DEVELOPMENTAL MEDICINE & CHILD NEUROLOGY ORIGINAL ARTICLE

Effects of a supported speed treadmill training exercise program on

impairment and function for children with cerebral palsy
THERESE E JOHNSTON 1 | KYLE E WATSON 2 | SANDY A ROSS 3 | PHILIP E GATES 4 | JOHN P GAUGHAN 5 |
RICHARD T LAUER 6 | CAROLE A TUCKER 6 | JACK R ENGSBERG 7

1 Department of Physical Therapy, University of the Sciences, Philadelphia, PA. 2 Department of Physical Therapy, Shriners Hospitals for Children, Philadelphia, PA. 3 School of
Health Professions, Maryville University, St Louis, MO. 4 Medical Staff, Shriners Hospitals for Children, Shreveport, LA. 5 Department of Physiology, Temple University,
Philadelphia, PA. 6 Department of Physical Therapy, Temple University, Philadelphia, PA. 7 School of Medicine, Washington University, St Louis, MO, USA.
Correspondence to Dr Therese E Johnston at Department of Physical Therapy, Sansom College of Health Sciences, University of the Sciences in Philadelphia, 600 South 43rd Street, Box 39, Philadelphia,
PA 19104, USA. E-mail: t.johnston@usp.edu

This article is commented on by Low on page 680 of this issue.

PUBLICATION DATA AIM To compare the effects of a supported speed treadmill training exercise program (SSTTEP)
Accepted for publication 16th March 2011. with exercise on spasticity, strength, motor control, gait spatiotemporal parameters, gross motor
Published online 17th June 2011. skills, and physical function.
METHOD Twenty-six children (14 males, 12 females; mean age 9y 6mo, SD 2y 2mo) with spastic
ABBREVIATIONS cerebral palsy (CP; diplegia, n=12; triplegia, n=2; quadriplegia n=12; Gross Motor Function Classifi-
MCID Minimum clinically important cation System levels II–IV) were randomly assigned to the SSTTEP or exercise (strengthening)
difference group. After a twice daily, 2-week induction, children continued the intervention at home 5 days a
PBWSTT Partial body-weight-supported week for 10 weeks. Data collected at baseline, after 12-weeks’ intervention, and 4 weeks after the
treadmill training intervention stopped included spasticity, motor control, and strength; gait spatiotemporal parame-
PODCI Pediatric Outcomes Data Collection
ters; Gross Motor Function Measure (GMFM); and Pediatric Outcomes Data Collection Instrument
Instrument
SSTTEP Supported speed treadmill training (PODCI).
exercise program RESULTS Gait speed, cadence, and PODCI global scores improved, with no difference between
groups. No significant changes were seen in spasticity, strength, motor control, GMFM scores, or
PODCI transfers and mobility. Post-hoc testing showed that gains in gait speed and PODCI global
scores were maintained in the SSTTEP group after withdrawal of the intervention.
INTERPRETATION Although our hypothesis that the SSTTEP group would have better outcomes
was not supported, results are encouraging as children in both groups showed changes in
function and gait. Only the SSTTEP group maintained gains after withdrawal of intervention.

Developing the ability to walk is an important functional goal review on PBWSTT in pediatric rehabilitation.7 Two
for children with cerebral palsy (CP). A relationship has been reviews4,6 concluded that PBWSTT may be safe and effective
shown between locomotor ability and the performance of for improving gait speed, and one concluded that it may be
activities of daily living and social roles.1 Children with CP in beneficial for improving gross motor skills.4 However, a third
lower levels on the Gross Motor Function Classification review concluded that there is insufficient evidence to deter-
System (GMFCS) and, thus, greater mobility impairments mine if PBWSTT leads to improvements for children with
display a decrease in overall daily walking activity2 and an CP.5 In the review of PBWSTT in pediatric rehabilitation,7 it
increased energy cost of walking3 than those in higher was concluded that the evidence is weak for outcomes for chil-
GMFCS levels. Therefore, interventions targeted at improv- dren with CP, and that randomized trials are needed to exam-
ing walking ability in children in lower GMFCS levels are ine issues such as effectiveness and dosing. These reviews
important for developing methods to improve overall function suggest that research that is more stringent is needed to
and activity. determine the effectiveness of PBWSTT for children with CP.
Partial body-weight-supported treadmill training (PBWSTT) Changes in gait spatiotemporal parameters after PBWSTT
is more widely used in the rehabilitation of populations with have been reported. Two studies reported significant increases
neurological conditions than previously, including children in gait speed for children aged 6 to 14 years classified in
with CP. Literature on PBWSTT in CP consists mainly of case GMFCS level I8 (walks without restrictions but has limitations
reports and small non-randomized trials with or without com- in more advanced gross motor skills)9 or ages 6 to 14 years
parison groups. Recently, three separate reviews of PBWSTT classified in GMFCS levels III and IV10 (III, walks with
in children with CP have been published4–6 as well as one assistive mobility devices but has limitations walking outdoors

742 DOI: 10.1111/j.1469-8749.2011.03990.x ª The Authors. Developmental Medicine & Child Neurology ª 2011 Mac Keith Press
 14698749, 2011, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2011.03990.x by Cochrane Chile, Wiley Online Library on [01/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
and in the community; IV, self-mobility with limitations but What this paper adds
is transported or uses power mobility outdoors and in the • This study is, to our knowledge, the only multi-site randomized controlled trial
community).9 Another study reported no change in gait speed to compare a partial body-weight-supported treadmill training intervention
but significant increases in stride length for 3- to 6-year-old with an exercise intervention.
children in GMFCS levels II (walks without assistive mobility • Children with CP can make gains in gait and function after an intensive
12-week intervention.
devices but has limitations walking outdoor and in the com- • No differences were found in strength, spasticity, or motor control as a result
munity)9 and III. Therefore, PBWSTT interventions have the of the treadmill training.
potential to impact upon these parameters. Several studies • Only the SSTTEP group maintained gains after withdrawal of the intervention.
have reported gains in Gross Motor Function Measure
(GMFM) scores after PBWSTT across various ages and 68kg; equipment limits); age 6 to 13 years; and ability to follow
GMFCS levels, with most change reported in dimension E multiple-step commands during training and data collection.
(walking, running, and jumping activities).4,6 Children were excluded for the following reasons: a medical
Spasticity, muscle weakness, and impaired motor control condition that would be negatively affected by exercise; lower
are key impairments associated with CP. Knowledge of these extremity orthopedic surgery in past year, botulinum toxin A in
impairments is a part of any clinical examination or investiga- the past 6 months, or dorsal rhizotomy in past 2 years; flexion
tion of a child with CP because they provide key information contractures greater than 30 at the hip or greater than 20 at
about the physical state of the child. Furthermore, quantifica- the knee or plantarflexion contractures greater than 15 with
tion of impairments is particularly valuable if measures may be the knee extended; intrathecal baclofen; or athetoid or mixed
altered by an intervention. One small non-randomized study11 types of CP.
reported no change in spasticity as assessed by the Ashworth Institutional review board approval was obtained at each
scale or in motor control using a clinical ordinal scale after a site. Parents (one parent per child) and children, ages 7 and
12-week PBWSTT program. However, no study has used older, signed the approved consent and assent forms respec-
more rigorous objective measures to assess the effects of tively. Authorization required by the Health Insurance Porta-
PBWSTT on strength, spasticity, or motor control. bility and Accountability Act was obtained from parents.
As children in GMFCS levels III and IV present with more
limitations in functional walking than those in higher GMFCS Intervention
levels, these children may experience greater benefits after an Children were randomly assigned to the SSTTEP or exercise
intervention as small improvements in gait speed may signifi- group using a block randomization schedule at each site using
cantly improve their functional abilities.4 Therefore, a blocks of eight group assignments (four of each intervention
PBWSTT program may be ideal to enable these children to per block). Children in each group underwent 2 weeks of
achieve consistent walking patterns and speed. intensive training, in which two 30-minute sessions were con-
The purpose of this randomized controlled trial was to ducted 5 days a week under the supervision of a physical ther-
compare the effects of a home-based PBWSTT protocol, the apist at one of the study sites or the child’s home. The goals of
supported speed treadmill training exercise program (SST- the induction period were to establish each child’s program
TEP), with a home-based exercise program for spasticity, and to train the parents on its implementation. After the
strength, motor control, gait speed, gross motor skills, and induction period, parents took home all equipment and con-
physical function. The goal of the SSTTEP intervention was ducted the intervention for 30 minutes, five times per week
to encourage faster walking speeds (‘speed’ component in pro- for 10 weeks. This protocol was based on previous unpub-
tocol title) while walking on the treadmill with partial body- lished pilot work within our laboratory that suggested an
weight support (‘support’ component of protocol title). The intensive induction followed by 5 days per week of treatment
hypothesis was that children in the SSTTEP group would led to better outcomes than a similar program without the
show greater improvements than those in the home-based induction period. The duration of 12 weeks was based on neu-
exercise group. rophysiological training effects; this duration was also reported
in a study that found improvements in functional ambulation
METHOD category and gross motor function in non-ambulatory
Participants children with CP.12
Children with spastic diplegic, triplegic, or quadriplegic CP The SSTTEP group used a home folding treadmill and a
were recruited through and the study conducted at outpatient pediatric suspension walker (Kaye Products, Hillsborough,
clinics at three intervention sites in different regions of the NC, USA) that fitted over the treadmill. Each child was
USA: the Shriners Hospitals for Children in Philadelphia, PA, wore a harness for use with the walker. For the intervention,
Shreveport, LA, and St Louis University and Washington Uni- the physical therapist or parent sat on a therapy bench
versity in St Louis, MO. The following inclusion criteria were (Kaye Products) in front of the child to guide the child’s
used: spastic CP; marginal ambulatory function (defined as leg, if needed. (Fig. 1). Children were allowed to use their
decreased gait velocity <80% of age-expected value regardless ankle–foot orthoses if they were unable to walk overground
of GMFCS level, or GMFCS level III or IV); ability to without them or the therapist determined that the gait pat-
take eight steps independently with or without assistive tern on the treadmill was poor without them. The initial
devices; body weight less than 150 pounds (approximately training speed was based on the child’s baseline gait speed

Function and Gait Following Treadmill Training Therese E Johnston et al. 743
 14698749, 2011, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2011.03990.x by Cochrane Chile, Wiley Online Library on [01/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Data collection
Before and after the 12-week intervention period (2wks’
induction, 10wks’ home training), data were collected at each
child’s primary site by the same evaluator as at baseline to
measure spasticity, strength, motor control, gait spatiotempo-
ral parameters, gross motor skills, and physical function. An
additional data collection session was performed 4 weeks after
the intervention ended to quantify any carryover effects. Two
sites were able to blind the evaluators to group assignment
(representing 16 children) but the third site was unable to
because of personnel issues. All evaluators were blinded to
results and trained on each measure for which written study-
specific protocols were available at each site. Inter- and
intrarater reliability testing was not performed. All evaluators
routinely used these measures within their practice settings
before the study.
Spasticity of the plantarflexors and knee flexors was assessed
with the child in a semi-supine position on a KinCom com-
puterized dynamometer (Chattanooga Group, Chattanooga,
TN, USA).13–15 This test was chosen rather than a more clini-
cal measure such as the Ashworth scale, which has question-
able validity.16,17 Using a computerized dynamometer allows
Figure 1: Photograph of set-up for the supported speed treadmill training the limb to be moved at a constant speed across tests and
exercise program (SSTTEP) group. across participants and meets the criteria for a spasticity test to
measure velocity-dependent resistance to movement.14,17 For
the test, the child’s hip was flexed approximately 25 and the
determined by gait analysis and adjusted as needed based on knee was in full extension. The pelvis, thigh and lower leg
the child’s response while walking on the treadmill. The were secured with straps. For the ankle, the child was
goals for this group during the induction period were to instructed not to move and to remain as relaxed as possible as
decrease body-weight support to less than 30% and then the ankle joint rotated passively from a plantarflexed to a
increase speed toward normal values. Decisions to alter dorsiflexed position (maximum of 40 of plantarflexion and
body-weight support and speed were based on the child’s maximal dorsiflexion for each child). The ankle was rotated
ability to achieve and maintain a foot–flat or heel–toe pat- three times at 5 ⁄ s for gravity correction. For the spasticity
tern, initiate swing and achieve knee extension in stance, tests, 10 repetitions were randomly performed at angular
and obtain consistent foot placement. velocities of 15, 60, 90, and 180 ⁄ s. Values representing the
The exercise group participated in an exercise program amount of work required to move the ankle passively through
based on impairments and functional tasks, with an emphasis the range of motion were determined for each angular velocity
on strengthening exercises and on standing-weight-bearing for each child. The process was similar for the knee, with a
activities. Specific components were forward step-ups, squats, range of motion from 20 to 90 of flexion. Spasticity values of
upper and lower extremity progressive resistive exercise, and the ankle plantarflexor and knee flexor were calculated from
core strengthening. Assistance was provided by the physical the slope of a linear regression of the angular velocity–work
therapist (induction period) or the parents (at home) as values. Test–retest reliability for this test is moderate for chil-
needed, and assistive devices were allowed to be used for the dren with CP.13 One site was unable to collect spasticity data
weight-bearing exercises. The program was determined indi- because the dynamometer was unable to accelerate quickly
vidually for the children, taking into account their individual enough to obtain the desired angular velocities. The sample
standing abilities and overall strength and endurance for each size for the spasticity assessment was 18 (10 in SSTTEP, eight
exercise. However, as with the SSTTEP group, the duration in exercise).
of each child’s session was 30 minutes. Strength was measured using the same set-up as the spastic-
All home training for each group was monitored by each ity assessment.15,18 The child exerted the maximum amount of
site’s primary physical therapist through weekly telephone dorsi- and plantarflexion or knee flexion ⁄ extension concentric
calls to the parents. At that time, decisions were made about force over the range of motion during a 10 ⁄ s isokinetic test.
increasing the speed for the SSTTEP group or advancing the Testing through the range of motion at a slow speed allows
exercise program. Advances in the exercise program first the child time to generate force and provides strength data
included increasing the number of repetitions performed and for more than just one angle, including end-range knee
then adding resistance by cuff weights. Parents of children in extension.18 The maximum gravity-corrected values normal-
each group kept weekly logs to record that each session had ized by dividing by body mass were used as the measure of
been completed. strength. For all strength tests, three trials were performed

744 Developmental Medicine & Child Neurology 2011, 53: 742–750

 14698749, 2011, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2011.03990.x by Cochrane Chile, Wiley Online Library on [01/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
and the highest value from the three trials was defined as the identified. Based on pilot work and the work of
maximal strength. Engsberg et al.,22 an SD of 0.19m ⁄ s was used to determine
Motor control of the quadriceps was tested in the same sample size.
position as spasticity and strength testing.19 Testing for The dependent variables were treated as continuous vari-
motor control in this manner allows an objective measure to ables for all analyses. Means, sums, and SDs are presented for
be obtained,19 and testing one joint in isolation decreases the all variables. The experimental design was a three-factor
complexity of the task, allowing the child to focus on one (treatment group, time, trial) mixed design with repeated mea-
motion. The dynamometer was set to the isokinetic mode, sures on one factor (time). The null hypothesis was that there
with an angular velocity of 10 ⁄ s. A ‘target’ force (20% of the was no difference in the measured parameters among the
maximum value determined during strength testing) was dis- treatment groups or between time periods (0, 12, 16). Before
played on the computer monitor as a horizontal line on a analysis, all data were tested for normality using the Shapiro–
force–angle diagram. The child was then asked to match that Wilk test. The data for the dependent variables were signifi-
force as the knee was moved toward extension from cantly non-normal. To apply analysis of variance (ANOVA)
maximum knee flexion. The standard deviation (SD) of the methods, a ‘normalized-rank’ transformation23 was applied to
force–angle data was used as the outcome value because it the data before analysis. The rank-transformed data were ana-
quantified the variability of the force application over the lyzed using a mixed-model ANOVA for repeated measures
range of motion. followed by multiple comparisons to detect significant individ-
Gait speed, cadence, and stride length were measured by ual mean differences at each time. Multiple pair-wise compari-
three-dimensional motion analysis. Children walked at their sons used an experiment-wise type I error of 0.05. Differences
self-selected speed using their commonly used assistive device between group means, times, etc. (rejection of the null
and ankle–foot orthoses if unable to walk without them. Gross hypothesis) were considered significant if the probability of
motor skills were assessed using dimensions A to E of the chance occurrence was less than or equal to 0.05 using two-
GMFM, a standard criterion-referenced test designed to assess tailed tests.
change in gross motor function in children with CP.20 Physi- The minimum clinically important difference (MCID) has
cal function was measured using the parent report (paper and been calculated for several commonly used measures for chil-
pencil version) of the Pediatric Outcomes Data Collection dren with CP in order to establish the clinical meaningfulness
Instrument (PODCI).21 The PODCI has been shown to have of change.24 Therefore, the MCID for each child was used as a
good validity, reliability, and sensitivity for children aged 2 to secondary measure to begin to understand changes that
18 years. It includes scales of global function, physical sports occurred in the GMFM, PODCI, and the gait spatiotemporal
and activities, transfers and mobility, upper extremity function, parameters. MCIDs reported by Oeffinger et al.24 (PODCI
and comfort. The parent report for the PODCI global func- and gait spatiotemporal parameters) and Wang and Yang25
tion score, and the transfers and mobility scores, were used as (GMFM) were used to examine the data in this study. Oeffinger
the primary measures of physical function. Children were et al.24 only reported data for children classified in GMFCS
required to complete at least 40 out of 50 full sessions (80% levels I to III, so level III MCID data were used for the children
adherence) to participate in data collection. classified in level IV in our study.

Statistical analysis RESULTS

From a priori data analysis, the projected sample size Thirty-eight children were screened and 26 completed the
with a=0.05 (two-tailed test) and a power of 80% was 54 study, with 14 assigned to the SSTTEP group and 12 to the
participants. Gait speed was used as the primary outcome exercise group (Table I and Table SI [supporting material
measure, with a clinically meaningful increase of 0.10m ⁄ s online], and Fig. 2). The SSTTEP group consisted of seven

Table I: Assistive device use and body-weight support (BWS) for the children enrolled in the study

Initial percentage End percentage Initial treadmill End treadmill

Group Assistive device BWS BWS speed (m ⁄ s) speed (m ⁄ s)

SSTTEP None (n=1) 35.4 (SD 14.2) 15.7 (SD 17.0) 0.36 (SD 0.18) 0.64 (SD 0.33)
Cane (n=1)
Anterior walker (n=1)
Posterior walker (n=11)
Exercise None (n=1)
Canes (n=1)
Forearm crutches (n=2)
Posterior walker (n=6)
Gait trainer (n=2)

For the supported speed treadmill training exercise program (SSTTEP) group, percentage BWS and treadmill speed were included for the initial
induction training session and the last home session.

Function and Gait Following Treadmill Training Therese E Johnston et al. 745
 14698749, 2011, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2011.03990.x by Cochrane Chile, Wiley Online Library on [01/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Screened for study (n=38)

Did not meet inclusion

criteria (n=4)

Randomized (n=34)

Allocated to SSTTEP Allocated to exercise

(n=18) (strengthening) group (n=16)
Received allocated Received allocated
intervention (n=16) intervention (n=16)
Did not receive allocated Did not receive allocated
intervention (n=2) intervention (n=0)

Lost to follow-up (n=2) Lost to follow-up (n=4)

Discontinued participation Discontinued participation
part way though intervention part way though intervention
period period

Analyzed (n=14) Analyzed (n=12)

Excluded from analysis Excluded from analysis
(n=0) (n=0)

Figure 2: Consort flow diagram. SSTTEP, supported speed treadmill training exercise program.

males and seven females (mean age 9y 7mo, SD 2y 2mo). Eight exercise, p=0.007) after the intervention, but the gains were
children had diplegic and six quadriplegic CP, and were in only maintained in the SSTTEP group (p=0.545 between the
GMFCS levels II (n=1), III (n=9), and IV (n=4). In the exercise 12- and 16-week periods). Cadence improved for children in
group, the seven males and five females had a mean age of the exercise group after 12 weeks (p<0.001), with no signifi-
9 years 6 months (SD 2y 4mo), with four with diplegic, two cant decline in values after the washout period (p=0.170
with triplegic, and six with quadriplegic CP. GMFCS levels between the 12- and 16-week periods). Post-hoc testing for
were II (n=1), III (n=6), and IV (n=5). stride length showed a gain for the SSTTEP group ( p=0.001)
Six participants were lost to follow-up at the 12-week point after the intervention was withdrawn. However, there was a
because of personal and family reasons not related to the inter- decline in stride length for both groups after the intervention
vention. Two children (in the SSTTEP group) did not partici- (SSTTEP, p=0.001; exercise, p=0.005).
pate in data collection after the washout period. All children There were no significant changes ( p=0.31) in GMFM
obtained an adherence rate of at least the required 80%. Study scores in either group or differences between groups ( p=0.66;
power was calculated to be 12.7% for the 26 children who Table II). The PODCI global score improved for all partici-
completed the study. pants ( p=0.003) but there was no difference between groups
After the intervention there were no differences within or ( p=0.73). Post-hoc testing showed gains in the PODCI global
between groups for spasticity, strength, or motor control score in the SSTTEP group ( p=0.001) after the intervention,
(Table II). The variability, as indicated by the SD, was large which was maintained after the intervention was withdrawn
for all measures. Changes were found in all spatiotemporal ( p=0.866 between the 12- and 16-week periods). There
parameters (p<0.001), with no difference between groups were no significant changes ( p=0.31) in the PODCI
(Table II). Gait speed and cadence improved, with inconsis- transfers and mobility scores across all children, and no
tent changes seen with stride length. Post-hoc testing showed difference between groups over time ( p=0.47). To compare
an increase in gait speed for both groups (SSTTEP, p=0.008; the effects of the interventions measured on different scales

746 Developmental Medicine & Child Neurology 2011, 53: 742–750

 14698749, 2011, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2011.03990.x by Cochrane Chile, Wiley Online Library on [01/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table II: Means (SD) for outcome measures for the 26 children (14 in supported speed treadmill training exercise program [SSTTEP], 12 in exercise)

Baseline 12wks Washout

SSTTEP Exercise SSTTEP Exercise SSTTEP Exercise

Plantarflexor spasticity (J ⁄  ⁄ s) 0.0013 (0.0012) 0.0030 (0.0024) 0.0016 (0.0024) 0.0030 (0.0021) 0.0012 (0.0018) 0.0026 (0.0013)
Knee flexor spasticity (J ⁄  ⁄ s) 0.0088 (0.0114) 0.0032 (0.0044) 0.0074 (0.0133) 0.0072 (0.0137) 0.0083 (0.0139) 0.0053 (0.0044)
Knee extension strength (N ⁄ kg) 3.90 (3.09) 3.09 (3.15) 3.58 (2.82) 3.80 (4.22) 3.66 (3.25) 3.69 (3.66)
Knee flexion strength (N ⁄ kg) 2.47 (1.45) 2.35 (2.04) 2.43 (1.54) 2.98 (3.26) 2.57 (1.65) 2.54 (2.09)
Dorsiflexion strength (N ⁄ kg) 0.86 (1.21) 0.62 (0.75) 0.69 (0.78) 0.77 (0.66) 1.02 (1.54) 0.62 (0.53)
Plantarflexion strength (N ⁄ kg) 3.44 (1.91) 3.06 (3.62) 3.23 (1.45) 3.14 (3.32) 3.65 (2.13) 3.35 (3.17)
Motor control SD (N) 28.3 (14.9) 27.5 (16.4) 22.1 (9.0) 27.8 (5.6) 26.8 (11.6) 24.4 (8.7)
GMFM 62.7 (17.5) 58.4 (26.9) 63.3 (16.2) 60.1 (25.1) 65.3 (16.5) 60.6 (26.7)
PODCI global 50.4 (11.2) 50.9 (14.9) 59.1 (11.4) 52.0 (22.6) 60.0 (10.0) 55.4 (21.7)
PODCI transfers and mobility 46.4 (23.0) 60.6 (26.7) 55.0 (22.9) 55.4 (21.7) 56.9 (20.7) 49.9 (36.2)
Gait speed (m ⁄ s) 0.50 (0.26) 0.44 (0.35) 0.62 (0.31) 0.50 (0.39) 0.63 (0.28) 0.44 (0.34)
Cadence (steps ⁄ min) 76.9 (33.9) 53.3 (24.0) 82.2 (38.2) 60.7 (26.8) 81.2 (38.6) 55.5 (24.8)
Stride length (m) 0.71 (0.27) 0.61 (0.29) 0.68 (0.29) 0.55 (0.29) 0.78 (0.28) 0.64 (0.31)

The sample size for the spasticity assessment was decreased compared with other measures to 18 (10 in SSTTEP, eight in exercise) as one site’s
data could not be used. GMFM, Gross Motor Function Measure; PODCI, Pediatric Outcomes Data Collection Instrument.

over time, the within-group standardized effects over time Minor anticipated adverse events were noted for three chil-
were calculated for each parameter with the 95% confidence dren: two complained of leg ⁄ knee discomfort off the treadmill,
intervals (Fig. 3). which resolved without intervention; and one child developed
As shown in Table II, there was large variability in the a blister beneath his ankle–foot orthosis during the induction
data, which contributed to the lack of statistical significance period.
between and ⁄ or within groups for the tested measures.
Therefore, MCID findings are important for understanding DISCUSSION
the potential clinical significance of the results. Table III and This study found gains in gait spatiotemporal parameters and
Table SII (supporting material online) show the MCID functional activity (PODCI), regardless of intervention group.
results for each child for the GMFM,25 the PODCI, and gait Therefore, both groups benefited from participation in the
spatiotemporal parameters.24 These results show mixed study. However, there were no significant changes in the
responses, with some children experiencing positive changes impairment level measures of spasticity, strength, or motor
and others negative changes across groups. Overall, it control for either group. Our hypothesis that the SSTTEP
appeared that the SSTTEP group had greater changes in the group would show greater gains overall was not supported.
PODCI and gait speed, and the exercise group had greater Despite this finding, the results of this study are encouraging.
change in the GMFM. Children were able to participate successfully in an intensive

Plantarflexor spasticity
Knee flexor spasticity
Knee extension strength
Knee flexion strength
Plantarflexion strength
Dorsiflexion strength
Quadriceps motor control
Gait speed
Cadence
Stride length
GMFM
PODCI global
PODCI transfers and mobility

1.8 –1.6 –1.4 –1.2 –1 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Standardized effect size

Figure 3: Within-group standardized effects over time and their 95% confidence intervals for outcome measures. The solid line represents the supported
speed treadmill training exercise program (SSTTEP) group, the dashed line represents the exercise group. Results are considered significant if the 95%
confidence interval line does not include zero. GMFM, Gross Motor Function Measure; PODCI, Pediatric Outcomes Data Collection Instrument.

Function and Gait Following Treadmill Training Therese E Johnston et al. 747
 14698749, 2011, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2011.03990.x by Cochrane Chile, Wiley Online Library on [01/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table III: Children in each group who showed minimal clinically important differences (MCIDs) after intervention

Group GMFM PODCI global PODCI transfers Gait speed Cadence Stride length

SSTTEP (n) )1 (3) )2 (0) )2 (2) )2 (0) )2 (0) )2 (4)

0 (7) )1 (2) )1 (0) )1 (1) )1 (1) )1 (1)
1 (4) 0 (3) 0 (3) 0 (8) 0 (5) 0 (4)
1 (0) 1 (1) 1 (1) 1 (4) 1 (0)
2 (6) 2 (6) 2 (4) 2 (3) 2 (4)
ND (3) ND (2) ND (1) ND (1)
Exercise (n) )1 (2) )2 (2) )2 (1) )2 (1) )2 (0) )2 (3)
0 (2) )1 (0) )1 (0) )1 (0) )1 (1) )1 (0)
1 (8) 0 (3) 0 (4) 0 (6) 0 (5) 0 (4)
1 (2) 1 (2) 1 (0) 1 (0) 1 (1)
2 (1) 2 (1) 2 (3) 2 (4) 2 (2)
ND (4) ND (4) ND (2) ND (2) ND (2)

Means (SDs) are calculated from the MCID codes to indicate overall positive or negative change. MCID codes: )2, large negative change (effect
size 0.8); )1, medium negative change (effect size 0.5 for Pediatric Outcomes Data Collection Instrument [PODCI] and gait. For the Gross Motor
Function Measure [GMFM] )1 indicates a negative change as no distinction was made between large and medium changes). 0, no significant
change; 1, medium positive change (effect size 0.5 for PODCI and gait. For the GMFM, 1 indicates a negative change as no distinction was made
between large and medium changes); 2, large positive change (effect size 0.8). SSTTEP, supported speed treadmill training exercise program;
ND, no data.

12-week, primarily home-based, intervention program with enough to find a statistical difference in the outcomes
positive outcomes. measured, which may also be the case in our study.
Dodd and Foley10 examined gait speed after a 6-week Although there was no difference between groups, the post-
treadmill training intervention for children aged 5 to 14 years hoc findings for gait speed and the PODCI are important.
with CP classified in GMFCS levels III and IV. They found Children in both groups increased gait speed after the inter-
no difference in gait speed between a non-intervention com- vention. However, only the SSTTEP group maintained this
parison group and those using a treadmill. However, gains improvement after the intervention was withdrawn. These
were reported within the treadmill group with an effect size findings suggest that motor learning may have been a factor
of 1.02 and mean increase in gait speed of 0.7m ⁄ s, with no based on task specificity and repetition with walking on the
change within the comparison group. Willoughby et al.26 treadmill. In addition, only children in the SSTTEP group
found no changes in gait speed for children classified in showed gains in PODCI global scores, which were maintained
GMFCS levels III and IV, who were randomly assigned to a without continued intervention, suggesting carryover effects.
9-week treadmill or overground walking program conducted Further research is needed to determine how long carryover
twice a week. In our study, both groups received an interven- effects might last or if additional periodic training might be
tion and improved in gait speed. However, children in our needed to maintain gains.
study participated in a more intensive program than that of The results for spasticity and motor control support previ-
Dodd and Foley10 or Willoughby et al.26 Finally, in a small ous work indicating no significant change as a consequence of
pilot study by Begnoche and Pitetti,27 a younger sample (ages PBWSTT.11 As with the present investigation, Cherng et al.11
2–9y) of children with CP significantly improved step length reported no changes in spasticity or motor control after non-
after a 4-week treadmill training intervention, although gait randomized intervention (n=8). Their measures for spasticity
speed and cadence did not change. These findings are oppo- (modified Ashworth scale) and motor control (graded scale)
site to ours. differed from those used in our study, yet the results were the
The focus of the exercise program was on muscles impor- same. Cherng et al.11 suggested that as assessments of spastic-
tant for walking and on standing-weight-bearing activities, ity and motor control were performed under a ‘static’ condi-
which were challenging for many children, especially those tion, they might not be related to the ‘dynamic’ condition
classified in GMFCS level IV. The inclusion of weight- associated with the PBWSTT intervention, and muscle tone
bearing activities might help to explain the lack of significant and motor control might not be associated with walking
differences between groups for gait parameters, the PODCI, performance or gross motion function. The unlinking of
and the GMFM in this study. Dobkin et al.28 found compara- spasticity and motor control to gait and gross motion function
ble gains in walking speed and distance between a locomotor has some merit, and little relationship has been reported
program using body-weight-supported treadmill training and between spasticity, gait speed, and GMFM scores.29 In an
an overground ambulation training program for people with unpublished investigation, no correlation was found between a
acute incomplete spinal cord injury. In their study, the over- measure of ankle motor control30 and gait speed and gross
ground ambulation program involved more weight-bearing motor function.
activities than are typically performed in an acute rehabilita- No difference was found in strength as a result of the
tion setting. As indicated by Dobkin et al.,28 the differences PBWSTT. The result is an interesting contrast to what is
between the two interventions may not have been significant characteristic of many previous studies in which a strength

748 Developmental Medicine & Child Neurology 2011, 53: 742–750

 14698749, 2011, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2011.03990.x by Cochrane Chile, Wiley Online Library on [01/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
training regime was used, and strength and function changes research with this population. The initial goal of the study was
were assessed.31,32 Scholtes et al.32 used a progressive resis- to have 54 children complete it, to give adequate statistical
tance strength training program for children with CP (three power; however, the final number of children was much smal-
times a week for 12wks). They found that although strength ler (50% of this goal) within the funding period for the grant.
increased, there was no change in mobility. However, another The biggest challenge to recruitment was the 2-week clinic-
study31 had children with CP perform an ankle-strengthening based induction period, which was sometimes difficult for par-
program (three times a week for 12wks). It found that strength ents to accommodate around their home and work demands.
and gross motor function improved, and that gait speed was The large variability due to a small sample size possibly led to
improved for participants strengthening the plantarflexors. In a type II error. Based on the results of this study, a sample size
both of these studies, the strategy was to improve the impair- of 58 children per group would be needed to detect a differ-
ment and determine if function improved. In the current ence of 0.1m ⁄ s in gait speed between groups with 80% power
study, the strategy was to improve the function directly (i.e. (p=0.05 two-tailed test).
gait for SSTTEP group) and to determine if function and The inability to blind the evaluators at one site was another
impairments improved. In terms of the International Classifi- potential limitation whose effects are unknown; however, the
cation of Functioning, Disability and Health, these measures data from that site showed increases and decreases in different
reflect Body Function (strength) and Activity (gait and gross variables for children in each group, similar to what occurred
motor function) and their interactions.33 Understanding the at the other two sites. The lack of data on spasticity at one site
mechanisms leading to improvement in function would be also was a limitation, thus decreasing the sample size for this
helpful in developing training protocols for children with CP. measure. Another potential limitation was that most of the
A recent report34 has suggested that activity-based interven- intervention was conducted at home. Although this method
tions are more important for improving activity than impair- may have increased adherence, some control over routine
ment-based interventions. decision-making was potentially lost despite the regular
The use of MCID is becoming important in rehabilitation, telephone calls to parents. Another limitation was the use of
as a statistical change may not be clinically meaningful, and a MCID as data were unavailable for the PODCI and gait
meaningful clinical change may not achieve statistical signifi- spatiotemporal parameters for children classified in GMFCS
cance. Therefore we decided to examine additionally the out- level IV. Therefore data from level III were used; however, the
comes of this study in relation to MCID values available in the accuracy of their use for level IV is unknown. Finally, measure-
literature. As seen in Table III, the MCID results are mixed: ment error is not known for two of the impairment tests per-
some children showed gains in most or all areas, others formed in this study, so it is not known if the error affected the
showed a mix of improvements and declines, and some findings.
showed no change and ⁄ or decline. These data are challenging
as many variables could account for differences such as CONCLUSION
GMFCS level, baseline values for the measures studied, the Although our hypothesis that the SSTTEP group would have
amount of body weight support and speed for the SSTTEP better outcomes was not supported, the results are encourag-
group, and the intensity of the exercises completed by the ing as children in both groups showed important changes in
exercise group. A larger study with sufficient power is needed the functional measures. However, only children in the
to examine these relationships. SSTTEP group were able to maintain gains in some variables
MCID suggested greater clinical change in GMFM for the once the intervention was withdrawn. Future research should
exercise group and in the PODCI and gait speed for the SST- include a randomized controlled trial with a non-intervention
TEP group. This result should be interpreted with caution comparison group and a larger sample size; this would provide
owing to our small sample size. MCID data for spatiotemporal greater protection against a type II error and determine
gait parameters showed an interesting trend in that a greater the factors that contribute to which children make the most
number of children aged 10 years and younger improved than gains.
did those older than 10 years. Of note, our sample size of
younger compared with older children was small; however, ACKNOWLEDGEMENTS
perhaps an intensive intervention, such as those used in this This study was funded by Shriners Hospitals for Children, grant
study, may have a greater impact on younger than older chil- number 9147. We acknowledge Rosemary Norris and Debra Banks
dren. Future studies should examine differences in outcomes for their efforts with training and monitoring progress of the children,
across age groups to guide the appropriate timing of the inter- and Anna Means and Trent Wierick for assistance with data
vention. collection. The funding agency was not involved in study design, data
There are several limitations of this study. First, there was collection, data analysis, manuscript preparation, or publication
considerable variability in the data, thus weakening the overall decisions.
power of the study. We chose to work with marginally ambu-
latory children, which is a large undertaking and has an impact ONLINE MATERIAL ⁄ SUPPORTING INFORMATION
on potential issues with variability. The issue of variability was Additional material and supporting information for this paper may be
also reported by Dodd and Foley10 and is problematical for found online.

Function and Gait Following Treadmill Training Therese E Johnston et al. 749
 14698749, 2011, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2011.03990.x by Cochrane Chile, Wiley Online Library on [01/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
REFERENCES
1. Lepage C, Noreau L, Bernard PM. Association between patients with cerebral palsy. Arch Phys Med Rehabil 2000; 81: 24. Oeffinger D, Bagley A, Rogers S, et al. Outcome tools used
characteristics of locomotion and accomplishment of life 301–6. for ambulatory children with cerebral palsy: responsiveness
habits in children with cerebral palsy. Phys Ther 1998; 78: 13. Pierce SR, Lauer RT, Shewokis PA, Rubertone JA, Orlin and minimum clinically important differences. Dev Med
458–69. MN. Test-retest reliability of isokinetic dynamometry for the Child Neurol 2008; 50: 918–25.
2. Bjornson KF, Belza B, Kartin D, Logsdon R, McLaughlin assessment of spasticity of the knee flexors and knee extensors 25. Wang HY, Yang YH. Evaluating the responsiveness of 2 ver-
JF. Ambulatory physical activity performance in youth with in children with cerebral palsy. Arch Phys Med Rehabil 2006; sions of the gross motor function measure for children with
cerebral palsy and youth who are developing typically. Phys 87: 697–702. cerebral palsy. Arch Phys Med Rehabil 2006; 87: 51–6.
Ther 2007; 87: 248–57. 14. Engsberg JR, Olree KS, Ross SA, Park TS. Quantitative clin- 26. Willoughby KL, Dodd KJ, Shields N, Foley S. Efficacy of
3. Johnston TE, Moore SE, Quinn LT, Smith BT. Energy cost ical measure of spasticity in children with cerebral palsy. Arch partial body weight-supported treadmill training compared
of walking in children with cerebral palsy: relation to the Phys Med Rehabil 1996; 77: 594–9. with overground walking practice for children with cerebral
Gross Motor Function Classification System. Dev Med Child 15. Engsberg JR, Ross SA, Olree KS, Park TS. Ankle spasticity palsy: a randomized controlled trial. Arch Phys Med Rehabil
Neurol 2004; 46: 34–8. and strength in children with spastic diplegic cerebral palsy. 2010; 91: 333–9.
4. Willoughby KL, Dodd KJ, Shields N. A systematic review of Dev Med Child Neurol 2000; 42: 42–7. 27. Begnoche DM, Pitetti KH. Effects of traditional treatment
the effectiveness of treadmill training for children with cere- 16. Damiano DL, Quinlivan JM, Owen BF, Payne P, Nelson and partial body weight treadmill training on the motor skills
bral palsy. Disabil Rehabil 2009; 31: 1971–9. KC, Abel MF. What does the Ashworth scale really measure of children with spastic cerebral palsy. A pilot study. Pediatr
5. Mutlu A, Krosschell K, Spira DG. Treadmill training with and are instrumented measures more valid and precise? Dev Phys Ther 2007; 19: 11–9.
partial body-weight support in children with cerebral palsy: a Med Child Neurol 2002; 44: 112–8. 28. Dobkin B, Apple D, Barbeau H, et al. Weight-supported
systematic review. Dev Med Child Neurol 2009; 51: 268–75. 17. Scholtes VA, Becher JG, Beelen A, Lankhorst GJ. Clinical treadmill vs over-ground training for walking after acute
6. Mattern-Baxter K. Effects of partial body weight supported assessment of spasticity in children with cerebral palsy: a crit- incomplete SCI. Neurology 2006; 66: 484–93.
treadmill training on children with cerebral palsy. Pediatr ical review of available instruments. Dev Med Child Neurol 29. Ross SA, Engsberg JR. Relationships between spasticity,
Phys Ther 2009; 21: 12–22. 2006; 48: 64–73. strength, gait, and the GMFM-66 in persons with spastic
7. Damiano DL, DeJong SL. A systematic review of the 18. Engsberg JR, Olree KS, Ross SA, Park TS. Maximum effort diplegia cerebral palsy. Arch Phys Med Rehabil 2007; 88:
effectiveness of treadmill training and body weight support resultant knee joint torques in children with spastic diplegia. 1114–20.
in pediatric rehabilitation. J Neurol Phys Ther 2009; 33: J Appl Biomech 1998; 14: 52–61. 30. Engsberg JR, Van Emmerik RE, Ross SA, Collins DR. Use
27–44. 19. Engsberg JR, Wagner JM, Reitenbach AK, Hollander KW. of relative phase as a measure of motor control at the ankle
8. Provost B, Dieruf K, Burtner PA, et al. Endurance and gait A measure of motor control at the knee in cerebral palsy. in persons with cerebral palsy: a preliminary study. J Appl
in children with cerebral palsy after intensive body weight- J Appl Biomech 2001; 17: 335–43. Biomech 2008; 24: 382–90.
supported treadmill training. Pediatr Phys Ther 2007; 19: 20. Russell DJ, Rosenbaum PL, Avery LM, Lane M. Gross 31. Engsberg JR, Ross SA, Collins DR. Increasing ankle strength
2–10. Motor Function Measure (GMFM-66 & GMFM-88) User’s to improve gait and function in children with cerebral palsy:
9. Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Manual. Clinics in Developmental Medicine No. 159. Lon- a pilot study. Pediatr Phys Ther 2006; 18: 266–75.
Galuppi B. Development and reliability of a system to clas- don: Mac Keith Press, 2002. 32. Scholtes VA, Becher JG, Comuth A, Dekkers H, Van DijkL,
sify gross motor function in children with cerebral palsy. Dev 21. Daltroy LH, Liang MH, Fossel AH, Goldberg MJ. The PO- Dallmeijer AJ. Effectiveness of functional progressive resis-
Med Child Neurol 1997; 39: 214–23. SNA pediatric musculoskeletal functional health question- tance exercise strength training on muscle strength and
10. Dodd KJ, Foley S. Partial body-weight-supported treadmill naire: report on reliability, validity, and sensitivity to change. mobility in children with cerebral palsy: a randomized con-
training can improve walking in children with cerebral palsy: Pediatric Outcomes Instrument Development Group. Pedi- trolled trial. Dev Med Child Neurol 2010; 52: e107–13.
a clinical controlled trial. Dev Med Child Neurol 2007; 49: atric Orthopaedic Society of North America. J Pediatr Orthop 33. World Health Organization. The International Classification
101–5. 1998; 18: 561–71. of Functioning, Disability and Health – Children and Youth
11. Cherng RJ, Liu CF, Lau TW, Hong RB. Effect of treadmill 22. Engsberg JR, Ross SA, Collins DR, Park TS. Effect of selec- Version ICF-CY. Geneva: WHO Press, 2007.
training with body weight support on gait and gross motor tive dorsal rhizotomy in the treatment of children with cere- 34. Kim WH, Park EY. Causal relation between spasticity,
function in children with spastic cerebral palsy. Am J Phys bral palsy. J Neurosurg 2006; 105: 8–15. strength, gross motor function, and functional outcome in
Med Rehabil 2007; 86: 548–55. 23. Conover WJ, Iman RL. Rank transformations as a bridge children with cerebral palsy: a path analysis. Dev Med Child
12. Schindl MR, Forstner C, Kern H, Hesse S. Treadmill train- between parametric and nonparametric statistics. Am Stat Neurol 2011; 53: 68–73.
ing with partial body weight support in nonambulatory 1981; 35: 124–9.

750 Developmental Medicine & Child Neurology 2011, 53: 742–750