; RT ¼ right; LT ¼ left; FWRT ¼ forward right; FWLT ¼ forwardleft; BWRT ¼ backward right; BWLT ¼ backward left. *P < .05; **P < .01.


achieved a ceiling effect during the pretraining assessment.The posttraining assessment results were possibly due totraining effect.

CONCLUSIONS

Four weeks of CGT and PWSTT improved gait in patientswith PD. The UPDRS-III motor score and balance (both thebalance indices and the LOS in 8 directions) improved onlyin the PWSTT group. Improvement of the UPDRS-motorscore, MLI score, and LOS total score was significantly betterin the PWSTT group compared with the CGT group.Therefore we conclude that PWSTT can be a recommendedas an intervention of choice in patients with PD, especially inpatients with balance impairment.

ACKNOWLEDGMENT

We thank Dr. K. Thennarasu, Department of Biostatistics,National Institute for Mental Health and Neurosciences, andDr. K. P. Suresh, Scientist (Biostatistics), National Institute ofAnimal Nutrition and Physiology, Bangalore, India, for theirstatistical advice.

REFERENCES1. Morris ME. Movement disorders in people with Parkinson disease: A

model for physical therapy. Phys Ther 2000;80:578-597.2. Blin O, Ferrandez AM, Pailhous J, Serratrice G. Dopa-sensitive and

dopa-resistant gait parameters in Parkinson’s disease. J Neurol Sci1991;103:51-54.

3. Bloem BR, Beckley DJ, van Dijk JG, Zwinderman AH, Remler MP,Roos RA. Influence of dopaminergic medication on automatic posturalresponses and balance impairment in Parkinson’s disease. Mov Disord1996;11:509-521.

4. Miyai I, Fujimoto Y, Ueda Y, et al. Treadmill training with body weightsupport: Its effect on Parkinson’s disease. Arch Phys Med Rehabil 2000;81:849-852.

5. Miyai I, Fujimoto Y, Yamamoto H, et al. Long-term effect of bodyweight-supported treadmill training in Parkinson’s disease: A ran-domized controlled trial. Arch Phys Med Rehabil 2002;83:1370-1373.

6. Toole T, Maitland CG, Warren E, Hubmann MF, Panton L. The effectsof loading and unloading treadmill walking on balance, gait, fall risk,and daily function in Parkinsonism. NeuroRehabilitation 2005;20:307-322.

7. Mehrholz J, Friis R, Kugler J, Twork S, Storch A, Pohl M. Treadmilltraining for patients with Parkinson’s disease. Cochrane Database SystRev 2010;1:CD007830.

8. Protas EJ, Mitchell K, Williams A, Qureshy H, Caroline K, Lai EC. Gaitand step training to reduce falls in Parkinson’s disease. Neuro-Rehabilitation 2005;20:183-190.

9. Cakit BD, Saracoglu M, Genc H, Erdem HR, Inan L. The effects of in-cremental speed-dependent treadmill training on postural instabilityand fear of falling in Parkinson’s disease. Clin Rehabil 2007;21:698-705.

10. Visintin M, Barbeau H, Korner-Bitensky N, Mayo NE. A new approachto retrain gait in stroke patients through body weight support andtreadmill stimulation. Stroke 1998;29:1122-1128.

11. Visintin M, Barbeau H. The effects of body weight support on thelocomotor pattern of spastic paretic patients. Can J Neurol Sci 1989;16:315-325.

12. Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of clinical diagnosisof idiopathic Parkinson’s disease: A clinico-pathological study of 100cases. J Neurol Neurosurg Psychiatry 1992;55:181-184.

13. Ganesan M, Pasha SA, Pal PK, Yadav R, Gupta A. Direction specificpreserved limits of stability in early progressive supranuclear palsy: Adynamic posturographic study. Gait Posture 2012;35:625-629.

14. Ganesan M, Pal PK, Gupta A, Sathyaprabha TN. Dynamic posturog-raphy in evaluation of balance in patients of Parkinson’s disease withnormal pull test: Concept of a diagonal pull test. Parkinsonism RelatDisord 2010;16:595-599.

15. Berg KO, Wood-Dauphinee SL, Williams JI, Maki B. Measuring balancein the elderly: Validation of an instrument. Can J Public Health 1992;83:S7-S11.

16. Tinetti ME. Performance-oriented assessment of mobility problems inelderly patients. J Am Geriatr Soc 1986;34:119-126.

17. Carpenter MG, Allum JH, Honegger F, Adkin AL, Bloem BR. Posturalabnormalities to multidirectional stance perturbations in Parkinson’sdisease. J Neurol Neurosurg Psychiatry 2004;75:1245-1254.

18. Jacobs JV, Horak FB, Tran VK, Nutt JG. Multiple balance tests improvethe assessment of postural stability in subjects with Parkinson’s disease.J Neurol Neurosurg Psychiatry 2006;77:322-326.

http://refhub.elsevier.com/S1934-1482(13)01035-6/sref1























































PM&R Vol. 6, Iss. 1, 2014 33

19. Nashner LM. Analysis of movement control in man using the movableplatform. Adv Neurol 1983;39:607-619.

20. Ashburn A, Fazakarley L, Ballinger C, Pickering R, McLellan LD,Fitton C. A randomised controlled trial of a home based exerciseprogramme to reduce the risk of falling among people with Parkinson’sdisease. J Neurol Neurosurg Psychiatry 2007;78:678-684.

21. Hirsch MA, Toole T, Maitland CG, Rider RA. The effects of balancetraining and high-intensity resistance training on persons with idio-pathic Parkinson’s disease. Arch Phys Med Rehabil 2003;84:1109-1117.

22. Dibble LE, Addison O, Papa E. The effects of exercise on balance inpersons with Parkinson’s disease: A systematic review across thedisability spectrum. J Neurol Phys Ther 2009;33:14-26.

23. Schieppati M, Hugon M, Grasso M, Nardone A, Galante M. The limitsof equilibrium in young and elderly normal subjects and in parkinso-nians. Electroencephalogr Clin Neurophysiol 1994;93:286-298.

24. De Nunzio AM, Nardone A, Schieppati M. The control of equilibriumin Parkinson’s disease patients: Delayed adaptation of balancing strategyto shifts in sensory set during a dynamic task. Brain Res Bull 2007;74:258-270.

25. Horak FB, Dimitrova D, Nutt JG. Direction-specific postural instabilityin subjects with Parkinson’s disease. Exp Neurol 2005;193:504-521.

26. Toole T, HirschMA, Forkink A, LehmanDA,Maitland CG. The effects ofa balance and strength training program on equilibrium in Parkinsonism:A preliminary study. NeuroRehabilitation 2000;14:165-174.

27. Ditor DS, Kamath MV, MacDonald MJ, Bugaresti J, McCartney N,Hicks AL. Effects of body weight-supported treadmill training on heartrate variability and blood pressure variability in individuals with spinalcord injury. J Appl Physiol 2005;98:1519-1525.

28. Fernando CK, Basmajian JV. Biofeedback in physical medicine andrehabilitation. Biofeedback Self Regul 1978;3:435-455.

29. Rubinstein TC, Giladi N, Hausdorff JM. The power of cueing tocircumvent dopamine deficits: A review of physical therapy treatmentof gait disturbances in Parkinson’s disease. Mov Disord 2002;17:1148-1160.

30. Platz T, Brown RG, Marsden CD. Training improves the speed of aimedmovements in Parkinson’s disease. Brain 1998;121:505-514.

31. Fisher BE, Petzinger GM, Nixon K, et al. Exercise-induced behavioralrecovery and neuroplasticity in the 1-methyl-4-phenyl-1,2,3,6-tetra-hydropyridine-lesioned mouse basal ganglia. J Neurosci Res 2004;77:378-390.

32. Fox CM, Ramig LO, Ciucci MR, Sapir S, McFarland DH, Farley BG.The science and practice of LSVT/LOUD: Neural plasticity-principledapproach to treating individuals with Parkinson disease and otherneurological disorders. Semin Speech Lang 2006;27:283-299.

33. Fisher BE, Wu AD, Salem GJ, et al. The effect of exercise trainingin improving motor performance and corticomotor excitability inpeople with early Parkinson’s disease. Arch Phys Med Rehabil 2008;89:1221-1229.

















































erikhornung

Typewritten Text

FN: 17-510 12/17

effect of partial weight-supported treadmill gait training ... · training on balance in patients...

Documents