DEVELOPMENTAL MEDICINE & CHILD NEUROLOGY ORIGINAL ARTICLE

Walking ability and predictors of performance on the 6-minute walk

test in adults with spastic cerebral palsy
GRETHE MAANUM 1 | REIDUN JAHNSEN 1 , 2 | KATHRINE F FRØSLIE 1 , 2 , 3 | KERSTIN L LARSEN 1 | ANNE KELLER 4

1 Sunnaas Rehabilitation Hospital, Nesoddtangen, and Faculty of Medicine, University of Oslo, Oslo, Norway. 2 Oslo University Hospital, Rikshospitalet, Oslo, Norway.
3 Department of Biostatistics, Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway. 4 Oslo University Hospital, Ullevaal Hospital, Oslo, Norway.
Correspondence to Dr Grethe Maanum, Sunnaas Rehabilitation Hospital, 1450 Nesoddtangen, and Faculty of Medicine, University of Oslo, Oslo, Norway. E-mail: grethe.manum@sunnaas.no

PUBLICATION DATA
Accepted for publication 7th December 2009.
Published online 12th February 2010.
LIST OF ABBREVIATIONS
6MWT 6-minute walk test
6MWD 6-minute walk test distance
FMS Functional Mobility Scale
SF-36 General Health Survey Short-Form 36
TUG Timed Up and Go
AIM To describe walking ability and identify factors predicting walking capacity in adults with
spastic cerebral palsy (CP) assessed with the 6-minute walk test (6MWT).
METHOD A cross-sectional clinical study as part of the recruitment process for a randomized con-
trolled trial on the effects of botulinum toxin A. Data analysed were the 6MWT, Timed Up and Go
(TUG) test, Borg Scale, spasticity, muscle strength, popliteal angle, pain, fatigue, type of CP, foot
deformity, Gross Motor Function Classification System (GMFCS) levels, Functional Mobility Scale
(FMS) scores, a gait questionnaire, interview, and demographic data.
RESULTS In total, 126 persons were included (53 males, 73 females; mean age 39y [SD 12y]; 59
with unilateral and 67 with bilateral spastic CP; GMFCS level I, n=12; level II, n=94; level III, n=20).
Mean distance on the 6MWT was 485m (SD 95m) with FMS scores reflecting independent walking
performance in daily life. Multiple regression analysis identified sex, type of CP, popliteal angle,
pain, and TUG values as significant predictors, with TUG values as the strongest predictor
(standardized regression coefficient =)0.57, p<0.001).
INTERPRETATION Our results demonstrate that 39% of the participants had declined one GMFCS
level from adolescence to their present age, and that the TUG was the strongest predictor for the
6MWT. This implicates the importance of focusing specifically on the different elements of func-
tional mobility in further studies.

Cerebral palsy (CP) is a group of permanent disorders of
movement and posture, often accompanied by secondary mus-
culoskeletal problems.1 Literature on children with CP sup-
ports a link between the levels of impairment and walking
ability.1 However, the relation between self-reported declined
walking ability and data on clinical measurements in adults
needs more investigation.
Surveys of adults with CP report that contractures, spastic-
ity, pain, fatigue, as well as reduced muscle strength and bal-
ance, are associated with declined walking ability.2–4
The Gross Motor Function Classification System
(GMFCS) categorizes gross motor function into five levels,
with level I describing the highest level of function and level V
the lowest.5 A study of individuals with CP aged up to
21 years showed that those classified at GMFCS levels III to
V have their peak motor function in childhood, whereas indi-
viduals at levels I or II have a stable gross motor function.6
However, surveys of adults with CP indicate that even persons
at GMFCS levels I or II experience increasing walking diffi-
culties in young adulthood.4
Adequate postural stability, selective motor control, muscle
strength, and endurance are necessary prerequisites for per-
forming basic gross motor skills such as walking, including the
complex ability to stay upright and be moving at the same
time.7 Walking in the community also requires sensory, per-
ceptual, and cognitive abilities to anticipate and react ade-
quately to crowds and unexpected events.8
In studies on gait and mobility ⁄ balance, the 6-minute walk
test (6MWT)9 and the Timed Up and Go (TUG) test10 are
frequently used measures for the activity dimension of the
International Classification of Functioning, Disability and
Health (ICF).11 The 6MWT, which measures the maximum
distance a person can walk within 6 minutes, was originally
developed as an endurance measure in chronic heart failure.12
For elderly and stroke populations, the 6MWT is considered
a general measure of overall functional walking capacity,
involving the integrated response of multiple body sys-
tems.9,13,14 The TUG was originally developed as a balance
test for the elderly.15 By measuring, in time, the ability to
stand up, walk, turn around, and sit down, the TUG captures
the complex interaction between balance and movement,
including planning, initiating, executing, and completing a ser-
ies of linked movements that are common in daily activities.8
The purpose of this study was to describe walking ability in
a population of adults with spastic CP experiencing declined
walking function, and to identify clinical and demographic

e126 DOI: 10.1111/j.1469-8749.2010.03614.x ª The Authors. Journal compilation ª Mac Keith Press 2010

factors predicting walking capacity, as measured with the
6MWT.
METHOD
This study is based on data from the baseline screening before
inclusion in a randomized controlled trial on the effects of bot-
ulinum toxin A in adults with spastic CP at the Sunnaas Reha-
bilitation Hospital, Oslo, Norway.
The participants were recruited through advertisements in
newspapers and on the websites of the Norwegian Cerebral
Palsy Association and Sunnaas Rehabilitation Hospital from
2007 to 2008. The first author was in contact with all the
responders. Potential eligible responders were questioned by
semi-structured telephone interview about changes in walking
function from adolescence to their present age.
Inclusion criteria were (1) uni- or bilateral spastic CP, (2)
age 18 to 65 years, (3) GMFCS levels I to III, (4) ability to
walk independently continuously for 6 minutes, (5) ability to
provide informed consent, and (6) a phone interview reported
declined walking ability compared with adolescence such as
reduced walking distance, increasing stiffness, spasticity, pain,
or balance difficulties. Exclusion criteria were (1) other condi-
tions that could affect the walking ability, (2) orthopaedic sur-
gery within the past 18 months, (3) injections of botulinum
toxin A within the preceding 3 months, (4) pregnancy, and (5)
intellectual disability documented in medical records.
The Commissioner for the Protection of Privacy in
Research and the Regional Ethics Committee in south-eastern
Norway approved the study. All participants gave written
informed consent.
Measurements
A test protocol provided detailed descriptions and instructions
for all data elements collected during one visit to Sunnaas
Rehabilitation Hospital. Data from medical records, classifica-
tion of type of CP according to the Surveillance of Cerebral
Palsy in Europe,16 functional level by GMFCS,5,17 and Func-
tional Mobility Scale (FMS),18 6MWT, TUG test, and self-
administered questionnaires were performed or collected by
the first author. The first author, assisted by one of three
skilled physiotherapists, used a consistent technique with stan-
dardized procedures for the outcomes of body structure ⁄ func-
tion. No formal reliability testing was performed; however,
the measurements were the same as those used in the motion
laboratories at Sunnaas Rehabilitation Hospital and at Oslo
University Hospital, developed and adjusted through joint
training sessions with consensus processes.
Activity ⁄ participation
Walking ability was determined using the GMFCS, the FMS,
and a questionnaire on walking function (Appendix SI, pub-
lished online). The GMFCS level at adolescence was rated on
information from the participants. Present GMFCS level was
rated based on observations. The GMFCS has demonstrated
validity and reliability for classification of present and earlier
gross motor function in adults with CP (intraclass correlation
coefficient [ICC]=0.93–0.95).17 The FMS, a 6-level scale
14698749, 2010, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2010.03614.x by Cochrane Chile, Wiley Online Library on [30/08/2023]. 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
What this paper adds
• Thirty-nine per cent of ambulant adults with spastic CP had declined
one GMFCS level from adolescence to their present age.
• 6MWD and TUG values varied significantly over the GMFCS levels.
• The TUG test was the strongest predictor of 6MWT.
• Age effects and TUG elements need specific investigation.
rating mobility over three distances (5m, 50m, and 500m), was
used as a measure of walking performance and rated based on
information from the participants. The FMS has established
psychometric properties for children with CP (interrater reli-
ability ICC=0.95; construct validity r=0.75–0.89).18 The self-
administered questionnaire on walking function (Appendix SI)
included information on causes of change in walking ability,
walking performance in different contexts, and on limiting
symptom(s) related to walking. It was developed by studying
other questionnaires used in adult CP research,2,3 and through
workshops with the present authors and the named person
acknowledged. Applicability was assessed in the ordinary clinic
before the start of the study.
The 6MWT was used to measure functional walking capac-
ity. The participants were instructed to walk as fast and as long
as possible for 6 minutes. Each participant underwent the
6MWT in a quiet hospital corridor with a 30m marked track,
and chairs as turning points instead of cones to support turn-
ing if needed. The investigator repeated set phrases every min-
ute during the walk.9 The total distance walked was recorded
as the 6MWT distance (6MWD, to the nearest metre). The
6MWT has demonstrated test–retest reliability for adults with
CP (ICC=0.97).19 Perceived exertion was graded immediately
after 6 minutes using the Borg Scale (rating of perceived exer-
tion, 6–20), a measure of exercise intensity with reported cor-
relations of ratings and heart rates of r>0.80 and test–retest
reliability of ICC=0.72.19,20
The TUG was used to measure functional mobility. The
participants sat on a standardized armchair. They were
instructed to rise, walk as quickly and safely as possible to a
marker on the floor 3m away, turn around, walk back to the
chair, and sit down on set cues. Timing started on the word
‘start’ and finished once the participant’s back touched the
back of the chair.10 Participants performed the test three
times; the fastest time measured in seconds with one decimal
place was used. The TUG has demonstrated test–retest reli-
ability of ICC=0.99 in children with CP.21 Despite some simi-
larity with 6MWT, the TUG has several tasks involved that
are potentially destabilizing, thus testing functional mobility
and anticipatory aspects of postural control.8
Pain intensity and impact on daily life were measured using
the ‘bodily pain’ domain of the General Health Survey Short-
Form 36 (SF-36), on a scale from 0 (worst pain) to 100 (no
pain).22 The SF-36 is validated in the Norwegian popula-
tion,23 and the ‘bodily pain’ domain has demonstrated test–
retest reliability of ICC=0.90.24
Fatigue was measured with the Fatigue Severity Scale, a
questionnaire using a scale from 1 to 7.25 The questionnaire is
validated in the Norwegian population, with average scores of
Walking Ability in Adults with CP Grethe Maanum et al. e127
 14698749, 2010, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2010.03614.x by Cochrane Chile, Wiley Online Library on [30/08/2023]. 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
at least 5 defined as severe fatigue,25 and has demonstrated dures are valuable tools when asymptotic properties of the
test–retest reliability of kappa=0.74 for the average score.26 estimators can be debated.36
The principal component analysis, ridge modelling, and the
Body structure ⁄ function bootstrap routines were performed in the R language, version
Foot deformity, defined as tibiocalcaneal angle being R.2.9.1 (http://www.r-project.org/). All other analyses were
greater than 5 varus or valgus, was determined by visual cate- performed with SPSS software (version 15.0; SSPS Inc., Chi-
gorization with the participants in standing position.27 This cago, IL, USA). Probability values less than 0.05 were consid-
assessment showed no disagreement with a secondary photo- ered statistically significant.
graph-based assessment for 66 of the participants.
Unilateral popliteal angle, expressing ‘functional’ hamstring RESULTS
length,7 was used as an indicator of range of motion in lower Of 201 respondents, 156 met the inclusion criteria and were
limbs (Appendix SII, published online). Intrarater reliability of enrolled for further testing. After clinical investigation another
ICC=0.77 is reported for children with CP.28 30 individuals were excluded, and 126 individuals were
The Modified Ashworth Scale, which measures the degree included in the study (Fig. S1, published online). Demo-
of resistance to fast passive movement, was used as an indica- graphic and clinical characteristics are presented in Table I.
tor of spasticity in lower limbs (Appendix SII).29,30 Interrater
reliability with a general kappa value of 0.74 to 0.82 is Walking ability
reported for comparable populations.30 Most participants (n=94) were classified at GMFCS level II.
Manual muscle testing as described by Hislop,31 which However, according to GMFCS level at adolescence, 32 had
scores muscle strength on a scale from 0 to 5, was used as an
indicator of muscle strength in lower limbs (Appendix SII). Table I: Demographic, social, and clinical data of participants (n=126)
This has demonstrated convergent construct validity with
dynamometry (r>0.76).32 Because manual muscle testing is n (%)a
prone to examiner bias, all the tests were performed by the
Age in years (y), mean (SD) 39 (12)
same assessor.31,32
Range, y 18–65
As impairment measures for the lower limbs, the popliteal Sex
angles, Modified Ashworth Scale, and manual muscle testing Female 73 (58)
Male 53 (42)
scores were individually transformed into summary measures
Type of cerebral palsyb
calculated as the mean of the measurements.33 Spastic unilateral 59 (47)
Spastic bilateral 67 (53)
GMFCS levelc
Statistical analysis
I 12 (10)
Descriptive statistics were used to characterize the sample. II 94 (74)
The TUG displayed a skewed distribution and was log trans- III 20 (16)
Education (duration)
formed. There were no missing data. Differences in TUG
First level (9y) 25 (20)
scores and 6MWD between the GMFCS levels were analysed Second level (12y) 49 (39)
with the Kruskal–Wallis test and the Mann–Whitney U-test. Third level (>12y) 52 (41)
Work ⁄ source of income
Bivariate relations between variables were explored in cross-
Paid work ‡20% 81 (64)
tables, box plots, and scatter plots, with corresponding Pear- Student 16 (13)
son’s and Spearman’s correlation coefficients. Disablement benefit ⁄ unemployedd 51 (40)
Comorbiditiese 61 (48)
Relevant literature was considered to identify possible pre-
Epilepsy 1 (1)
dictors for performance on the 6MWT, reflecting both Allergy 26 (21)
patient characteristics and the different ICF domains. The Anxiety ⁄ depression 16 (13)
Hypertension 11 (9)
regression model building included univariate regression anal-
Hypothyroidism 7 (6)
yses, analyses of several custom models considered to be of Otherf 4 (3)
importance, and stepwise procedures. Every step of the analy- Orthopaedic surgery lower limbg 62 (49)
Gastrocnemius ⁄ soleus lengthening 47 (37)
sis involved F-tests, Akaike’s Information Criterion and model
Hamstring lengthening 11 (9)
diagnostics, and discussions of the clinical importance of the Iliopsoas lengthening 11 (9)
findings. Interdependencies between predictors were explored Otherh 9 (7)
by principal component analysis34 and multicolinearity diag- a
Numbers presented unless otherwise stated. bBased on data from
nostics. The impact of this on the regression coefficients was medical records and clinical assessments. cClassified by first author.
explored by ridge regression.34 The final multiple model was d
For many with less than 100% paid work and some of the students
based on variables identified in these procedures, with redun- the income was combined with disablement benefit, so the
percentages of work ⁄ studies and disablement benefit add up to more
dant variables removed. Bootstrap methods with 10 000 repli- than 100. eObtained from medical records. fNamely diabetes type II,
cates were used to calculate the regression confidence intervals psoriasis, and hypercholesterolemia. gObtained from medical records.
(CIs), as the traditional formula-based CIs do not account for For several participants, this was more than one operation, so the
percentages of orthopaedic surgery of lower limbs add up to more
potential bias in the marginal distributions of the regression than 100. hRotational osteotomies and foot ⁄ toe corrections. GMFCS,
coefficients.35 Despite their shortcomings, bootstrap proce- Gross Motor Function Classification System.

e128 Developmental Medicine & Child Neurology 2010, 52: e126–e132

 14698749, 2010, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2010.03614.x by Cochrane Chile, Wiley Online Library on [30/08/2023]. 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: Present walking ability described by GMFCS level, FMS score, and questionnaire (n=126)

GMFCS levela, n (%)

Adolescence Present Shift to lower GMFCS levels

I 44 (35) 12 (10)
II 80 (64) 94 (74) 32 (25) from level I to II
III 2 (2) 20 (16) 18 (14) from level II to III

FMS, n (%)

Score Description 5m 50m 500m

6 Independent on all surfaces 14 (11) 5 (4) 1 (1)

5 Independent on level surfaces 103 (82) 101 (80) 87 (69)
4 Use one crutch or two sticks 8 (6) 12 (10) 25 (20)
3 Use two crutches 1 (1) 7 (6) 6 (5)
2 Use walking frame 0 (0) 1 (1) 5 (4)
1 Use wheelchair 0 (0) 0 (0) 2 (2)

Questionnaireb, n (%)

Is able to use public transport 98 (78)

Walking for exercise 83 (66)
Reported changes compared with adolescencec
Increased stiffness 100 (79)
Decreased balance 68 (54)
Decreased walking distance 76 (60)
Increased pain 81 (64)
Reported limitations in walking functiond
Fatigue 69 (55)
Stiffness 81 (64)
Balance 65 (52)
Pain 73 (58)

a
Classification of Gross Motor Function Classification System (GMFCS) level in adolescence based on interview data compared with present
classification based on observation. bQuestionnaire developed for this study. c,dThe percentages add up to more than 100 owing to changes or
limitations caused by more than one of the items. FMS, Functional Mobility Scale.

declined from GMFCS level I to level II (Table II). Examples
of statements from the interviews and questionnaires were,
‘Now I need a rail while climbing stairs’ and ‘I don’t do moun-
taineering anymore’.
TUG values and 6MWD varied significantly across the
GMFCS levels (p>0.001). The median scores (interquartile
range) on the TUG were as follows: GMFCS level I, 5.2s
(4.7–5.9); level II, 6.5s (5.9–8.4); and level III, 10.8s (9.9–13.8).
Median 6MWDs were as follows: GMFCS level I, 611m
(542–652); level II, 491m (435–543); and level III, 376m (317–
423).
FMS scores showed that the participants were more likely to
use assistive devices outdoors (Table II). Additional informa-
tion from the interviews showed a trend towards increasing fre-
quency in the use of assistive devices. Stiffness was the most
frequently reported cause of declined walking ability (Table II).
Predictors of 6MWT
During the model-building process, we noted that type of CP,
spasticity, muscle strength, popliteal angle, FMS, GMFCS,
and the TUG test were interrelated, with bivariate correlation
coefficients ranging from r=0.31 (0.14–0.46) to r=0.63 (0.51–
0.73; GMFCS ⁄ FMS), all factors having a correlation with
6MWD of r>0.38 (0.22–0.52). Both the GMFCS and FMS
were significantly associated with 6MWD in the univariate
analysis (r=0.58 [0.45–0.69] and r=0.56 [0.43–0.67]) respec-
tively. Because most participants were at GMFCS level II and
FMS 5, and because of colinearity in models with the TUG
and FMS, GMFCS and FMS were not included in the multi-
ple regression analysis. Descriptive statistics of the variables in
the final multiple model are presented in Tables I and III.
Univariate analyses (Table IV) showed that neuromuscular
and musculoskeletal impairments, such as increased spasticity,
reduced muscle strength, increased popliteal angle, equinova-
rus or -valgus foot, as well as increased pain, fatigue, and time
used in the TUG, were related to decreased 6MWD. Sex was
included in the multiple regression analyses as a known
predictor for 6MWD.
In the final multiple regression analysis, sex, type of CP,
popliteal angle, pain, and TUG values were identified as sig-
nificant predictors of 6MWD. The adjusted estimates and CIs
are shown in Table IV. Females had a shorter 6MWD than
males, as did bilateral CP compared with unilateral CP. An
increase of 1 in popliteal angle resulted in a 2.6m shorter
6MWD; transferring this result to clinical practice, a 10
increase in popliteal angle would have resulted in 26m shorter
6MWD. The regression coefficient for pain reflected a 1.3m
longer 6MWD, with a higher score on the 0 to 100 scale indi-
cating less pain. The regression coefficient for a log-trans-
formed TUG of )377 reflected a shorter 6MWD with
increasing TUG values (in seconds). Participants with a TUG
value at the lower quartile for the group (5.9s) walked on

Walking Ability in Adults with CP Grethe Maanum et al. e129

 14698749, 2010, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2010.03614.x by Cochrane Chile, Wiley Online Library on [30/08/2023]. 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
DISCUSSION
Table III: Variables used in the regression analysis (n=126)
This study investigated walking ability and the influence of
Variablea Mean (SD)b Range
demographic, clinical, and self-reported factors on functional
walking capacity in adults with CP who reported a declined
Dependent variable walking function compared with their function during adoles-
6-minute walk test distance (m) 485 (95) 260–730
Potential predictor
cence. We identified CP-related neuromuscular deficits, pain,
Borg Scalec (6–20) 13.5 (2.1) 9–20 and sex as important factors predicting functional walking
Foot deformityd, n (%) 75 (60) Yes ⁄ no capacity. Self-reported walking function in adolescence com-
Popliteal anglee (degrees) 51 (13) 25–90
Spasticityf 1.2 (0.8) 0.1–3.3
pared with current GMFCS classification indicated a shift in
Muscle strengthg 4.4 (0.4) 3.4–5.0 GMFCS level for 39% of the participants.
Timed Up and Goh (s), median (Q1–Q3) 6.8 (5.9–9.6) 3.6–24 As expected, type of CP was a significant predictor for
Log-transformed Timed Up and Go 0.86 (0.14) 0.56–1.38
Bodily paini (0–100) 54.8 (25.2) 0–100
6MWD, which implies that individuals with spastic bilateral
Fatigue severity scalej (0–7) 4.8 (1.4) 1.1–7.0 CP have more neurological impairments.1 Because CP is a
a
multifactorial disorder, and there is conflicting evidence on
Age, sex, and type of CP are described in Table I. bNumbers
how motor impairments relate to each other and to function
presented unless otherwise stated. cPerception of physical exertion,
higher scores indicating more exertion. dOccurrence of equinovarus or for children with CP,29,33 it seems logical that our results indi-
-valgus deformity. eUnilateral popliteal angle for the right and left side cate a complex interrelation between the variables investi-
calculated as the mean of the measurements, higher score indicating
gated.
reduced ‘functional hamstring length’. fModified Ashworth Scale (0–4)
for eight different muscle groups calculated as the mean of the Increased mean popliteal angle was associated with
measurements, higher score indicating more spasticity in lower limb.
g
decreased 6MWD. By choosing unilateral popliteal angle
Manual muscle testing (0–5) for 14 different muscle groups calculated
instead of bilateral popliteal angle,7 we measured the ‘func-
as the mean of the measurements, lower score indicating less muscle
strength in lower limbs. hHigher score indicating more functional tional hamstring length’ with the potential effect of a contra-
mobility impairment. iGeneral Health Survey Short-Form 36 (SF-36) lateral hip flexor tightness, a factor considered relevant when
domain of bodily pain, higher score indicating less pain and less effect
choosing a lower-limb range of motion measure related to
of pain on daily activities. jHigher score indicating more fatigue.
Q1–Q3, interquartile range. walking function. The effect of mean spasticity in the adjusted
analyses may be confounded by other effects, such as the mean
average 100m further than those at the upper quartile for the popliteal angle. This is supported by the association between
group (9.6s). Log-transformed TUG became the strongest reduced range of motion and increased spasticity ⁄ more
predictor, with a standardized regression coefficient of )0.57 impairment in spastic CP,7,29 the ‘increased stiffness’ as the
(Table IV). The coefficient of multiple determination (R2) most frequently reported cause of declined walking function,
showed that this model explained 67% of the variance of the and the fact that the modified Ashworth Scale does not dis-
6MWD. These results were similar to the results from ridge criminate between dynamic spasticity and spastic contrac-
regression models. tures.29

Table IV: Results from univariate and multiple linear regression analyses with 6-minute walk test distance as the dependent variable (n=126)

Unadjusted estimates Adjusted estimates

Bootstrap Bootstrap

Potential predictors Ba p Betab Ba 95% CI Ba p Betab Ba 95% CI

Sex )29 0.089 )0.15 )29 )63; )4 )29 0.008 )0.15 )29 )50; )9
Age )2.3 0.001 )0.31 )2.3 )3.6; )1.0
Type of cerebral palsy )98 <0.001 )0.52 )98 )126; )70 )31 0.007 )0.16 )31 )55; )9
Foot deformityc )43 0.011 0.22 )43 )9; )78
Popliteal angled (degrees) )2.6 <0.001 )0.39 )2.6 )3.6; )1.4 )1.5 0.001 )0.20 )1.5 )2.3; )0.6
Spasticitye )40 <0.001 )0.45 )40 )53; )28
Muscle strengthf 133 <0.001 0.51 133 96; 173
Borg Scaleg )12.8 0.001 )0.29 )12.8 )20; )6
Log-transformed TUGh )508 <0.001 )0.77 )508 )606; )424 )377 <0.001 )0.57 )377 )478; )291
Bodily paini 1.3 <0.001 0.34 1.3 0.7; 1.9 0.5 0.009 0.14 0.5 0.1; 1.0
Fatigue Severity Scalej )16.2 0.008 )0.24 )16.2 )28; )4.9

Eleven clinically relevant potential predictors were included. The adjusted estimates are those from the final model, using multiple linear
regression analysis (R2=0.67). aRegression coefficient. bStandardized regression coefficient. Coding of categorical variables: male=0, female=1;
spastic unilateral=0, spastic bilateral=1. cOccurrence of equinovarus or -valgus deformity classified as no=0, yes=1. dUnilateral popliteal angle for
the right and left side calculated as the mean of the measurements, higher score indicating reduced ‘functional hamstring length’. eModified
Ashworth Scale (0–4) for eight different muscle groups calculated as the mean of the measurements, higher score indicating more spasticity in
lower limb. fManual muscle testing (0–5) for 14 different muscle groups calculated as the mean of the measurements, lower score indicating less
muscle strength in lower limbs. gPerception of physical exertion, higher score indicating more exertion. hHigher score indicating more functional
mobility impairment. iGeneral Health Survey Short-Form 36 (SF-36) domain of bodily pain (0–100), higher score indicating less pain. jHigher score
indicating more fatigue. TUG, Timed Up and Go.

e130 Developmental Medicine & Child Neurology 2010, 52: e126–e132


When adjusted, mean lower-limb muscle strength was not a
significant predictor. This may reflect the narrow range of
muscle strength and the clinical impression that the partici-
pants had satisfactory muscle strength for walking.37 However,
we recognize that manual muscle testing has limited ability to
discriminate muscle strength scores.31,32 Thus, we hypothesize
that muscle strength is important for maintaining functional
mobility in more demanding mobility tasks than walking in a
corridor.8,38 Foot deformity was excluded in the adjusted anal-
ysis as a consequence of other variables being more important.
As the TUG is a measure of functional mobility, the
interrelations between the TUG, GMFCS, FMS, and
6MWD were expected. We considered it important to use
a measure with a potential ability to detect anticipatory
aspects of postural control, as well as sensory and percep-
tual abilities required in functional mobility,8 thus challeng-
ing the complex impairments mentioned in the definition
of CP.1 The literature is conflicting about the ability of the
TUG to differentiate between the different GMFCS lev-
els.21,39 We found a significant difference in TUG values
across these levels. Thus, we suggest that the TUG can
give useful information about functional mobility in daily
activities in comparable populations. Recognizing the
limitations of the TUG as a specific measure of balance, it
may be sufficient for assessing changes of functional bal-
ance in follow-up programmes. Although the 6MWT and
TUG are different measures, there is some overlap in tasks,
which may overestimate the effect of the TUG on
6MWD. Further, it is recognized that the performance of
the TUG is related to age, possibly because of its depen-
dence on muscle strength as well as anticipatory postural
control, both of which decrease with age.8 Thus the TUG
may confound the effects of age and muscle strength in
our final model.
Our finding that pain was a significant predictor of func-
tional walking capacity, confirms the earlier reported impact
of pain on declined ambulation.2,4 In contrast, fatigue was
adjusted out as consequence of other variables being more
important. However, it is reasonable to hypothesize a greater
impact from pain and fatigue if walking performance had been
investigated.4
Despite the normally distributed age range from 18 to
65 years, we did not find any effect of age. This was surpris-
ing,3 but may be due to the impact of the TUG or to other
CP-related impairments being more important predictors
than age. Moreover, we recognize the fundamental challenge
of age effects in a cross-sectional study.
As with samples of healthy populations, sex was a significant
predictor in our final model; height and body mass index are
likely to contribute to this finding.40 The Borg Scale was not a
significant predictor, which probably reflects that most of the
participants were more limited by neuromuscular impairment
than by cardiovascular factors. This has also been demon-
strated for stroke populations.14
The walking capacity in these high-level ambulant partici-
pants was better than in a population of adults with CP
recruited from a rehabilitation centre,19 but reduced compared
14698749, 2010, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2010.03614.x by Cochrane Chile, Wiley Online Library on [30/08/2023]. 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
with the general population.40 The results show that even per-
sons at GMFCS level I experience declined walking ability. As
others have demonstrated stability until young adulthood,6
our results demonstrate the need for studies on adults. In addi-
tion, there are few studies exploring different mobility aspects
that, potentially, protect from or predispose to declined walk-
ing in individuals at GMFCS levels I and II. Thus, our results
support the need for a measure differentiating the children at
GMFCS level I.41
We recognize that a limitation of this study is the selected
group of high-level independent walkers. The recruitment
through advertisements also led to more female participants
than males. Though widespread in clinical use, the lower limb
assessments have limitations to their reliability.28,29,32 To
compensate for this we used a strict protocol, and one experi-
enced clinician conducted all the assessments. The use of
mean scores may have limited the measurements for popliteal
angle, spasticity, and muscle strength. However, the purpose
was to have clinically relevant scores in spastic lower limbs,
and therefore sum scores were judged to be sufficiently pre-
cise.33
Although we have attempted to explore a wide range of
different subsets of predictor variables to support our final
model, subset selection procedures tend to be slightly over-
fit, and the effect size estimates may, therefore, be biased.42
Our bootstrap estimation ensured that possible biases in
the distribution of the estimators were accounted for in the
calculations of the CIs. However, a potential slight overes-
timation of effect sizes remains, with other subsets of pre-
dictor variables potentially being equally good or superior
to those we have presented.34,42,43 Therefore, our model
requires further validation in studies on similar samples.
The CIs in the final model may reflect both measurement
errors and the known heterogeneity in spastic CP.7 The
results support the need for more studies with adequate
sample sizes addressing subgroups, as well as studies with a
longitudinal design exploring the effects of ageing on walk-
ing ability.
CONCLUSION
The present study demonstrates that in this population of
independent adult walkers reporting declined walking ability
compared with adolescence, 39% had changed GMFCS level.
The multiple regression analyses identified sex, type of CP,
popliteal angle, pain, and performance on the TUG as signifi-
cant predictors, with the TUG as the strongest predictor. This
implicates the importance of focusing specifically on different
aspects of functional mobility in further studies with an ade-
quate sample size and a longitudinal design.
ACKNOWLEDGEMENTS
This study was performed as part of a randomized controlled trial,
and the project was supported by the East Regional Health Adminis-
tration and Sunnaas Rehabilitation Hospital (grant number 206
24 503). We are grateful to all the adults with CP who participated.
We acknowledge Professor Katharina S Sunnerhagen, and the staff at
Sunnaas Gait laboratory.
Walking Ability in Adults with CP Grethe Maanum et al. e131

SUPPORTING INFORMATION
Additional supporting information may be found in the online version
of this article:
Figure S1: Summary of recruitment of adults with CP responding to
the advertisement.
Appendix SI: Questionnaire concerning present walking ability.
REFERENCES
1. Rosenbaum P, Paneth N, Leviton A, Goldstein M, Bax M.
The definition and classification of cerebral palsy. Dev Med
Child Neurol 2007; 49(Suppl 109): 1–44.
2. Andersson C, Mattsson E. Adults with cerebral palsy: a sur-
vey describing problems, needs, and resources, with special
emphasis on locomotion. Dev Med Child Neurol 2001; 43:
76–82.
3. Jahnsen R, Villien L, Egeland T, Stanghelle JK, Holm I.
Locomotion skills in adults with cerebral palsy. Clin Rehabil
2004; 18: 309–16.
4. Opheim A, Jahnsen R, Olsson E, Stanghelle JK. Walking
function, pain, and fatigue in adults with cerebral palsy: a
7-year follow-up study. Dev Med Child Neurol 2009; 51:
381–8.
5. Palisano R, Rosenbaum P, Walter S, Russell D, Wood E,
Galuppi B. Development and reliability of a system to clas-
sify gross motor function in children with cerebral palsy. Dev
Med Child Neurol 1997; 39: 214–23.
6. Hanna SE, Rosenbaum PL, Bartlett DJ, et al. Stability and
decline in gross motor function among children and youth
with cerebral palsy aged 2 to 21 years. Dev Med Child Neurol
2009; 51: 295–302.
7. Gage JR, Schwartz MH, Koop SE, Novacheck TF. The
Identification and Treatment of Gait Problems in Cerebral
Palsy, 2nd edn. Clinics in Developmental Medicine Nos
180–181. London: Mac Keith Press, 2009.
8. Shumway-Cook A, Wollacott MH. Motor Control: Trans-
lating Research into Clinical Practice. Philadelphia: Lippin-
cott Williams & Wilkins, 2007.
9. American Thoracic Society. ATS statement: guidelines for
the six-minute walk test. Am J Respir Crit Care Med 2002;
166: 111–7.
10. Podsiadlo D, Richardson S. The timed ‘‘Up & Go’’: a test of
basic functional mobility for frail elderly persons. J Am Geri-
atr Soc 1991; 39: 142–8.
11. World Health Organization. International Classification of
Functioning, Disability and Health. Geneva: World Health
Organization, 2001.
12. Guyatt GH, Sullivan MJ, Thompson PJ, et al. The 6-minute
walk: a new measure of exercise capacity in patients with
chronic heart failure. Can Med Assoc J 1985; 132: 919–23
13. Lord SR, Menz HB. Physiologic, psychologic, and health
predictors of 6-minute walk performance in older people.
Arch Phys Med Rehabil 2002; 83: 907–11.
14. Pang MY, Eng JJ, Dawson AS. Relationship between ambu-
latory capacity and cardiorespiratory fitness in chronic
stroke: influence of stroke-specific impairments. Chest 2005;
127: 495–501.
e132 Developmental Medicine & Child Neurology 2010, 52: e126–e132
Appendix SII: Clinical measures.
Please note: Wiley-Blackwell are not responsible for the content or
functionality of any supporting materials supplied by the authors. Any
queries (other than missing material) should be directed to the corre-
sponding author of the article.
15. Mathias S, Nayak US, Isaacs B. Balance in elderly patients:
the ‘‘get-up and go’’ test. Arch Phys Med Rehabil 1986; 67:
387–9.
16. Cans C. Surveillance of cerebral palsy in Europe: a collabora-
tion of cerebral palsy surveys and registers. Dev Med Child
Neurol 2000; 42: 816–24.
17. Jahnsen R, Aamodt G, Rosenbaum P. Gross Motor Function
Classification System used in adults with cerebral palsy:
agreement of self-reported versus professional rating. Dev
Med Child Neurol 2006; 48: 734–8.
18. Graham HK, Harvey A, Rodda J, Nattrass GR, Pirpiris M.
The Functional Mobility Scale (FMS). J Pediatr Orthop
2004; 24: 514–20.
19. Andersson C, Asztalos L, Mattsson E. Six-minute walk test
in adults with cerebral palsy. A study of reliability. Clin Reha-
bil 2006; 20: 488–95.
20. Borg GA. Psychophysical bases of perceived exertion. Med
Sci Sports Exerc 1982; 14: 377–81.
21. Gan SM, Tung LC, Tang YH, Wang CH. Psychometric
properties of functional balance assessment in children with
cerebral palsy. Neurorehabil Neural Repair 2008; 22: 745–53.
22. Ware J. SF-36 Health Survey Manual and Interpretation
Guide, 2nd edn. Boston: The Health Institute, New England
Medical Centre, 1997.
23. Loge JH, Kaasa S, Hjermstad MJ, Kvien TK. Translation
and performance of the Norwegian SF-36 Health Survey in
patients with rheumatoid arthritis. I. Data quality, scaling
assumptions, reliability, and construct validity. J Clin Epi-
demiol 1998; 51: 1069–76.
24. Linde L, Sorensen J, Ostergaard M, Horslev-Petersen K,
Hetland ML. Health-related quality of life: validity, reliabil-
ity, and responsiveness of SF-36, EQ-15D, EQ-5D, RAQoL,
and HAQ in patients with rheumatoid arthritis. J Rheumatol
2008; 35: 1528–37.
25. Lerdal A, Wahl A, Rustoen T, Hanestad BR, Moum T. Fati-
gue in the general population: a translation and test of the
psychometric properties of the Norwegian version of the fati-
gue severity scale. Scand J Public Health 2005; 33: 123–30.
26. Mattsson M, Moller B, Lundberg I, Gard G, Bostrom C.
Reliability and validity of the Fatigue Severity Scale in Swed-
ish for patients with systemic lupus erythematosus. Scand J
Rheumatol 2008; 37: 269–77.
27. Chang CH, Miller F, Schuyler J. Dynamic pedobarograph in
evaluation of varus and valgus foot deformities. J Pediatr Ort-
hop 2002; 22: 813–8.
28. Ten Berge SR, Halbertsma JP, Maathuis PG, Verheij NP,
Dijkstra PU, Maathuis KG. Reliability of popliteal angle
measurement: a study in cerebral palsy patients and healthy
controls. J Pediatr Orthop 2007; 27: 648–52.
14698749, 2010, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1469-8749.2010.03614.x by Cochrane Chile, Wiley Online Library on [30/08/2023]. 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
29. Ostensjo S, Carlberg EB, Vollestad NK. Motor impairments
in young children with cerebral palsy: relationship to gross
motor function and everyday activities. Dev Med Child Neurol
2004; 46: 580–9.
30. Ghotbi N, Ansari NN, Naghdi S, Hasson S, Jamshidpour B,
Amiri S. Inter-rater reliability of the Modified Ashworth
Scale in assessing lower limb muscle spasticity. Brain Inj
2009; 23: 815–9.
31. Hislop HJ. Daniels and Worthingham’s Muscle Testing:
Techniques of Manual Examination. St Louis, MO: Saun-
ders ⁄ Elsevier, 2007.
32. Bohannon RW. Measuring knee extensor muscle strength.
Am J Phys Med Rehabil 2001; 80: 13–8.
33. Ross SA, Engsberg JR. Relationships between spasticity,
strength, gait, and the GMFM-66 in persons with spastic
diplegia cerebral palsy. Arch Phys Med Rehabil 2007; 88:
1114–20.
34. Hastie TJ. The Elements of Statistical Learning: Data Min-
ing, Inference, and Prediction. New York: Springer, 2001.
35. Efron B, Tibshirani R. Bootstrap methods for standard
errors, confidence intervals, and other measures of statistical
accuracy. Stat Sci 1986; 1: 54–77.
36. Young GA, Daniels HE. Bootstrap bias. Biometrika 1990; 1:
179–85.
37. Abel MF, Damiano DL, Blanco JS, et al. Relationships
among musculoskeletal impairments and functional health
status in ambulatory cerebral palsy. J Pediatr Orthop 2003;
23: 535–41.
38. Shortland A. Muscle deficits in cerebral palsy and early loss
of mobility: can we learn something from our elders? Dev
Med Child Neurol 2009; 51(Suppl 4): 59–63.
39. Williams EN, Carroll SG, Reddihough DS, Phillips BA,
Galea MP. Investigation of the timed ‘up & go’ test in chil-
dren. Dev Med Child Neurol 2005; 47: 518–24.
40. Jenkins S, Cecins N, Camarri B, Williams C, Thompson P,
Eastwood P. Regression equations to predict 6-minute walk
distance in middle-aged and elderly adults. Physiother Theory
Pract 2009; 25: 516–22.
41. Whright V, Kavanaugh A, Moher R, et al. Measuring higher
level functional mobility skills in children with cerebral palsy
(CP) in GMFCS Level I: preliminary work on a proposed
challenge module for the gross motor function measure. Dev
Med Child Neurol 2009; 51(Suppl 2): 61–2.
42. Copas JB. Regression, prediction and shrinkage. J R Stat Soc
A 1983; 145: 311–54.
43. Miller AJ. Selection of subsets of regression variables.
J R Stat Soc A 1984; 147: 389–425.