DEVELOPMENTAL MEDICINE & CHILD NEUROLOGY SYSTEMATIC REVIEW

Exercise interventions improve postural control in children with

cerebral palsy: a systematic review
ROSALEE DEWAR 1 | SARAH LOVE 2 | LEANNE MARIE JOHNSTON 1
1 The University of Queensland, Brisbane, Qld; 2 Princess Margaret Hospital, Perth, WA, Australia.
Correspondence to Rosalee Dewar at Division of Physiotherapy, School of Health and Rehabilitation Sciences, The University of Queensland, Brisbane, Qld 4072, Australia.
E-mail: [email protected]

PUBLICATION DATA AIM The aim of this study was to evaluate the efficacy and effectiveness of exercise
Accepted for publication 28th October interventions that may improve postural control in children with cerebral palsy (CP).
2014. METHOD A systematic review was performed using American Academy of Cerebral Palsy and
Published online 18th December 2014. Developmental Medicine (AACPDM) and Preferred Reporting Items for Systematic Reviews
and Meta-Analyses (PRISMA) methodology. Six databases were searched using the following
ABBREVIATIONS keywords: (‘cerebral palsy’ OR ‘brain injury’); AND (‘postur*’ OR ‘balance’ OR ‘postural
AACPDM American Academy of Cerebral balance’ [MeSH]); AND (‘intervention’ OR ‘therapy’ OR ‘exercise’ OR ‘treatment’). Articles
Palsy and Developmental Medi- were evaluated based on their level of evidence and conduct.
cine RESULTS Searches yielded 45 studies reporting 13 exercise interventions with postural
BF&S Body functions and structures control outcomes for children with CP. Five interventions were supported by a moderate
F-BWS Full body weight support level of evidence: gross motor task training, hippotherapy, treadmill training with no body
FES Functional electrical stimulation weight support (no-BWS), trunk-targeted training, and reactive balance training. Six of the
ICF International Classification of interventions had weak or conflicting evidence: functional electrical stimulation (FES),
Functioning, Disability and hippotherapy simulators, neurodevelopmental therapy (NDT), treadmill training with body
Health weight support, virtual reality, and visual biofeedback. Progressive resistance exercise was an
NDT Neurodevelopmental therapy ineffective intervention, and upper limb interventions lacked high-level evidence.
No-BWS No body weight support INTERPRETATION The use of exercise-based treatments to improve postural control in
P-BWS Partial body weight support children with CP has increased significantly in the last decade. Improved study design
PRISMA Preferred Reporting Items for provides more clarity regarding broad treatment efficacy. Research is required to establish
Systematic Reviews and Meta- links between postural control impairments, treatment options, and outcome measures. Low-
Analyses burden, low-cost, child-engaging, and mainstream interventions also need to be explored.

Cerebral palsy (CP) is the most common cause of physical upper limb activities such as reaching,19 and during oral
disability in childhood, with an estimated incidence of 2.11 motor activities such as eating, swallowing, and speaking.20
per 1000 live births.1 Central to the definition and diagno- These limitations restrict participation across a broad range
sis of CP is impaired development of movement and pos- of life domains, including self-care, education, and recrea-
ture.2 Postural control dysfunction derives from primary tion.21 Despite the significant impact that postural control
brain injury, which causes deficits in postural networks. dysfunction has on the activity and participation of children
Motor (producing) networks are impacted by deficits such with CP, and indeed their caregivers, optimal interventions
as muscle spasticity, contracture, decreased isometric force for this core deficit are not well understood. While children
production and abnormal timing, and reduced amplitude with CP receive or participate in a wide range of passive or
of muscle recruitment. Perceptual (orienting) networks are active interventions aimed to improve movement and pos-
impacted by deficits including poor registration and/or ture, often the specific impact on postural control is not well
perception in visual, tactile, proprioceptive, and vestibular measured or documented. This systematic review seeks to
systems.3 Individually and collectively, these factors can examine reported exercise interventions for children with
result in problems with balance and/or orientation in chil- CP, to critique their efficacy and effectiveness for postural
dren with CP. To date, it is known that children with CP control outcomes, and to make recommendations for
show deficits in anticipatory postural adjustments,4–9 and improved therapeutic management of this fundamental attri-
reactive postural adjustments,10–12 as well as sensory4,13 bute of CP.
and musculoskeletal components14,15 of postural control, Postural control can be defined as the ability to con-
compared with children with typical development. This trol the body’s position in space for the purposes of sta-
dysfunction is known to contribute to limitations in gross bility and orientation.22,23 Postural stability, or balance,
motor skills that require balance,16 especially gait,17,18 during is the ability to maintain and/or regain the centre of

504 DOI: 10.1111/dmcn.12660 © 2014 Mac Keith Press

mass within the base of support where gravity is the key What this paper adds
vector.22,24 Stability tasks can be considered static, when • First systematic review of postural control exercise interventions for children
the body is stationary (e.g. when sitting or standing on a with cerebral palsy.
stable surface), or dynamic, when the body is moving, • Exercise interventions that improve postural control have increased in the
either during self-initiated internal perturbations (e.g. last decade.
walking), or in response to external perturbations initi-
• Improved study design has clarified efficacy of postural control exercise
approaches.
ated by other people or objects (e.g. being pushed, • Five exercise interventions reached moderate evidence level, however, no
or maintaining a stance on a moving bus).25 Postural interventions were rated strong.
orientation is the ability to attain and maintain an opti-
mal functional relationship between body segments, a and methodological rigor is required to provide adequate
task, and the environment (e.g. for writing, reaching, or clinical guidance.
looking).22,24 The outcomes and recommendations of these reviews
The effect and intent of postural control interventions prompted the study of many new clinical and mainstream
need to be evaluated with reference to a framework of core approaches to treating postural control dysfunction in chil-
postural control elements. Although numerous theoretical dren with CP. Key examples include hippotherapy, tread-
frameworks exist, the contemporary Systems Control The- mill training, upper limb therapy, strength training, and
ory is the most comprehensive for this purpose.22,26,27 This virtual reality technologies. However, it has been almost
theory describes postural control as a complex interaction 10 years since the last review of postural control interven-
between seven components: (1) neuromuscular synergies; tions in children with CP; therefore, an updated systematic
(2) internal representations; (3) adaptive mechanisms review, to evaluate the efficacy and effectiveness of tradi-
(including reactive postural adjustments); (4) anticipatory tional and contemporary exercise interventions, is needed.
mechanisms (including anticipatory postural adjustments); In this respect, efficacy is defined as the ability of an inter-
(5) sensory strategies; (6) individual sensory systems; and vention to improve postural control under ideal conditions,
(7) musculoskeletal components.22 Children with motor such as in a laboratory, and effectiveness is defined as the
disorders can show deficits in one or more of these compo- ability of an intervention to provide benefits during usual
nents. Similarly, interventions and outcomes can target one conditions of clinical care.31,32 Therefore, the aim of this
or more components. paper is to present a systematic review of exercise interven-
A burst of postural control intervention research in the tions reported for use in children with CP; to evaluate the
1980s and 1990s, relating to children with CP, prompted efficacy and effectiveness of these interventions for postural
the publication of three review articles.28–30 First of all, control outcomes, according to international standards;
Campbell28 published a non-systematic review of interven- and to recommend appropriate management of postural
tions for children with CP. This review proposed prelimin- control dysfunction in children with CP.
ary support for the following postural control
METHOD
interventions: gait training with real-time auditory biofeed-
This systematic review was conducted according to princi-
back or retrospective verbal feedback; neurodevelopmental
ples of American Academy of Cerebral Palsy and Develop-
therapy (NDT); therapeutic horseback riding; and inhibi-
mental Medicine (AACPDM) methodology for developing
tory casting. A decade later, Westcott and Burtner29 pre-
systematic reviews of treatment interventions,33 and Pre-
sented a second, non-systematic review of children with
ferred Reporting Items for Systematic Reviews and Meta-
motor disabilities (including CP) using the systems control
Analyses (PRISMA) guidelines.34–36 The study did not
approach. This review supported Campbell’s findings, and
require human participation; therefore, ethical approval
also supported some new interventions, including reactive
was not required.
balance training using platform perturbations and anticipa-
tory balance training with computer feedback. There was Search strategy
insufficient or conflicting evidence regarding outcomes of A systematic literature search of articles published between
interventions targeting musculoskeletal (strengthening), January 1980 and December 2013 was performed using the
sensory (vestibular stimulation), and motor (electrical stim- following electronic databases: PubMed, EMBASE, EB-
ulation) processes.29 To improve the existing body of SCOhost (MEDLINE and CINAHL), the Cochrane
research, the authors recommended that further studies Library, and PEDro. Search terms were designed to
include (1) outcome measures for both postural control include the population of interest (‘cerebral palsy’ OR
and motor function; (2) more task-specific training to ‘brain injury’), and intervention type (‘postur*’ OR ‘bal-
improve functional outcomes; and (3) evaluation of main- ance’ OR ‘postural balance’ [MeSH]), AND (‘intervention’
stream recreational activities for their potential impact on OR ‘therapy’ OR ‘exercise’ OR ‘treatment’). Secondary
postural control. In 2005, the first systematic review of searches included reference list checking of the included
postural control interventions for children with CP was articles, electronic searches for included interventions by
published by Harris and Roxborough.30 The authors sup- name and author, and citation tracking of all included arti-
ported Campbell’s28 view that progress in study quality cles. Two authors (RD and LJ or SL) examined the titles

Review 505
and abstracts of the articles identified by these searches. RESULTS
Full-text articles were retrieved if they fulfilled inclusion A total of 911 articles were identified, 890 from initial
criteria, or if further clarification regarding the fulfilment searches, and 21 from secondary searches. After duplicates
of inclusion criteria was required. If agreement on inclu- were removed, the titles and abstracts of 558 articles were
sion could not be reached following review by two of the screened. Of these, 154 full-text articles met initial criteria and
authors, the third author (LJ or SL) was consulted. were retrieved for review, with 45 studies meeting final inclu-
sion criteria. The flow of studies, and reasons for exclusion at
Inclusion and exclusion criteria each stage, is summarized in a PRISMA diagram (Fig. 1).
Articles were included if (1) they were full articles, pub- From the 45 included articles (evidence levels I–V), 13
lished in English, in peer-reviewed journals, after 1980; (2) intervention types were identified that purported to impact
study participants were children diagnosed with CP, and postural stability or postural orientation in children with
aged between 0 and 18 years; (3) they performed a land- CP: functional electrical stimulation (FES) (n=2); gross
based exercise intervention that required active participa- motor task training (n=4); hippotherapy (n=9); hippothera-
tion by the child; and (4) they reported the efficacy or py simulators (n=3); progressive resistance exercise (n=1);
effectiveness of the intervention, for improving postural reactive balance training (n=3); treadmill training with no
control, using at least one outcome measure of either pos- body weight support (no-BWS) (n=1); treadmill training
tural stability (static or dynamic balance), or postural ori- with partial or full body weight support (P-BWS or
entation (e.g. postural alignment). Articles were excluded if F-BWS respectively; n=5); trunk-targeted training (n=2);
they were non-systematic reviews or opinion articles, or if upper limb interventions (n=2); visual biofeedback (n=1);
they reported (1) passive interventions (e.g. orthotics, and virtual reality (n=8). Within the included articles,
equipment such as seating, or support garments); (2) NDT was used as a comparison treatment in seven (n=7,
water-based interventions; (3) medical or surgical interven- not adding to the study tally), and appeared in one other
tions; or (4) active exercise interventions without any where it was used as the sole intervention (n=1). Finally,
reported outcome measures for postural control. three systematic reviews were identified, including one dis-
cussing postural control interventions in general, and two
Data extraction and quality appraisal specifically discussing hippotherapy.
Two authors gathered data from each article using the Results were tabulated based on AACPDM guidelines. Arti-
appropriate AACPDM ‘study data extraction summary cles that were rated as level I to III (n=22) met criteria for full
form’; the forms used were relevant for either group or evaluation, which is provided in Table I, including citation,
single-subject research study designs. The forms recorded design, evidence level and conduct rating, intervention type,
information regarding participants, intervention(s), out- participants, results, outcome measures, and coding (according
come measure(s) (for postural control or other motor out- to the International Classification of Functioning, Disability
comes), results, and potential adverse effects. The quality and Health [ICF]: Children & Youth Version,40 and postural
of each included article was assessed in two steps: (1) by control component according to Systems Control Theory).
assignment of the level of evidence (for all studies); and Tables SIIa,b (online supporting information) provide addi-
then (2) by evaluation of conduct (for studies with level tional information regarding participant characteristics and
I–III evidence only [as determined using guidelines for intervention methodology for each study (level IV–V studies
each study type recommended by the AACPDM]). Group were also included in this table to comprehensively describe
research designs were assigned levels of evidence using the the scope of research available for each intervention type).
classification described by Sackett et al.37 (see Table SI, Tables SIIIa,b,c (online supporting information) report the
online supporting information), where level I studies are objective conduct item scores for level I to III studies with
most able to demonstrate that the intervention was respon- group, single-subject, and systematic review designs. Table
sible for the reported outcome. Conduct of level I to III SIV (online supporting information) documents reported
group studies was rated using a seven-item questionnaire, adverse events.
with studies scoring ‘yes’ on six or seven items rated as In the following sections, each intervention type with
strong, on four or five items rated as moderate, and on available level I to III evidence is critiqued for (1) the over-
three or less items rated as weak. Single-subject research all strength of the evidence presented; (2) the efficacy and/
design studies were assigned levels of evidence using the or effectiveness for improving postural control when con-
classification described by Logan et al.38 (see Table SI). sidering each component (‘body functions and structures’
Conduct of level I to III single-subject research design (BF&S), ‘activity’, and ‘participation’) of the ICF; (3) links
studies was rated using a 14-item questionnaire, with stud- between outcomes across the ICF; and (4) adverse events,
ies scoring ‘yes’ on between 11 and 14 items rated as if reported. Intervention types are discussed in alphabetical
strong, on 7 to 10 items rated as moderate, and on seven order.
or less items rated as weak. Systematic review studies were
rated using the classification of Sackett et al.37 Conduct of Functional electrical stimulation
systematic reviews was evaluated using Oxam and Two studies,41,42 both evidence level II (Table I), applied
Guyatt’s39 classification, which yields a score out of 10. FES to abdominal and lumbar muscles simultaneously, with

506 Developmental Medicine & Child Neurology 2015, 57: 504–520

 Records identified through Additional records identified
Identification database searching through other sources
(n=890) (n= 21)

Records after duplicates removed

(n=558)

Records excluded
Screening

(n=404)
Records screened for title Too old (n=43)
and abstract Not cerebral palsy (n=157)
(n=558) Not postural control assessment or treat (n=177)
Non-systematic review or opinion paper (n=26)
Withdrawn from publication (n=1)

Full-text articles excluded,

with reasons
(n=109)
Not English text (n=6)
Eligibility

Too old (n=17)

Full-text articles assessed Not CP/BI (n=4)
for eligibility Not postural control assessment or treat (n=22)
Equipment (seating, orthotics, support garments)
(n=154)
(n=19)
Water-based exercise (n=1)
Medical or surgical intervention (n=9)
Non-systematic review or opinion paper (n=21)
Abstract or conference proceedings only (n=10)
Included

Studies included in
qualitative synthesis
(n=45)

Figure 1: PRISMA flow diagram.

the aim of improving muscle strength and function. Both step-ups, walking and standing activities, and reaching to
studies used the following FES parameters with a sequence limits of stability). Of four studies, two were level II
of 10 seconds ‘on’ followed by 12 seconds ‘off’: intensity of (Table I). Improvement was reported for the ‘activity’, but
20 to 30mA; pulse width of 250ls; and frequency of 25 to not the BF&S, component. Thirty hours (five 1h sessions/
35Hz. Both studies used the same dosage of 10 to 18 hours wk for 6wks) of sit-to-stand and step-up exercises improved
(five or six 30min sessions/wk for 4–6wks), together with standing balance (‘activity’) and dynamic postural stability
rehabilitation (stretching, strengthening, and mobility activi- during gait (‘activity’) in children with CP aged between 5
ties, and Bobath treatments of inpatients in rehabilitation and 12 years (evidence level II, conduct moderate; see
hospitals) for children aged between 1 and 10 years with Table SIIIa).43 A lower dose of 10 hours (two 1h sessions/
spastic diplegia (Gross Motor Function Classification Sys- wk for 5wks) of walking, standing, sit-to-stand, and object
tem [GMFCS] Level not reported). FES, along with rehabil- pick-up activities improved dynamic balance during gait
itation, improved postural alignment (BF&S) to a greater (‘activity’) in 4- to 11-year-old children with CP (evidence
extent than rehabilitation alone (evidence level II, conduct level II, conduct moderate; see online Table SIIIa).44
weak; see Table SIIIa).41,42 No discomfort was reported by
children receiving FES (see Table SIV).42 Hippotherapy
Hippotherapy is the provision of sensory and motor input
Gross motor task training via the movements of a horse, with programmes designed
Gross motor task training involves repetition of simple by professionals with hippotherapy qualifications.53 From
functional gross motor exercises (e.g. sit-to-stand exercises, 11 studies, three level II or III studies, and two systematic

Review 507
 Table I: Summary of level I to III included studies

Design,
evidence level, PC
and conduct ICF (proposed component
References rating Intervention Participants Results Outcome measure code) measured

Functional electrical stimulation (FES)

Park et al.41 Group, II, Weak FES CP (spastic Improved postural symmetry in sitting Radiographic B (b7101) MSk
diplegic); (Cobb and kyphotic angle, p<0.05); measures: Cobb,
mean age=13y no change in lumbosacral angle kyphotic, and
6mo; n=26 lumbosacral angle
Improved sitting function (p<0.05) GMFM: B (sitting) A (d410–d429, APA, IR, MSk
d455)
Karabay et al.42 Group, II, Weak FES CP (spastic Improved postural symmetry in sitting Radiographic B (b7101) MSk
diplegic); (Cobb, kyphotic, and sacral angle, measures: Cobb,
2–10y; n=33 p<0.001) kyphotic, and sacral
angle
Improved sitting (p<0.05) GMFM: B (sitting) A (d410–d429, APA, IR, MSk
d455)

508 Developmental Medicine & Child Neurology 2015, 57: 504–520

Gross motor
task training
Katz-Leurer Group, II, Moderate Gross motor TBI and CP; No change in strength Muscle strength of B (b7300) MSk
et al.43 task training GMFCS I–II; lower limb
5–13y; n=20 (dynamometry)
No change in walking performance Walking speed over B (b770) APA, IR, MSk
10m
No change in walking efficiency Walking energy B (b455) APA, IR, MSk
expenditure index
Improved dynamic balance in Functional reach test A (d4105, d4106, APA, IR
standing (p<0.01) and walking (p<0.01) d4452)
Timed-up and go test A (d4103, d4104, APA, MSk
d4500)
Salem and Group, II, Moderate Gross motor CP; GMFCS I–III; Improved gross motor function: GMFM-88: D (standing) A (d410–d429, APA, IR, MSk
Godwin.44 task training 4–11y; n=10 standing (p=0.009) and walking, and E (walking, d455)
running, and jumping (p=0.017) running, and
jumping)
Improved dynamic balance: Timed-up and go test A (d4103, d4104, APA, MSk
walking (p=0.017) d4500)
Hippotherapy
Kang et al.45 Group, II, Weak Hippotherapy CP (hemi- and Improved sitting balance (p<0.05) Stabilometry (sitting): B (b755) IR, sensory,
diplegic pathway and velocity MSk
ambulatory); of COP while sitting
6–10y; n=45 still for 30s with
visual fixation
Hamill et al.46 SSRD, III, Moderate Hippotherapy CP; GMFCS V; No change in sitting postural control Sitting Assessment A (d4153, d4300, IR, APA, MSk
27–54mo; Scale d440) and B (b760)
sitting; n=3 No change in gross motor function GMFM-88: B (sitting) A (d410–d429, APA, IR, MSk
overall, or in sitting and total score d455)
 Table I: Continued

Design,
evidence level, PC
and conduct ICF (proposed component
References rating Intervention Participants Results Outcome measure code) measured
47
Kwon et al. Group, III, Hippotherapy CP; GMFCS I–II; Improved walking (stride length Temporal–spatial and B (b770, b710) APA, MSk
Moderate 4–10y; n=32 [p<0.001] and walking speed kinematic gait
[p=0.002]); no change parameters
in cadence, single limb support, or
pelvic and hip kinematics
Improved overall gross motor function GMFM-66/88: E and A (d410–d429, APA, IR, MSk
(p=0.003); improved walking, running, total score d455)
and jumping (p=0.042)
Improved balance in standing (p=0.004) Paediatric Balance A (d4103, d4104– APA, IR,
Scale d4106, d4153- sensory,
d4154, d4200, MSk
d4452)
Hippotherapy
simulator
Borges et al.48 Group, II, Weak Hippotherapy CP (spastic Improved postural control in sitting Stabilometry: B (b760) APA
simulator diplegic); (AP, p=0.0001; ML, p=0.0069) voluntary COP
GMFCS II–IV; movement in sitting
3–12y; n=40 No change in motor classification GMFCS A (d450–d469) APA, IR, MSk
Herrero et al.49 Group, II, Strong Hippotherapy CP; GMFCS I–IV; No change in sitting postural control Sitting Assessment A (d4153, d4300, APA, IR, MSk
simulator 4–18y; n=38 Scale d440)and B (b760)
Improved sitting function (odds GMFM-66: B (sitting) A (d410–d429, APA, IR, MSk
ratio=3.9; 95% CI=0.68–22.7); and total score d455)
no change in overall gross motor
function
NDT
Park et al.41 Group, II, Weak NDT CP (spastic Improved postural symmetry in sitting Radiographic B (b7101) MSk
diplegia); (Cobb angle only p<0.05); no change measures: Cobb,
8–16mo; n=26 in kyphotic and lumbosacral angle kyphotic, and
lumbosacral angle
Improved sitting function (p<0.05) GMFM: B (sitting) A (d410–d429, APA, IR, MSk
d455)
Salem and Group, II, Moderate NDT CP (quadriplegic Improved gross motor function: GMFM-88: D (standing) A (d410–d429, APA, IR, MSk
Godwin44 and diplegic); standing and E (walking, d455)
GMFCS I–III; 4– and walking, running, and jumping running, and
11y; n=10 (no p-value reported) jumping)
Improved dynamic balance: walking Timed-up and go test A (d4103, d4104, APA, MSk
(no p-value reported) d4500)
Borges et al.48 Group, II, Weak NDT CP (spastic Improved postural control: sitting Stabilometry: B (b760) APA
diplegia); (AP and ML, no p-value reported) voluntary COP
GMFCS II–V; 3– movement in sitting
10y; n=40 No change in motor classification GMFCS A (d450–d469) APA, IR, MSk
Karabay et al.42 Group, II, Weak NDT + CP (spastic Improved postural symmetry: sitting Radiographic B (b7101) MSk
conventional diplegic); (Cobb, kyphotic angle [p<0.001] and measures: Cobb,
PT 2–10y; n=33 sacral angel [p<0.003]) kyphotic, and sacral

Review
angle
Improved sitting (p<0.001) GMFM: B (sitting) A (d410–d429, APA, IR, MSk
d455)

509
 Table I: Continued

Design,
evidence level, PC
and conduct ICF (proposed component
References rating Intervention Participants Results Outcome measure code) measured

El-Shamy et al.50 Group, II, Moderate NDT CP (spastic Improved limits of stability (p<0.001) Biodex, DSL level 12: B (b760) APA, IR,
diplegic); movement of COP sensory
GMFCS I–II; 10– with visual feedback
12y; n=30 Improved (reduced) falls risk (p<0.05) Biodex: COP pathway A (d450) RPA, NMR
in response to
perturbation
calculation of falls risk
Improved functional standing balance Paediatric Balance A (d4103, d4104– APA, IR
(p<0.05) Scale d4106, d4153- sensory,
d4154, d4200, MSk

510 Developmental Medicine & Child Neurology 2015, 57: 504–520

d4452
Kwon et al.47 Group, III, NDT CP (bilateral Improved walking (cadence p=0.013; Temporal–spatial and B (B770,B710) APA, MSk
Moderate spastic); walking speed p=0.002); no change kinematic gait
GMFCS I–II; 4– in stride length, single limb support, parameters
10y; n=32 or pelvic and hip kinematics
No change in gross motor function GMFM-66/88: E and A (d410–d429, APA, IR, MSk
total score d455)
No change in standing balance Paediatric Balance A (d4103, d4104– APA, IR,
Scale d4106, d4153– sensory,
d4154, d4200, MSk
d4452)
Progressive resistance exercises
Bandholm et al.51 Group, II, Strong Progressive CP; GMFCS I; 5– Improved plantar flexion MVT only Ankle muscle strength: B (b7300) MSk
resistance 14y; n=15 (p<0.035) DF and PF
training dynamometer (MVT),
and EMG (torque
steadiness)
No change in gait pattern 3D gait kinematics and B (b770, b710) APA, IR, MSk
temporal–spatial
parameters
No change in static balance (standing) Stabliometry: standing B (b755) IR, MSk
COP sway
Improved spasticity of ankle PF in both Modified Ashworth B (7350) MSk
groups (p<0.001) Scale
No change in gross motor performance GMFM-66 A (d410–d429, APA, IR, MSk
d455)
Reactive balance
 Table I: Continued

Design,
evidence level, PC
and conduct ICF (proposed component
References rating Intervention Participants Results Outcome measure code) measured

El-Shamy et al.50 Group, II, Moderate Reactive balance CP (spastic Improved limits of stability (p<0.001) Biodex, DSL level 12: B (b760) APA, IR,
training with diplegic); movement sensory
Biodex balance GMFCS I–II; 10– of COP with visual
system 12y; n=30 feedback
Improved (reduced) falls risk (p<0.05) Biodex: COP pathway A (d450) RPA, NMR
in response
to perturbation
calculation of falls risk
Improved functional standing balance Paediatric Balance A (d4103, d4104– APA, IR,
(p<0.05) Scale d4106, d4153– sensory,
d4154, d4200, MSk
d4452)
Shumway- SSRD, II, Moderate Massed practice: CP; GMFCS I–II; Improved reactive balance in standing Stabilometry: COP B (b755) RPA, NMR
Cook et al.52 reactive 7–12y; n=6 (p<0.05) time to stabilization
balance and area in response
training to perturbation
No change in function in standing GMFM: D (standing) A (d410–d429, APA, IR, MSk
d455)
Systematic reviews
Harris and Systematic Review, Seating, ankle 12 studies; CP; 0 Improved postural control (various) Postural control in Various Various
Roxborough30 II, 8/9 foot orthotics, –19y following motor therapy and sitting, upper
lycra garments, balance training; conflicting extremity movement
motor therapy results for NDT; improved reactive control, standing
and balance balance following reactive balance balance, knee
training training, artificial hippotherapy extension,
protocols simulators, and rocker board training pathological
movement, head
control, independent
sitting balance,
segmental level of
control, dynamic
stability in gait,
function, goal
performance, comfort,
upper limb function,
posture alignment,
engagement with
toys, caregiving,
reactive balance, and
motor function

Review
511
 Table I: Continued

Design,
evidence level, PC
and conduct ICF (proposed component
References rating Intervention Participants Results Outcome measure code) measured

Zadnikar and Systematic Review, Hippotherapy 8 studies; CP; Improved postural control (various) Video, dynamic trunk/ Various Various
Kastrin53 II, 9/9 and therapeutic children or with all interventions head stability, barrel
horse riding adults test, functional reach
and test, Posture
hippotherapy Assessment Scale,
simultors Bertoti’s Postural
Assessment Scale,
GMFM, Peabody
Developmental Motor
Scales, Vineland
Adaptive Behavior
Scales, Self-
perception profile for
children, Child
Behaviour Checklist,

512 Developmental Medicine & Child Neurology 2015, 57: 504–520

Bruininks–Oseretsky
Test of Motor
Proficiency, passive
range of anterior–
posterior pelvic tilt,
photography,
stabilography, muscle
symmetry, and EMG
Tseng et al.54 Systematic Review, Equine-assisted 14 studies; CP; 0 Improved postural control (BPAS and Betoti’s Postural Various Various
II, 8/9 activities –18y SAS – see outcome measure Assessment Scale,
and therapies description) Sitting Assessment
(EAAT): following hippotherapy (involving Scale, surface EMG
hippotherapy children (hip adductor
and GMFCS I–IV); no change in all other asymmetry), Modified
therapeutic measures Ashworth Scale,
horse riding stride length, GMFM-
only 66, and GMFM-88
Treadmill training
Druz_ bicki et al.55 Group, II, Weak Treadmill CP; GMFCS II– Improved standing balance, Stabilometry: COP B (b235, b755) APA, IR,
training III; 6–14y; n=18 predominantly deviations and sensory
with Lukomat in eyes closed condition (p<0.05) pathway in standing,
eyes open and closed
Improved weight symmetry in standing Underfoot pressure B (b7603) IR, sensory,
(p value not reported) distribution MSk
Grecco et al.56 Group, II, Moderate Treadmill CP; GMFCS I–II; Improved standing balance in eyes Stabilometry: COP B (b235, b755) IR, sensory,
training, no- 3–11y; n=15 open and closed conditions for AP oscillation in MSk
BWS (p=0.03), and eyes open for ML standing, eyes open
(p=0.04) and closed
Improved functional balance in Berg Balance Scale A (d4103, d4104– APA, IR,
standing (p=0.01) d4106, d4153– sensory
d4154, d4200,
d4452)
 Table I: Continued

Design,
evidence level, PC
and conduct ICF (proposed component
References rating Intervention Participants Results Outcome measure code) measured

Trunk-targeted training
Unger et al.57 Group, II, Moderate Trunk-targeted CP; GMFCS I–III; Improved postural alignment: AP 2D photographic B (b7101) MSk
exercise 6–13y; n=27 angle sitting and standing, and posture analysis in
with vibration shoulder-to-seat height in sitting standing
(p<0.05)
Improved abdominal muscle thickness Ultrasound imaging: B (d7300) MSk
(RA, OE, OI, and TrA, p<0.05) resting abdominal
muscle thickness
Improved abdominal muscle strength Total sit-ups in one B (d7300) MSk
(p<0.001) minute
Improved functional walking ability 1min walk test A (d4500) APA, IR, MSk
(p<0.001)
Virtual reality
Ramstrand and Group, II, Weak Virtual reality: CP; GMFCS I–II; No change in standing balance Stabilometry: mean B (b235, b755) APA, IR,
Lygneg ard58 Nintendo 8–17y; n=16 velocity of COP NMR,
Wii, Wii Fit, during modified sensory
and Wii sensory organization
balance board test
No change in reactive balance Response to B (b755) RPA, NMR
perturbation from
platform: EMG on
distal leg muscles
No change in directional control and Rhythmic weight shift A (d4106) APA, IR
synchronization of movement in test
standing
Jelsma et al.59 SSRD, II, Strong Virtual reality: CP; GMFCS I–II; No results reported Goniometry lower limb B (b7100) MSk
Nintendo 7–14y; n=14 Improved balance in standing (p<0.043); Bruininks—Oseretsky A (d450, d4552) APA, IR,
Wii, Wii Fit, no change in running Test of Motor RPA,
and Wii Proficiency: balance sensory,
balance board and running speed, MSk
and agility
subsections
Deterioration in stair function Timed-up and down A (d4551) APA, IR, MSk
stairs test
Brien and SSRD, III, Strong Virtual reality: CP; GMFCS I; 13 Improved high level functional balance Community balance A (d4500, d4552, APA, IR,
Sveistrup60 2D virtual –18y; n=4 and mobility in everyday function (true and mobility scale d4553, d4106, sensory
world change from mean of the baseline d4101)
significant)
Improved functional walking capacity 6-min walk test A (d4501) APA, IR, MSk
(MDC CI 80%, significant)
No change in functional mobility Timed-up and down A (d4551) APA, IR, MSk
stairs
No change in gross motor function GMFM (total) A (d410–d429, APA, IR, MSk

Review
d455)

513
 reviews, were identified (Table I). Sixteen hours of therapy

APA, IR, MSk

voluntary torque; DF, dorsiflexion; PF, plantarflexion; EMG, electromyography; RA, rectus abdominis; OE, obliquus externus; OI, obliquus internus; TrA, transversus abdominis; MDC, mini-
Group, group research design; CP, cerebral palsy; B, ‘body functions and structures’; MSk, musculoskeletal; GMFM, Gross Motor Function Measure; A, ‘activity’; APA, anticipatory postural
APA, IR, SS
component

adjustments; IR, internal representations; TBI, traumatic brain injury; GMFCS, gross motor functional classification system; SSRD, single-subject research design; COP, centre of pressure;
(two 1h sessions/wk for 8wks) improved sitting balance

AP, anteroposterior; ML, mediolateral; CI, confidence interval; DSL, dynamic stability limits; RPA, reactive postural adjustments; NMR, neuromuscular response synergies; MVT, maximal
measured
(BF&S; evidence level II, conduct weak; see Table SIIIa)45
and standing balance (‘activity’; evidence level III, conduct
PC

moderate; see Table SIIIa)47 for ambulant, school-aged

children (GMFCS I–II). In contrast, no improvement was
seen for younger (2–5y), non-ambulant (GMFCS V) chil-
ICF (proposed

dren receiving lower-dose (8h) hippotherapy (one 1h ses-

sion/wk for 8wks) in sitting balance (BF&S and ‘activity’;
B (b760)

B (b770)
evidence level III, conduct moderate; see Table IIIb).46
code)

Two systematic reviews, published in 2011 and 2013, con-

cluded that children with less severe spastic CP were more
max displacement; (2)

likely to show improvements in postural control following

of COP on target and

4m walkway with two

Stabilometry: (1) time

maximum voluntary

hippotherapy compared with more severely affected

Outcome measure

children.53,54
COP displace

force plates

Hippotherapy simulators
Hippotherapy simulators were developed to imitate the
movement of a horse in an attempt to make hippothera-
py accessible in a clinical setting.49 Four studies were
(p=0.023), and voluntary displacement
Improved static and dynamic standing

identified, which included two level II studies and one

[p=0.015] and decreased asymmetry
Improved gait (increased step length
balance: maintaining COP on target

systematic review (Table I). These showed that simulator

training has a mixed ability to improve postural control
at BF&S and ‘activity’ components. Simulator training
for 2.5 hours (one 15min session/wk for 10wks) did not
improve sitting balance (BF&S and ‘activity’) in a hetero-
geneous group of children with CP (GMFCS I–IV,
predominantly non-ambulant GMFCS IV) (evidence level
AP (p<0.05)

II, conduct strong; see Table SIIIa).49 A higher dose of

[p=0.021])

8 hours (two 40min sessions/wk for 6wks) improved

Results

sitting balance (BF&S) in another heterogeneous group

of children with CP (GMFCS II–IV) (evidence level II,
conduct weak; see Table SIIIa).48 A systematic review,53
CP; GMFCS I; 5–

which preceded these two studies, included predomi-

Participants

nantly sources of low-level evidence; although it included

11y; n=10

a meta-analysis of seven studies, which demonstrated

statistically significant improvement in postural control,
this result was confounded by the inclusion of simulator
studies along side other studies of equine-assisted activi-
ties and therapies (hippotherapy [n=3] and therapeutic
biofeedback
Intervention

horse riding [n=3]).53

Visual

Neurodevelopmental therapy
mal detectable change; SS, sensory strategies.

Because NDT has been an evolving concept since the

1940s, studies were included if they met any of the criteria
reported in a previous AACPDM review of NDT.62 As a
Group, II, Weak
evidence level,

result, NDT was identified as a comparison treatment in

and conduct

seven studies, of which six were level II or III (Table I).

Design,

An unspecified dose of NDT alone,41 or NDT (‘based on

rating

the Bobath technique’) combined with ‘conventional reha-

bilitation’ treatments (i.e. joint mobility, muscle strength-
ening, and mobility activities),42 improved postural
Visual biofeedback
Table I: Continued

Ledebt et al.61

alignment (BF&S, two of level II, conduct weak; see Table

SIIIa) over 6 weeks in children with spastic diplegia, aged
References

between 8 months and 10 years.41,42 An unknown dose of

therapy ‘focused on improving walking and balance
through facilitation and normalization of movement

514 Developmental Medicine & Child Neurology 2015, 57: 504–520

patterns’, delivered over 5 weeks, led to an average Treadmill training with no body weight support and
1.80 second decrease in ‘timed-up and go’ test scores in treadmill training with partial or full body weight support
children with quadriplegia and diplegia (GMFCS I–III), Treadmill training includes walking or running on a tread-
aged between 4 years and 11 years (‘activity’; level II, con- mill with F-BWS, P-BWS or no-BWS. Of six studies, two
duct moderate; see Table SIIIa). However, the statistical were level II (Table I). The first included training with no-
significance of this decrease was not reported.44 Eight BWS, and the second involved P-BWS or F-BWS with
hours (two 40min sessions/wk for 6wks) of NDT focusing robotic assistance using the Lokomat.55,56 Treadmill train-
on trunk control improved postural control in sitting ing with no-BWS, for children with CP GMFCS I–II,
(BF&S) in children with spastic diplegia (GMFCS II–V), improved BF&S and ‘activity’ components:56 training for
aged 3 to 10 years (level II, conduct weak, see Table SIIIa).48 7 hours (two 30min sessions/wk for 7wks) improved stand-
The statistical significance of this change was not reported. ing balance (BF&S), which demonstrated improved overall
NDT (composed of stretching, strengthening exercises, balance (‘activity’) (level II, conduct moderate; see Table
standing exercises, postural reactions exercises, reflex-inhib- SIIIa).56 Training with P-BWS or F-BWS and robotic
iting patterns, and gait training exercises) for 72 hours (three assistance (Lokomat) for 15 hours (five 45min sessions/wk
120min sessions/wk for 12wks) improved limits of stability for 4wks), by ambulant and semi-ambulant children with
(BF&S) and standing balance (‘activity’), and reduced fall spastic diplegia (GMFCS II–III), improved standing bal-
risk (‘activity’), in children with spastic diplegic CP, ance (BF&S), and foot loading symmetry with eyes open
GMFCS I–II (evidence level II, conduct moderate; see (BF&S; level II, conduct weak; see Table SIIIa);55 no
online Table SIIIa).50 In contrast, 8 hours (two 30min ses- ‘activity’-level measures were evaluated. One study
sions/wk for 8wks) of NDT did not improve standing bal- reported intermittent discomfort and ‘a few’ skin abrasions
ance (‘activity’) in children with bilateral spastic CP from robotic orthosis use; these issues were resolved by
(GMFCS I–II), aged 4 to 10 years (level III, conduct moder- adjustment of the supports (see Table SIV).63
ate; see Table SIIIa).47
Trunk-targeted training
Progressive resistance exercise Trunk-targeted training involves exercises aimed at
Progressive resistance exercise involves resisted motion or improving trunk muscle strength and control. Two studies
lifting tasks, with structured increases in training loads, to were identified, of which one was level II (Table I).57 In
improve muscle strength. One study (evidence level II) was this study, trunk-strengthening exercises were performed
identified (Table I). Performing resisted ankle and knee whilst participants were positioned on a vibrating platform
exercises for a total of 6 hours (two 15 min sessions/wk for for between 1.5 and 2.5 hours over 4 weeks (two 5–10min
12wks), combined with rehabilitation (12h), did not sessions/wk for 2wks, followed by four or five 5–10min ses-
improve standing balance (BF&S; as measured by stabil- sions/wk for a further 2wks). This protocol improved pos-
ometry [sway path length], p>0.05) or gait kinematics tural alignment (BF&S), increased resting abdominal
(BF&S) in ambulant children with CP (GMFCS I), aged muscle thickness (BF&S), and increased functional muscle
5 to 14 years (level II, conduct strong; see online strength (BF&S). All improvements were maintained after
Table SIIIa).51 4 weeks, except resting muscle thickness of the transversus
abdominis and internal oblique, and performance on the 1-
Reactive balance training minute walk test (evidence level II, conduct moderate; see
Reactive balance training involves repeated practice of Table SIIIa). The link between BF&S (transversus abdo-
balance recovery, when standing on a support surface minis and internal oblique thickness) and ‘activity’ (1min
that is perturbed without warning in a forward, back- walk test) was not evaluated.
ward, or lateral direction. Three studies were identified,
of which two were level II (Table I).50,52 Training using Upper limb interventions
a laboratory-based force platform for approximately Two upper limb intervention studies were identified, each
2 hours (one 20–25min session/d for 5d [100 perturba- using a different approach: constraint-induced movement
tions/session]) improved standing balance (BF&S) in therapy64 or force use therapy65 (see Table SIIa). Both
ambulant children with CP (GMFCS I–II) (level II, con- studies were classified as level IV (demonstrating low levels
duct moderate; see Table SIIIa). Training using the Bio- of evidence), and, therefore, did not meet criteria for fur-
dex balance system, for a higher dose of 18 hours (three ther evaluation.
30min sessions/wk for 12wks), improved limits of stabil-
ity (BF&S) and standing balance (‘activity’), and reduced Virtual reality
fall risk (‘activity’), in children with spastic diplegic CP Virtual reality involves balance training whilst playing
(GMFCS I–II) (evidence level II, conduct moderate; see computer games that create a virtual environment using
Table SIIIa].50 This protocol also included an unspeci- artificial sensory information to simulate real-life experi-
fied duration of anticipatory balance training (voluntary ences or activities.66 Virtual reality is used in rehabilitation
movements of the center of mass to the limits of stabil- to achieve therapy goals within a play environment.66 Out
ity with visual feedback). of eight studies, three were level II or III (Table I).

Review 515
Conflicting results were obtained on the impact of virtual control in ambulant children with CP when provided for
reality on the ICF BF&S and ‘activity’ components. Two at least 16 hours. It appears to impact multiple postural
studies involved the Nintendo Wii Fit. Five hours (four control components, including anticipatory and reactive
25min sessions/wk for 3wks) of supervised Wii Fit training, postural adjustments, and sensory and musculoskeletal
in a physiotherapy clinic, playing games aimed at improv- systems. Effective elements of hippotherapy are proposed
ing standing balance and weight shift, resulted in improved to include the horse’s movement, which has been suggested
standing balance (‘activity’) for children with spastic hemi- to challenge balance, improve posture and strength while
plegia, GMFCS I–II (level I, conduct strong; see Table incorporating sensory input,67 and simulate human gait
SIIIb).59 However, stair climbing was unchanged or, in and the vertical change in the centre of pressure;68 and the
some cases, deteriorated. A higher dose of 12.5 hours (five warmth and rhythm of the horse, which may promote
30min sessions/wk for 5wks) of unsupervised Wii Fit bal- relaxation, thereby reducing spasticity and increasing mus-
ance games at home did not improve standing balance cle length.53 Hippotherapy may also provide the opportu-
(BF&S and ‘activity’) for ambulant children with hemiple- nity for trunk muscle motor control training, similar to
gia or diplegia, GMFCS I–II (level II, conduct weak; see that achieved with therapy ball activities. The novelty of
Table SIIIa).58 The third study involved a 2D virtual world hippotherapy promotes extended engagement, and the
game. Intensive training for 7.5 hours (ten 45min sessions opportunity for massed practice of reactive postural
over 1wk) improved functional balance (‘activity’) for ado- adjustments.30 Two previous systematic reviews support
lescents with CP GMFCS I (level III, conduct strong; see the current findings, including the conclusion that hippo-
Table SIIIb).60 therapy is more effective for ambulant children with CP
(GMFCS I–II) than for children with more severe CP.53,54
Visual biofeedback Although moderate evidence supports the use of hippo-
One study of visual biofeedback was identified, which had therapy, the use of hippotherapy simulators cannot be sup-
a level II evidence rating (Table I). It involved standing on ported by this systematic review because of conflicting
a balance platform in a laboratory (no specifications pro- outcomes between studies and the levels of the ICF com-
vided) and keeping the centre of pressure, represented as a ponents considered. The limitations of the two level II
red dot on a computer screen, static, or shifting it to a tar- studies48,49 which explored the use of hippotherapy simula-
get. Anticipatory training for 9 hours (three 30min ses- tors, when compared with the studies of hippotherapy,
sions/wk for 6wks) improved static standing balance included heterogeneity of participants (GMFCS I–IV); var-
(BF&S) and dynamic standing balance (BF&S) in ambulant ied intervention dose (2.5–8h); and a lack of individualiza-
children with CP, GMFCS I (level II, conduct weak; see tion in training protocols. Future research could address
Table SIIIa).61 No ‘activity’-level measures were included. these limitations. The systematic review53 of hippotherapy
simulator use provided little additional information because
DISCUSSION outcomes were pooled with both hippotherapy and thera-
This systematic review analysed 45 studies of children with peutic horse riding.
CP, and presented information on the use of 13 different Treadmill training with no-BWS was the second most
postural control interventions; 40 of these studies, and nine common approach, gaining moderate support, with one
of these interventions, had not been included in a previous level II study.56 For ambulant children with CP, training
systematic review on postural control interventions, pub- for at least 7 hours improved musculoskeletal and sensory
lished in 2005. There is moderate evidence to support the components, and anticipatory postural adjustments. The
use of five of these interventions: hippotherapy, treadmill use of a specific protocol, guiding the gradual increase in
training with no-BWS, trunk-targeted training, reactive treadmill speed for children who are walking independently
balance training, and gross motor task training. There is (GMFCS I–II), appears to be critical to the success of this
only weak or conflicting evidence to support the use of six approach. In contrast, there was only weak evidence sup-
of these as effective postural control interventions: hippo- porting the use of treadmill training with F-BWS or P-
therapy simulators, treadmill training with P-BWS or BWS for semi-ambulant children (GMFCS II–III), despite
F-BWS (including robotic assistance), NDT, virtual real- a higher dose of 15 hours (5wks94 sessions945min).55
ity, visual biofeedback, and FES. For the remaining two This is not surprising given that the goal of treadmill
interventions, either there were no high-level protocols training is to improve anticipatory and reactive postural
(level I–III) evaluating efficacy (upper limb interventions) adjustments, which are reduced when robotic support is
or the evidence suggests that there is no improvement in present.69 A systematic review of P-BWS treadmill train-
postural control (progressive resistance exercise). With the ing, including robotic assistance, for children with CP,
possible exception of NDT, it was noted that all of the supports these findings, concluding that while treadmill
effective interventions were reported by studies involving training with P-BWS is effective for improving important
ambulant children with CP (GMFCS I–III). gait elements, such as endurance, speed, and gait pat-
Of the five intervention types that gained moderate sup- terns,70 there is a need for further research, with standard-
port, hippotherapy was the most commonly reported (five ized protocols, to determine if it is effective in improving
level I–III studies). Hippotherapy improved postural postural control for children with lower-level mobility.

516 Developmental Medicine & Child Neurology 2015, 57: 504–520

 Trunk-targeted training (one level II study) was a novel functional goals. This approach has high clinical utility
combination of interventions (strengthening exercises on a because it requires no technical equipment, and can be
vibrating platform) that was shown, by moderate-level evi- delivered in most settings. Further research is required to
dence, to improve anticipatory, sensory, and musculoskele- determine the minimum required dose, as total training
tal components of postural control in ambulatory children times ranged from 10 to 30 hours in the studies consid-
with CP, when provided in short bursts totalling only 1.5 ered by this review.
to 2.5 hours.57 The authors were unable to determine the Six intervention approaches showed weak or conflicting
relative effectiveness of the ‘strength’ and ‘vibration’ com- evidence. Hippotherapy simulators and treadmill training
ponents of this protocol. They proposed that vibration with P-BWS or F-BWS have been addressed in previous
‘activated weak and dormant muscles’, and reduced the paragraphs. Overall, NDT was the most commonly evalu-
need to use weights or high repetitions to strengthen ated intervention in this category (six level II–III studies),
abdominal muscles for postural control.57 A separate as it was frequently included as a comparison when testing
review of vibration therapy supports this concept, propos- potential new interventions. Unfortunately, although it
ing that it perturbs the gravitational field to activate and appeared in six high-level studies, it is difficult to assign
strengthen muscles, stimulate peripheral sensory receptors, more than a weak evidence level to this intervention for
and evoke postural responses.71 In children with CP, vibra- several reasons. Primarily, studies often lacked a clear
tion alone can improve gross motor function and bone description of the specific framework or intervention con-
density;72 further research is needed to establish the effects tent of the NDT component provided, or how fidelity was
of vibration on postural control. Trunk muscle motor con- maintained across participants or therapists. Furthermore,
trol training has been shown to improve anticipatory pos- it was combined with other therapies and so the effective
tural adjustments73,74 and gait57 in adult populations, and postural control element was not possible to determine. At
so appears to be a contender for further treatment develop- times, dose was delivered in unspecified or variable for-
ment. A mechanism for achieving efficient trunk muscle mats. In other cases, participants were heterogeneous in
activation and strengthening in children with atypical mus- GMFCS or age. It would be useful to address these meth-
cle tone, spasticity, and motor control issues remains a odological limitations in future research so that the relative
challenge. benefit of NDT for postural control in children with CP
There was moderate evidence to support the use of reac- can be more clearly understood, particularly for children
tive balance training alone as an intervention to improve with more severe motor impairment, where it may be of
reactive postural adjustments in ambulant children with particular benefit.
CP, with at least 2 hours of training, but it did not Virtual reality (three level I–III studies)58–60 is gaining
improve self-initiated movement control (Gross Motor popularity as a result of the increasing availability of rele-
Function Measure [GMFM]).52 However, when reactive vant home- and laboratory-based technologies. Although it
balance training (Biodex system) was combined with antici- received weak support as a postural control intervention, it
patory training (voluntary leaning towards the limits of sta- is proposed to influence anticipatory and sensory compo-
bility), both reactive postural adjustments and self-initiated nents through practice of voluntary movement, in conjunc-
movement control did improve.50 A comparison of these tion with feedback through visual (screen) and/or tactile
two studies confirms the impact of specificity of training (hand control) modalities. It is suggested to fulfil three
on postural control. Reactive balance training may improve important requisites for motor learning: (1) movement rep-
recovery from external perturbations, such as a trip or etition; (2) active participation; and (3) performance feed-
movements when standing on a bus; if the goal of therapy back.76 Lack of rigorous research conduct makes it difficult
is to improve control of self-initiated motor function, then to draw conclusions about the effects of virtual reality on
anticipatory training may be more appropriate. More postural control. One author recommends that virtual real-
research is warranted to determine the extent to which ity be used only as an adjunct to other therapies, and not
reactive balance training alone influences other postural as a replacement.59 A review of virtual reality use by chil-
control elements, ICF components, and postural orienta- dren with CP, which included motor but not postural con-
tion, and not just stability. trol outcomes, agrees that substantial benefits could be
The use of gross motor task training (two level II stud- gained from using virtual reality; however, the current evi-
ies) was supported by moderate-level evidence to improve dence is weak.66
postural control in ambulant children with CP when pro- Visual biofeedback (one level I–III study)61 and FES
vided for at least 10 hours.43,44 Gross motor task training (two level I–III studies)41,42 are the remaining two inter-
affects most postural control elements because the devel- vention types to show weak evidence. Both demonstrated
opment of efficient anticipatory and reactive postural broad efficacy in improving postural stability and orienta-
adjustments occurs in parallel with the attainment of a tion. Both approaches require more detailed reporting of
gross motor skill, and variability in practice can then participant groups and/or evaluation of higher participant
fine-tune control of that task.3,75 Both papers also numbers, as well as more detailed treatment protocols, to
highlight the concept of specificity of practice,43 which establish a sufficient level of guidance for evidence-based
is important when designing programmes to address practice.

Review 517
 Two interventions showed no impact on postural con- excluded from this review and, therefore, some effective
trol. Progressive resistance exercise (one level II study)51 exercise interventions may not be represented here. This
showed no effect on children with CP. This finding is sup- highlights the importance of including postural control
ported by studies with adult populations, which have sug- measures in future studies of exercise interventions. This
gested that, rather than strengthening, neuromuscular review also reports on many positive gains made in pos-
control training is required to improve postural control.77 tural control intervention design and evaluation in the last
This recommendation is consistent with the moderate sup- decade. It highlights where further research is needed if
port found for functional strength training during gross intervention types are to achieve higher levels of evidence.
motor tasks, as discussed earlier. Finally, there were no In particular, a focus on improving treatment description
high-level upper limb interventions suitable for reporting and fidelity, establishing dosage and measuring both short-
at this stage. Ballaz et al.65 contend that investigating pos- and long-term effects for subgroups in the CP population,
tural control outcomes following constraint-induced move- is required. Finally, there is a need for further research
ment therapy and force use therapy is important to assess into programmes that are innovative, provide multidimen-
the effect of reducing asymmetric upper limb function on sional impact on components of postural control, and
balance (postural stability) and postural symmetry (orienta- include functional exercise. Mainstream programmes, such
tion). More rigorous research is required before it is possi- as yoga, Pilates, and tai chi, which are popular with adults,
ble to recommend these approaches specifically as effective are now being offered to children. The intent of these pro-
postural control interventions. grammes is to improve posture and motor control; how-
ever, their efficacy for children with CP has yet to be
Outcome measures investigated.
This review highlights the importance of selecting appropri-
ate outcome measures when assessing the potential impact CONCLUSION
of each intervention. Progress has been made, but broader Exercise interventions documented to improve postural
evaluation is required to establish a thorough understanding control for children with CP are increasing. This review
of the effects of each intervention on (1) overall postural sta- has identified five potentially effective interventions, six
bility and orientation functions; (2) postural control ele- that require more investigation, and two that are probably
ments according to the systems theory approach; and (3) ineffective. Further research is required for children with
‘function’ (BF&S), ‘activity’, and ‘participation’ according to different types and severities of CP to establish (1) respon-
the ICF. The majority of studies reported in this review sive and reliable postural control outcome measures; (2)
measured postural stability outcomes, especially the ability effective treatment selection and dose guidelines; and (3)
to ‘maintain’ or ‘restore’ balance; however, fewer studies possible efficacy of mainstream exercise interventions that
have measured the ability to ‘achieve’ balance, or improve have demonstrated effectiveness for improving postural
orientation of body segments relative to the task and envi- control in adults with brain injury, such as Pilates, yoga,
ronment. Furthermore, new measures may better quantify and tai chi.
the impact on postural control elements; for example, the
recently described Balance Evaluation Systems Test (BES- A CK N O W L E D G E M E N T S
Test)27 or the miniBESTest,78 which were developed to dif- We would like to thank the Research and Innovation Unit of the
ferentiate the six postural control elements of the systems Cerebral Palsy League, Queensland Australia, who supported this
control approach. Neither of these tests has been validated study, including Jill Duff, Information and Resource Coordinator,
for use with children. Finally, our review shows that out- and Robyn Smedley, Research and Evaluation Officer, for their
comes were generally measured at the ICF BF&S (impair- assistance in accessing the full-text articles required for this
ment) or ‘activity’ (basic motor skill) levels, with no review. The authors state that they had no interests that could be
exploration of the impact on ‘participation’. There is great perceived as posing a conflict or bias.
scope for future research to explore the potential carry-over
from postural control interventions to ‘activity’ and ‘partici- SUPPORTING INFORMATION
pation’ functions, which may be assessed by using, for exam- The following additional material may be found online:
ple, the Canadian Occupational Performance Measure, Goal Table SI: Levels of evidence for group and single-subject
Attainment Scale, or the The Children’s Assessment of Par- design studies.
ticipation and Enjoyment. Table SII: (a) Summary of group studies – interventions and
participants. (b) Summary of single-subject studies – interventions
Limitations and future direction for research and participants.
It is possible that other exercise interventions may affect Table SIII: (a) Conduct of group design studies. (b) Conduct
postural control to some degree. If an intervention study of single-subject design studies. (c) Conduct of systematic review.
did not include a postural control outcome measure, it was Table SIV: Reported adverse events.

518 Developmental Medicine & Child Neurology 2015, 57: 504–520

REFERENCES
1. Oskoui M, Coutinho F, Dykeman J, Jette N, Prings- 17. Woollacott MH, Shumway-Cook A. Postural dysfunc- 35. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA
heim T. An update on the prevalence of cerebral palsy: tion during standing and walking in children with cere- statement for reporting systematic reviews and meta-
a systematic review and meta-analysis. Dev Med Child bral palsy: what are the underlying problems and what analyses of studies that evaluate health care interven-
Neurol 2013; 55: 509–19. new therapies might improve balance? Neural Plast 2005; tions: explanation and elaboration. PLoS Med 2009; 6:
2. Rosenbaum P, Paneth N, Leviton A, et al. A report: the 12: 211–9. e1000100.
definition and classification of cerebral palsy April 2006. 18. Kurz MJ, Arpin DJ, Corr B. Differences in the 36. Moher D, Liberati A, Tetzlaff J, Altman DG. Research
Dev Med Child Neurol Suppl 2007; 109: 8–14. dynamic gait stability of children with cerebral palsy and methods & reporting–Preferred reporting items for sys-
3. Hadders-Algra M, Carlberg EB. Postural Control: A typically developing children. Gait Posture 2012; 36: tematic reviews and meta-analyses: the PRISMA state-
Key Issue in Developmental Disorders. London: Mac 600–4. ment. Br Med J 2009; 339: 332.
Keith Press, 2008. 19. Van Der Heide JC, Fock JM, Otten B, Stremmelaar E, 37. Levels Evidence Working Group. The Oxford Levels
4. Nashner L, Shumway-Cook A, Marin O. Stance posture Van Eykern LA, Hadders-Algra M. Postural control of Evidence 2. Available from: http://www.cebm.net/
control in select groups of children with cerebral palsy: during reaching in preterm children with cerebral palsy. index.aspx?o=5653 (accessed 22 May 2013).
deficits in sensory organization and muscular coordina- Dev Med Child Neurol 2004; 46: 253–66. 38. Logan LR, Hickman RR, Harris SR, Heriza CB. Single
tion. Exp Brain Res 1983; 49: 393–409. 20. Larnert G, Ekberg O. Positioning improves the oral and subject research design: recommendations for levels of
5. Woollacott MH, Burtner P, Jensen J, Jasiewicz J, Ron- pharyngeal swallowing function in children with cerebral evidence and quality rating. Dev Med Child Neurol 2008;
cesvalles N, Sveistrup H. Development of postural palsy. Acta Paediatr 1995; 84: 689–93. 50: 99–103.
responses during standing in healthy children and chil- 21. Imms C. Children with cerebral palsy participate: 39. Oxman AD, Guyatt GH. Validation of an index of the
dren with spastic diplegia. Neurosci Biobehav Rev 1998; a review of the literature. Disabil Rehabil 2008; 30: quality of review articles. J Clin Epidemiol 1991; 44:
22: 583–9. 1867–84. 1271–8.
6. de Graaf-Peters VB, Blauw-Hospers CH, Dirks T, Bak- 22. Shumway-Cook A, Woollacott MH. Normal Postural 40. World Health Organization. International Classification
ker H, Bos AF, Hadders-Algra M. Development of pos- Control. In: Motor Control: Translating Research Into of Functioning, Disability, and Health: Children &
tural control in typically developing children and children Clinical Practice, 4th edn. Baltimore, MD: Lippincott Youth Version (ICF-CY). Geneva: World Health Orga-
with cerebral palsy: possibilities for intervention? Neurosci Williams & Wilkins, 2012: 161–94. nization, 2010.
Biobehav Rev 2007; 31: 1191–200. 23. Massion J. Postural control system. Curr Opin Neurobiol 41. Park ES, Park CI, Lee HJ, Cho YS. The effect of elec-
7. Bigongiari A, De Andrade E, Souza F, Franciulli PM, 1994; 4: 877–87. trical stimulation on the trunk control in young children
Neto SER, Araujo RC, Mochizuki L. Anticipatory and 24. Saether R, Helbostad JL, Riphagen II, Vik T. Clinical with spastic diplegic cerebral palsy. J Korean Med Sci
compensatory postural adjustments in sitting in children tools to assess balance in children and adults with cere- 2001; 16: 347–50.
with cerebral palsy. Hum Movement Sci 2011; 30: 988–97. bral palsy: a systematic review. Dev Med Child Neurol _ Dogan A, Arslan MD, G€
42. Karabay I, ulseren D, Ozgirgin
8. Girolami GL, Shiratori T, Aruin AS. Anticipatory pos- 2013; 55: 988–99. N. Effects of functional electrical stimulation on trunk
tural adjustments in children with hemiplegia and diple- 25. Santos MJ, Kanekar N, Aruin AS. The role of anticipa- control in children with diplegic cerebral palsy. Disabil
gia. J Electromyogr Kinesiol 2011; 21: 988–97. tory postural adjustments in compensatory control of Rehabil 2012; 34: 965–70.
9. Tomita H, Fukaya Y, Ueda T, et al. Deficits in task- posture: 1. Electromyographic analysis. J Electromyogr 43. Katz-Leurer M, Rotem H, Keren O, Meyer S. The
specific modulation of anticipatory postural adjustments Kinesiol 2010; 20: 388–97. effects of a ‘home-based’ task-oriented exercise pro-
in individuals with spastic diplegic cerebral palsy. J Neu- 26. Horak FB. Clinical measurement of postural control in gramme on motor and balance performance in children
rophysiol 2011; 105: 2157–68. adults. Phys Ther 1987; 67: 1881–5. with spastic cerebral palsy and severe traumatic brain
10. Burtner P, Qualls C, Woollacott M. Muscle activation 27. Horak FB, Wrisley DM, Frank J. The balance evalua- injury. Clin Rehabil 2009; 23: 714–24.
characteristics of stance balance control in children with tion systems test (BESTest) to differentiate balance defi- 44. Salem Y, Godwin EM. Effects of task-oriented training
spastic cerebral palsy. Gait Posture 1998; 8: 163–74. cits. Phys Ther 2009; 89: 484–98. on mobility function in children with cerebral palsy.
11. Roncesvalles M, Woollacott M, Burtner P. Neural fac- 28. Campbell SK. Efficacy of physical therapy in improving NeuroRehabilitation 2009; 24: 307–13.
tors underlying reduced postural adaptability in children postural control in cerebral palsy. Pediatr Phys Ther 45. Kang H, Jung J, Yu J. Effects of hippotherapy on the
with cerebral palsy. NeuroReport 2002; 13: 2407–10. 1990; 2: 135–40. sitting balance of children with cerebral palsy: a ran-
12. Woollacott M, Shumway-Cook A, Hutchinson S, Ciol 29. Westcott SL, Burtner P. Postural control in children. domized control trial. J Phys Ther Sci 2012; 24: 833–6.
M, Price R, Kartin D. Effect of balance training on Phys Occup Ther Pediatr 2004; 24: 5–55. 46. Hamill D, Washington K, White OR. The effect of hip-
muscle activity used in recovery of stability in children 30. Harris SR, Roxborough L. Efficacy and effectiveness of potherapy on postural control in sitting for children
with cerebral palsy: a pilot study. Dev Med Child Neurol physical therapy in enhancing postural control in chil- with cerebral palsy. Phys Occup Ther Pediatr 2007; 27:
2005; 47: 455–61. dren with cerebral palsy. Neural Plast 2005; 12: 229–43. 23–42.
13. Westcott SL, Lowes LP, Richardson PK. Evaluation of 31. Law K, Howick J. Glossary. http://www.cebm.net 47. Kwon J-Y, Chang HJ, Lee JY, Hia Y, Lee PK, Kim
postural stability in children: current theories and assess- (accessed 16 June 2014). YH. Effects of hippotherapy on gait parameters in chil-
ment tools. Phys Ther 1997; 77: 629–45. 32. Portney L, Watkins M. Foundations of Clincial dren with bilateral spastic cerebral palsy. Arch Phys Med
14. Lowes LP, Westcott SL, Palisano RJ, Effgen SK, Orlin Research: Applications to Practice. Upper Saddle River, Rehab 2011; 92: 774–9.
MN. Muscle force and range of motion as predictors of NJ: Prentice Hall Publishing, 2008. 48. Borges MBS, Werneck MJDS, Silva MDLD, Gandolfi
standing balance in children with cerebral palsy. Phys 33. Treatment Outcomes Committee AACPDM. AACPDM L, Pratesi R. Therapeutic effects of a horse riding simu-
Occup Ther Pediatr 2004; 24: 57–77. Methodology to Develop Systematic Reviews of Treat- lator in children with cerebral palsy. Arq Neuropsiquiatr
15. Woollacott M, Burtner P. Neural and musculoskeletal ment Interventions (Revision 1.2) 2008 Version. http:// 2011; 69: 799–804.
contributions to the development of stance balance con- www.aacpdm.org/resources/systematicReviewsMethodol- 49. Herrero P, G  et al. Study
omez-Trullen EM, Asensio A,
trol in typical children and in children with cerebral ogy.pdf (accessed 9 July 2013). of the therapeutic effects of a hippotherapy simulator in
palsy. Acta Paediatr 1996; 85: 58–62. 34. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred children with cerebral palsy: a stratified single-blind ran-
16. Liao H-F, Hwang A-W. Relations of balance function reporting items for systematic reviews and meta-analy- domized controlled trial. Clin Rehabil 2012; 26: 1105–13.
and gross motor ability for children with cerebral palsy. ses: the PRISMA statement. PLoS Med 2009; 6: 50. El-Shamy SM, Abd El Kafy EM. Effect of balance train-
Percept Motor Skills 2003; 96: 1173–84. e1000097. ing on postural balance control and risk of fall in chil-

Review 519
 dren with diplegic cerebral palsy. Disabil Rehabil 2014; 59. Jelsma J, Pronk M, Ferguson G, Jelsma-Smit D. The 69. Borggraefe I, Schaefer JS, Klaiber M, et al. Robotic-
36: 1176–83. effect of the Nintendo Wii Fit on balance control and assisted treadmill therapy improves walking and standing
51. Bandholm T, Jensen BR, Nielsen LM, et al. Neuroreha- gross motor function of children with spastic hemiplegic performance in children and adolescents with cerebral
bilitation with versus without resistance training after cerebral palsy. Dev Neurorehabil 2013; 16: 27–37. palsy. Eur J Paediatr Neuro 2010; 14: 496–502.
botulinum toxin treatment in children with cerebral 60. Brien M, Sveistrup H. An intensive virtual reality pro- 70. Mattern-Baxter K. Effects of partial body weight sup-
palsy: a randomized pilot study. NeuroRehabilitation gram improves functional balance and mobility of ado- ported treadmill training on children with cerebral palsy.
2012; 30: 277–86. lescents with cerebral palsy. Pediatr Phys Ther 2011; 23: Pediatr Phys Ther 2009; 21: 12–22.
52. Shumway-Cook A, Hutchinson S, Kartin D, Price R, 258–66. 71. Pozo-Cruz B, Adsuar J, Parraca J, Pozo-Cruz J, Olivares
Woollacott M. Effect of balance training on recovery of 61. Ledebt A, Becher J, Kapper J, et al. Balance training with PR, Gusi N. Using whole-body vibration training in
stability in children with cerebral palsy. Dev Med Child visual feedback in children with hemiplegic cerebral palsy: patients affected with common neurological diseases: a
Neurol 2003; 45: 591–602. effect on stance and gait. Mot Control 2005; 9: 459. systematic literature review. J Altern Complem Med 2012;
53. Zadnikar M, Kastrin A. Effects of hippotherapy and 62. Butler C, Darrah J. Effects of neurodevelopmental treat- 18: 29–41.
therapeutic horseback riding on postural control or bal- ment (NDT) for cerebral palsy: an AACPDM evidence 72. Rauch F. Vibration therapy. Dev Med Child Neurol 2009;
ance in children with cerebral palsy: a meta-analysis. report. Dev Med Child Neurol 2001; 43: 778–90. 51: 166–8.
Dev Med Child Neurol 2011; 53: 684–91. 63. Kurz MJ, Corr B, Stuberg W, Volkman KG, Smith N. 73. Endleman I, Critchley DJ. Transversus abdominis and
54. Tseng S-H, Chen H-C, Tam K-W. Systematic review Evaluation of lower body positive pressure supported obliquus internus activity during pilates exercises: mea-
and meta-analysis of the effect of equine assisted activi- treadmill training for children with cerebral palsy. Pedi- surement with ultrasound scanning. Arch Phys Med Rehab
ties and therapies on gross motor outcome in children atr Phys Ther 2011; 23: 232–9. 2008; 89: 2205–12.
with cerebral palsy. Disabil Rehabil 2013; 35: 89–99. 64. Zipp GPJ, Winning S. Effect of constraint induced 74. Tsao H, Hodges PW. Immediate changes in feedfor-
55. Dru_zbicki M, Rusek W, Szczepanik M, Dudek J, Snela movement therapy on gait, balance, and functional loco- ward postural adjustments following voluntary motor
S. Assessment of the impact of orthotic gait training on motor ability. Pediatr Phys Ther 2012; 24: 64–8. training. Exp Brain Res 2007; 181: 537–46.
balance in children with cerebral palsy. Acta Bioeng Bio- 65. Ballaz L, Huffenus A-F, Lamarre C, Koclas L, Lemay 75. Ledebt A, Bril B, Breniere Y. The build-up of anticipa-
mech 2010; 12: 53–8. M. Effect of Forced Use Therapy on posture in children tory behaviour an analysis of the development of gait
56. Grecco LA, Tomita SM, Christov~ao TC, Pasini H, with hemiplegic cerebral palsy: a pilot study. J Rehabil initiation in children. Exp Brain Res 1998; 120: 9–17.
Sampaio LMM, Oliveira CS. Effect of treadmill gait Med 2012; 44: 268–71. 76. Harders M. Virtual Reality for Rehabilitation. Virtual
training on static and functional balance in children with 66. Snider L, Majnemer A, Darsaklis V. Virtual reality as a Reality in Medicine. London: Springer, 2012: 161.
cerebral palsy: a randomized controlled trial. Braz J Phys therapeutic modality for children with cerebral palsy. 77. Hodges PW, Moseley GL. Pain and motor control of
Ther 2013; 17: 17–23. Dev Neurorehabil 2010; 13: 120–8. the lumbopelvic region: effect and possible mechanisms.
57. Unger M, Jelsma J, Stark C. Effect of a trunk-targeted 67. Shurtleff TL, Engsberg JR. Changes in trunk and J Electromyogr Kinesiol 2003; 13: 361–70.
intervention using vibration on posture and gait in chil- head stability in children with cerebral palsy after hippo- 78. Franchignoni F, Horak F, Godi M, Nardone A, Giord-
dren with spastic type cerebral palsy: a randomized con- therapy: a pilot study. Phys Occup Ther Pediatr 2010; 30: ano A. Using psychometric techniques to improve the
trol trial. Dev Neurorehabil 2013; 16: 79–88. 150–63. Balance Evaluation System Test: the mini-BESTest. J
58. Ramstrand N, Lygneg
ard F. Can balance in children with 68. Bertoti DB. Effect of therapeutic horseback riding on Rehabil Med 2010; 42: 323–31.
cerebral palsy improve through use of an activity promot- posture in children with cerebral palsy. Phys Ther 1988;
ing computer game? Technol Health Care 2012; 20: 531–40. 68: 1505–12.

520 Developmental Medicine & Child Neurology 2015, 57: 504–520