DEVELOPMENTAL MEDICINE & CHILD NEUROLOGY

ORIGINAL ARTICLE

Age-related changes in energy efficiency of gait, activity, and participation in children with cerebral palsy
CLAIRE KERR 1 | BRONA C MCDOWELL 2 | JACKIE PARKES 1 | MIKE STEVENSON 3 | AIDAN P COSGROVE 2
1 School of Nursing and Midwifery, Queen's University Belfast, Belfast, Northern Ireland. 2 Gait Analysis Service, Belfast Health and Social Care Trust, Musgrave Park Hospital, Stockman's Lane, Belfast, Northern Ireland. 3 Department of Epidemiology and Public Health, Queen's University Belfast, Belfast, Northern Ireland.
Correspondence to Dr Claire Kerr at Nursing and Midwifery Research Unit, Queen's University Belfast, 10 Malone Road, Belfast BT9 5BN, Northern Ireland. E-mail: c.kerr@qub.ac.uk

PUBLICATION DATA

AIM The aim of this study was to use a prospective longitudinal study to describe age-related
trends in energy efciency during gait, activity, and participation in ambulatory children with cerebral palsy (CP). METHOD Gross Motor Function Measure (GMFM), Paediatric Evaluation of Disability Inventory (PEDI), and Lifestyle Assessment Questionnaire-Cerebral Palsy (LAQ-CP) scores, and energy efciency (oxygen cost) during gait were assessed in representative sample of 184 children (112 male; 72 female; mean age 10y 9mo; range 416y) with CP. Ninety-four children had unilateral spastic CP, 84 bilateral spastic CP, and six had other forms of CP. Fifty-seven were classied as Gross Motor Function Classication System (GMFCS) level I, 91 as level II, 22 as level III, and 14 as level IV). Assessments were carried out on two occasions (visit 1 and visit 2) separated by an interval of 2 years and 7 months. A total of 157 participants returned for reassessment. RESULTS Signicant improvements in mean raw scores for GMFM, PEDI, and LAQ-CP were recorded; however, mean raw oxygen cost deteriorated over time. Age-related trends revealed gait to be most inefcient at the age of 12 years, but GMFM scores continued to improve until the age of 13 years, and two PEDI subscales to age 14 years, before deteriorating (p<0.05). Baseline score was consistently the single greatest predictor of visit 2 score. Substantial agreement in GMFCS ratings over time was achieved (jlw=0.740.76). INTERPRETATION These ndings have implications in terms of optimal provision and delivery of services for young people with CP to maximize physical capabilities and maintain functional skills into adulthood.

Accepted for publication 9th July 2010. Published online 28 September 2010.
ABBREVIATIONS

GMFM LAQ-CP

Gross Motor Function Measure Lifestyle Assessment Questionnaire-Cerebral Palsy PEDI Paediatric Evaluation of Disability Inventory PEDI CAS PEDI caregiver assistance scaled scores

Cerebral palsy (CP) is the leading cause of motor impairment in childhood, caused by damage to, or malformation of, the developing brain.1 Although the brain lesion is static, its consequences may change as the child develops, necessitating a lifelong demand on therapy services in some cases. Knowledge of the natural progression of the condition may help families, service planners, and researchers to anticipate the likely future needs of this population. The most reliable data about the numbers of people affected by CP in developed countries come from case registers, the aims of which are to systematically compile data that can be used for surveillance and service planning and to provide a framework for research.2 The data collected are primarily of an epidemiological nature and provide a useful starting point for the development of more precise clinical research. Several large retrospective population-based studies have contributed to the knowledge-base of the long-term prole of people with CP. For example, Wu et al.3 investigated prognostic factors for ambulation and Day et al.4 studied changes in ambulatory status through adolescence and young
The Authors. Journal compilation Mac Keith Press 2010

adulthood. The prognostic charts resulting from these two studies provide useful age estimates for the achievement and retention of ambulatory skills. More recently, several prospective longitudinal studies have been reported. The Ontario Motor Growth Study5 mapped motor development in children with CP using the using the Gross Motor Function Measure (GMFM)6,7 and the Gross Motor Function Classication System (GMFCS).8 The resultant prognostic curves have been widely utilized by clinicians and researchers. A continuation of this study is currently in progress.9 Another study employed cross-sectional and longitudinal analyses to construct population-based gross motor developmental curves for children aged up to 15 years with hemiplegia and diplegia.10 Furthermore, prospective longitudinal work from the Shriners Group of Hospitals11 and from the Paediatric Rehabilitation in the Netherlands programme (PERRIN)12,13 is contributing to a more comprehensive picture of the young person with CP. The aims of this paper are to describe the natural progression of the ambulatory form of CP over time and to identify
DOI: 10.1111/j.1469-8749.2010.03795.x 61

predictors of change. A comprehensive assessment framework was used, of which selected measures of energy efciency during gait, activity, and participation are presented in this article.

What this paper adds

Longitudinal population-based trends have been found in energy efficiency, activity, and participation in children with ambulatory CP. Evidence supports the fact that there is a decline in motor function from puberty onwards. Our results indicate that participation is stable throughout childhood and the adolescent years. This study confirms that GMFCS levels in children with CP remain stable over time.

METHOD Ethical approval Approval for the study was granted from the local research ethics committee, and all parents (and children when possible) provided written informed consent. Study design This was a 5-year prospective longitudinal study, which used a population-based approach to the identication, approach, and recruitment of a representative sample of children with ambulatory CP. Participants All children with a diagnosis of early impairment CP who could ambulate more than 10m independently (assistive devices permitted), who were born between 1 June 1987 and 1 June 1999, and who were alive and resident in Northern Ireland on 1 June 2003 were potentially eligible to participate. Information on the presumed timing of CP was available from the Northern Ireland Cerebral Palsy Register.14 Children who had undergone lower limb surgery within the previous year or had received a botulinum toxin A injection to the lower limbs within the previous 6 months were excluded; however, appointments were scheduled over the data collection period in order to minimize exclusions for these reasons. The Northern Ireland Cerebral Palsy Register identied 487 potentially eligible participants, of whom a representative sample of 184 (38%) were recruited to the study. Families were contacted either by their childs orthopaedic surgeon (n=312) or by a community paediatric professional (n=175). More families responded to an invitation from the orthopaedic service than to invitations from the community professionals, but overall no evidence existed of any systematic biases in demographic or key clinical characteristics between those children who participated and those who did not. Full details of the recruitment process and associated analyses are provided elsewhere.15 Measurement tools Energy efciency during gait was assessed using an oxygen consumption protocol and was recorded by a portable Cosmed K4b2 device (Cosmed, Rome, Italy). Participants had not eaten for 2 hours before the test and wore their normal footwear, orthotic devices, and or assistive devices (if any). The test involved a 5-minute walking protocol, described comprehensively elsewhere.16 Resting oxygen rate during the nal 2 minutes of an initial rest period, walking oxygen rate during the nal minute of the walk period, and mean walking speed during the nal minute of the walk period were calculated. Gross motor function was evaluated using the GMFM.7 Children were assessed barefoot using the original 88-item instrument. Children were also classied using the GMFCS.8 Participants aged over 12 years were classied using the
62 Developmental Medicine & Child Neurology 2011, 53: 6167

GMFCS 6 to 12 years descriptors as the criteria for older children were not available at the time of assessment. Functional capabilities and the amount of caregiver assistance required in the areas of self-care, mobility, and social function were recorded using the Paediatric Evaluation of Disability Inventory (PEDI).17 The PEDI is primarily designed for the functional evaluation of children aged from 6 months to 7 years 6 months, but can be used with older children if their functional abilities are less than those expected of a typically developing child aged 7 years and 6 months.17 A research assistant completed the PEDI by interviewing the childs parent carer. The impact of the childs disability on both the child and family unit was assessed using the Lifestyle Assessment Questionnaire-Cerebral Palsy (LAQ-CP).18 This questionnaire addresses physical independence, clinical burden, mobility, economic burden, social integration, and schooling, and was completed by the childs parent carer.

Definitions CP and clinical subtypes were dened according to the Surveillance of Cerebral Palsy in Europe guidelines.1 Early impairment CP was dened as motor impairment secondary to damage to the developing brain before the end of the neonatal period (i.e. within the rst 28d after birth). Intellectual impairment was dened as moderate where the IQ (or best clinical estimate) was 50 to 70 and severe where the IQ was less than 50. Deprivation was dened using the Carstairs Index.19 Children were assigned to a deprivation quintile ranging from quintile 1 (least deprived) to quintile 5 (most deprived) based on their electoral ward of residence. Perceptual problems, rated by parents, related to depth and spatial perception only. For clarity of description the following terms are used in this paper: energy efciency refers to the net oxygen cost of walking; activity refers to GMFM and PEDI scores, with GMFM reecting capacity (what the child can do in a standardized environment) and PEDI reecting performance (what the child actually does in their own environment); and participation refers to the LAQ-CP score. It is of note that the LAQ-CP is not a pure measure of participation, but was the best available tool at commencement of this study. Procedure Participants attended a regional hospital-based gait analysis laboratory on two occasions approximately 2 years 7 months apart. The same battery of physical and psychosocial tests were completed on each visit and were administered by the

same research physiotherapist research assistant. Assessments lasted from 2.5 to 3.5 hours. During an initial interview, caregivers subjectively rated their childs vision, hearing, perception, communication, and concentration as none, mild, moderate, or severe in addition to discussing aspects of the childs health care. Participants then underwent the oxygen consumption protocol while their caregiver completed the LAQ-CP. Afterwards, a research physiotherapist carried out the GMFM assessment while a research assistant interviewed the childs caregiver with the PEDI. GMFCS ratings were independently obtained at the time of assessment from the childs caregiver and the research physiotherapist. A third GMFCS rating was obtained, by mail, from the childs community physiotherapist. Epidemiological data relating to the childs IQ and seizure activity were obtained from the Northern Ireland Cerebral Palsy Register.

In all analyses, two-tailed signicance tests were employed, with the threshold of signicance being 5%. Furthermore, missing data were not replaced; casewise deletion was simply applied.

Data analysis Data were initially reviewed descriptively and then analysed for differences between assessment points. Trends associated with ageing were then identied and explored in the context of preselected clinical variables. Data were analysed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA) and Stata version 9.1 (StataCorp LP, TX, USA). Continuous data were summarized using means and SDs. Categorical variables were summarized using frequency tables. The net oxygen cost was calculated as follows: Net O2 cost=(walking O2 rate)resting O2 rate) average walking speed. Paired t-tests were used to establish if signicant differences existed between visit 1 and visit 2 data on each of the measurement tools. Difference scores were then calculated for all the measurement tools (visit 2 scorevisit 1 score) and their distribution was checked by observation of PP plots. Difference scores were subjected to simple linear regression (by least squares) to determine if signicant age-related trends were evident. If signicant age-related trends were identied, multiple linear regression was then performed to determine which preselected clinical variables contributed signicantly to the model. In each instance the visit 2 score was the dependent variable, with baseline score and age assigned as covariates. The role of the following preselected clinical variables was investigated: sex, CP type, incidence of orthopaedic surgery between assessment points, GMFCS level, IQ, incidence of seizures, and impairments of perception and concentration. For the purpose of the analyses, all xed factors were dichotomized with the exception of the GMFCS, in which the following distinctions were made: (i) level I, (ii) level II, and (iii) levels III and IV. This analysis was checked by the bootstrap method (using 1000 iterations) to estimate condence intervals (CIs). Stability of the GMFCS over time was determined by compiling cross-tabulations and then using the kappa statistic with linear weighting (jlw). The strength of agreement was dened as moderate, substantial, and almost perfect (jlw=0.410.60, 0.610.80, and 0.811.0 respectively).20 Finally, bias analysis was undertaken to test for heterogeneity in proportions (v2 test) in characteristics of children who attended visit 1 but did not attend visit 2.

RESULTS Demographics and clinical characteristics Of the initial 184 children assessed, 94 had unilateral spastic CP (51%), 84 (46%) had bilateral spastic CP, and six children had non-spastic forms (four had ataxia, one had dyskinesia, and one had hypotonia). A total of 151 children returned for the second assessment, with the parents of a further six children providing questionnaire data by post, giving a second response rate of 86% (n=157). Twenty-four children underwent surgery during the interim period. Subsequently, one child was excluded from all analyses owing to a change in diagnosis between the assessments. There were no statistically signicant differences between those who re-attended (n=157) and those who did not (n=26) in terms of age, sex, deprivation quintile, GMFCS level, or associated impairments. The mean time between assessments was 2 years 7 months (SD 2mo). The GMFCS distribution of participants and clinical characteristics are detailed in Table I. Data distribution Although ceiling effects in raw scores on the PEDI and, to a lesser extent, the GMFM were observed, the difference scores for these measures followed a normal distribution as evidenced by visual inspection of the PP plots. Changes over time Statistically signicant differences in raw scores on all of the measurement tools were noted between the rst and second assessments. Increases in mean GMFM and PEDI scores were recorded, indicating improvements in abilities (p=0.004 for GMFM; p0.001 for all PEDI subscales). Similarly, improvement in participation over time was reected by a decrease in

Table I: Demographic data

Visit 1, mean (SD) n Sex (M:F) Age (y) Age range (y) Height (m) Weight (kg) GMFCS Level I Level II Level III Level IV Missing SCPE classication Unilateral spastic Bilateral spastic Other 184 112:72 10.8 (3.6) 4.617.5 1.4 (0.2) 38.8 (16.9) 57 (31%) 91 (49.5%) 22 (12%) 14 (7.5%) Visit 2, mean (SD) 157 (86%) 99:58 13.3 (3.5) 7.119.9 1.5 (0.2) 48.1 (17.3) 55 (35%) 70 (44.5%) 16 (10%) 10 (6.5%) 6 (4%) 82 70 5

94 84 6

M, male; F, female; GMFCS, Gross Motor Function Classication System; SCPE, Surveillance Of Cerebral Palsy in Europe.

Energy Efficiency, Activity, and Participation in CP Claire Kerr et al. 63

the mean LAQ-CP score (p<0.001). In contrast, an increased cost of walking was noted by the increase in the mean net oxygen cost difference score over time (p=0.017). Mean scores at each assessment and difference scores on all measurement tools for those children with data available at each time point are presented in Table II.

Table III: Regression analyses results: measurement tool difference score as dependent variable, age at first assessment as independent variable. Statistically significant models are shown in bold
Measurement tool r Model Quadratic Linear F(df,n) p-value

Age-related trends Regression analyses using the difference scores for each measurement tool and the age at rst assessment revealed that net oxygen cost, GMFM, and the PEDI caregiver assistance scale (PEDI CAS) scores for mobility and social function altered signicantly over time (see Table III), although in all instances very weak relationships were identied. The relationship between the net oxygen cost and age was best described by a quadratic model (r=0.079; F2,82=3.45; p=0.035), with the turning point of the curve occurring at approximately 12 years of age, suggesting that gait is least efcient at this age. A linear relationship was identied between GMFM-66 and age (r=0.146; F1,119=20.37; p<0.001), with increasing age being associated with decreasing motor abilities. Figure S1 (online supplementary material) shows that the rate of development of motor skills is greatest in younger children, but this rate slows linearly until a position of no change is achieved at approximately 13 years of age (95% CI 11.916.3y). Beyond age 13, deterioration in motor abilities was noted. A similar deteriorating linear trend was identied for PEDI CAS mobility (r=0.071; F1,140=10.67; p=0.001), with increased dependence on assistance from caregivers noted from the age of 14 years. Finally, a very weak deteriorating logarithmic relationship was identied between PEDI CAS social function and age (r=0.028; F1,140=4.86; p=0.029). Other contributing variables Multiple linear regression was carried out for those measures that demonstrated statistically signicant age-related trends. Results were checked by the bootstrap method. The results are summarized in Table SI (online supplementary material).

Net oxygen cost of walking 0.079 GMFM-66 0.146 PEDI Functional scaled score Self-care 0.004 Mobility 0.018 Social function 0.013 Caregiver assistance scaled score Self-care 0.017 Mobility 0.071 Social function 0.028 LAQ-CP 0.034

0.04 3.50(2,82) 20.37(1,119) <0.001

Linear Linear Linear

0.53(1,140) 2.55(1,140) 1.86(1,140)

0.47 0.11 0.18 0.12 0.001 0.03 0.08

Linear 2.42(1,140) Linear 10.67(1,140) Logarithmic 4.86(1,140) Quadratic 2.57(2,145)

GMFM-66, gross motor function measure-66; PEDI, paediatric evaluation of disability inventory; LAQ-CP, lifestyle assessment questionnaire-cerebral palsy.

The visit 2 net oxygen cost score was signicantly dependent on baseline net oxygen cost score, surgery, IQ, concentration, and GMFCS level. This was conrmed by the bootstrap method, with the exception of the role of GMFCS (p=0.051). As demonstrated in Table SI, baseline oxygen cost was the greatest predictor of the visit 2 score (p<0.001). Better visit 2 net oxygen cost scores (i.e. lower scores) were also associated with an IQ over 70 (p=0.005), having surgery between the study assessment points (p=0.012), poorer concentration (p=0.005), and being classied as GMFCS level I compared with GMFCS levels III and IV (p=0.030). Baseline score was also the most inuential factor on the second GMFM score (p<0.001), with GMFCS level, age, and perception also signicantly contributing. However, results from the bootstrap method differed slightly in that the GMFCS level was not considered to be a signicant variable (p=0.0570.087) and that seizure activity was a more likely contributor (p=0.033). If applying the regression model, those children who could ambulate unaided were likely to have

Table II: Mean (SD) scores on measurement tools for all participants (data available at both assessment points)
Visit 1 2003 05, mean (SD) Net oxygen cost of walking (n=85) GMFM-66 (n=121) PEDI (n=142) Functional scaled score Self-care Mobility Social function Caregiver assistance scaled score Self-care Mobility Social function LAQ-CP (n=148) 0.188 (0.105) 78.43 (14.18) Visit 2 2006 07, mean (SD) 0.204 (0.110) 80.24 (14.35) Mean difference score (SD) visit 2visit 1 0.017 (0.049) 1.81 (6.73) 95% CI of the difference 0.0050.046 0.603.02 Clinical implication Deterioration Improvement

74.70 (14.20) 78.26 (15.59) 79.56 (15.56) 75.79 (17.01) 82.62 (15.48) 90.44 (15.28) 32.87 (14.99)

81.36 (15.41) 84.26 (15.25) 84.53 (15.78) 83.10 (15.92) 87.68 (13.91) 93.60 (14.38) 28.48 (15.00)

6.66 (9.71) 6.00 (9.95) 4.96 (11.61) 7.31 (11.17) 5.05 (10.75) 3.17 (10.99) )4.39 (9.51)

5.058.27 4.357.65 3.046.89 5.459.16 3.276.84 1.344.99 )5.93 to )2.84

Improvement Improvement Improvement Improvement Improvement Improvement Improvement

GMFM-66, Gross Motor Function Measure-66; PEDI, Paediatric Evaluation of Disability Inventory; LAQ-CP, Lifestyle Assessment QuestionnaireCerebral Palsy.

64 Developmental Medicine & Child Neurology 2011, 53: 6167

better second visit GMFM scores than those who required walking aids (GMFCS level I compared with levels III and IV; p=0.029). GMFM scores tended to deteriorate less quickly in younger children (p=0.001) and in children who had a moderate to severe perceptual problem (p=0.012). Full agreement between the regression and bootstrap methods was evident in the analyses of both PEDI subscales. Only baseline score and GMFCS level were shown to contribute signicantly to the PEDI CAS mobility visit 2 score, with lower GMFCS levels resulting in poorer PEDI CAS mobility scores (GMFCS level I compared with levels III and IV; p=0.034). Factors that signicantly affected the visit 2 PEDI CAS social function score were baseline score, incidence of orthopaedic surgery between assessments, IQ, and concentration. Unsurprisingly, baseline score was again the greatest predictor (p<0.001). Improved visit 2 PEDI CAS social function scores were associated uniquely with an IQ over 70 (p=0.002), no mild concentration problems (p=0.014), and having orthopaedic surgery between the assessments (p=0.006).

Stability of the GMFCS over time Substantial agreement in GMFCS level was noted over the 2 years 7 month assessment period, as indicated by jlw values of 0.74 to 0.76 and detailed in Table IV. This stability was noted with all three assessors: parents, clinical physiotherapist, and research physiotherapist. DISCUSSION This study presents age-related trends in oxygen cost, activity, and participation in a representative sample of children with CP. Signicant improvements in raw scores for activity (GMFM and PEDI) and participation (LAQ-CP) were noted in the sample over two time points, 2 years 7 months apart. However, this trend was not mirrored by the participants energy expenditure during gait, as a statistically signicant deterioration in the mean raw net oxygen cost was noted over time. Interestingly, very weak relationships with age were detected, with relationships between age and net oxygen cost, GMFM, PEDI CAS mobility, and PEDI CAS social function scores achieving statistical signicance. The role of multiple covariates was investigated for those measures demonstrating age-related changes in an attempt to quantify predictor variables. Of the proposed predictors, GMFCS level, IQ, perception, concentration, incidence of surgery between visits, age, presence of seizures, and baseline score in the measurement tool under investigation were shown to be inuential. Further-

Table IV: Stability of Gross Motor Function Classification System ratings over two assessments, 2 years and 6 months apart
% Absolute agreement 80 76 79

Rater Parent Clinical physiotherapist Research physiotherapist

n 150 92 151

jlw 0.74 0.74 0.76

95% CI 0.650.83 0.630.84 0.680.84

jlw, linear weighted kappa; CI, condence interval.

more, considerable stability in parent and therapist rating of GMFCS level over the assessment period was also demonstrated. The quadratic relationship associated with the age-related trend in energy efciency suggests that gait is most inefcient at around 12 years of age. Whilst perhaps contrary to clinical intuition, it may be that walking efciency deteriorates with the onset of puberty and changing demands of the childs education. The incidence of surgery between visits, IQ, and concentration were noted to inuence this model; however, baseline oxygen cost was identied as the greatest predictor of the visit 2 score, accounting for 69% of its variance. Contrary to these ndings, Day et al.4 reported an overall picture of stability in ambulation during adolescence and young adulthood, although other authors have documented a decline in ambulatory ability in adults with CP.2123 When selecting energy efciency variables, we elected to continue using net oxygen cost as our main analysis variable instead of the more recently advocated net non-dimensional oxygen cost.24 The reasons for this were twofold: rstly, to maintain consistency with reporting over time16 and, secondly, rendering the net oxygen cost data dimensionless simply required multiplication by a constant (20.1 g) and thus would not have any effect on the statistical analysis. More detailed investigation of the longitudinal prole of energy efciency of walking in this sample is warranted but is beyond the scope of this paper. A weak linear trend reecting a decrease in activity (GMFM) with increasing age was observed, with children classied as GMFCS level I tending to deteriorate less rapidly than those classied as level III or IV. This concurs in part with the ndings of Voorman et al.12 and Hanna and colleagues.9 Van Eck et al.13 reported declines in motor performance for children classied as GMFCS level II, IV, or V, but speculated that this anomaly in the hypothesized trend (decreasing function with increasing GMFCS level) may be due to intellectual or ne motor abilities. Somewhat counterintuitively, the current study also revealed that loss of motor function was slower among children with moderate to severe perceptual problems (as rated by parents) than among their peers with no or mild perceptual difculties. This may be due to an increased focus on retention of motor skills or accommodation to perceptual difculty over time, or indeed the nding may simply be spurious because of the lack of objectivity in our measurement of perception. Interestingly, the two statistical methods differed slightly when predicting the GMFM score for visit 2, with the GMFCS level just approaching statistical signicance in the bootstrap model (p=0.057) and seizure activity approaching statistical signicance in the regression model (p=0.067). Other authors have shown age, intellectual impairment, epilepsy, GMFCS level, manual ability, and selective motor control to be signicant predictors of gross motor function.12,2527 Signicant age-related trends were also detected with two of the PEDI CAS scores, mobility and social function, although with both of these scales of deterioration occurred later in the adolescent years (approximately age 1617y) than with energy efciency and GMFM. The PEDI CAS scores are reective of
Energy Efficiency, Activity, and Participation in CP Claire Kerr et al. 65

a childs performance,28 and it is interesting that the deterioration in motor capacity (GMFM) was also reected in motor performance (PEDI CAS mobility score). Interestingly, those children who underwent orthopaedic surgery between the assessments in the current study experienced a slower decline in PEDI CAS social function score than their non-operated peers (p=0.035). This may be due to increased effort on the part of the child and family to maintain an active social life during a period of increased dependence on health and therapy services. The stability of GMFCS ratings by parents and research clinical physiotherapists adds further validity and concurs with previous work.2931 This nding gives condence in the predictive ability of the GMFCS and also suggests that communication between health professionals and families in relation to functional ability can be transparent. Use of the recently published GMFCS 12 to 18 years age band descriptors32 may further increase the stability of the GMFCS ratings in this population. This study has identied the early adolescent years as a critical time in the development and retention of skills, with gait efciency reduced at age 12 and gross motor skills from age 13. These deteriorations may result from imbalance between physiological capacity and environmental demands, for example transition to a larger school resulting in wheelchair use for changing classes, but may also be inuenced by puberty, educational demands, peer pressure, self-awareness, and developing independence of the young adolescent. However, alarmingly, it is at this transition stage that therapy services frequently begin to reduce as educational and extracurricular demands increase, and the onus for maintenance of skills falls on the child and family. Whether the targeted provision of therapy and or services at this point would ameliorate this decline is not known; however, one would suspect that a more holistic approach would be required. A limitation of the study is the low response rate (38% of total population); however, this is offset to some extent by the representative nature of the recruited sample. Also of some concern is the relatively small number of children recruited to the study with non-spastic forms of CP. We attribute this to children with mixed forms of the condition being overclassied as having spastic CP. In terms of the outcome measures employed, we felt the PEDI was the most appropriate instru-

ment for the population and age range of children under investigation; however, we acknowledge its limitations with older, more functionally able children. Finally, completion of a practice walk prior to the collection of energy efciency data would have been desirable, but this was omitted in an attempt to minimize participant fatigue. Strengths of the current study include the populationbased approach, the representative nature of the recruited population,15 the high re-attendance rate (86%) at the second assessment, a single assessor and assistant measurement team, and a comprehensive measurement framework. Frequently, studies with two measurement points are considered cross-sectional; however, the current study was conceived within a longitudinal framework and analysed by longitudinal means such as analysis of change and analysis of covariance.33 We hope to continue to follow this population into young adulthood in order to determine how best to describe CP and deal with its consequences. In conclusion, this study has demonstrated a peak in gait inefciency at age 12, a decline in gross motor function from puberty onwards, and a similar slow decline in the later adolescent years in gross motor performance and social function in children with ambulatory CP. However, these declining trends are not reected in childrens participation, which remained stable over the study period. This has implications for both health and education in terms of how and when services are delivered to this population in order to optimize and maintain motor performance.
ACKNOWLEDGEMENTS
The authors wish to thank the children and families who participated in this research and our research assistant, Joy Batoy. We also would like to thank Dr Chris Cardwell, Lecturer, School of Medicine, Dentistry and Biomedical Sciences, Centre for Clinical and Population Sciences, Queens University Belfast, for his advice and assistance with statistical analysis. We acknowledge the support of the Northern Ireland Health and Social Care Research and Development Ofce, project grant RSG 1708 01.

ONLINE SUPPLEMENTARY MATERIAL

Additional material and supporting information may be found in the online version of this article.

REFERENCES
1. Surveillance of Cerebral Palsy in Europe (SCPE). Surveillance of Cerebral Palsy in Europe: a collaboration of cerebral palsy surveys and registers. Dev Med Child Neurol 2000; 42: 81624. 2. Cans C, Surman G, McManus V, Coghlan D, Hensey O, Johnson A. Cerebral palsy registries. Semin Pediatr Neurol 2004; 11: 1823. 3. Wu YW, Day SM, Strauss DJ, Shavelle R. Prognosis for ambulation in cerebral palsy: a population-based study. Pediatrics 2004; 114: 126471. 4. Day SM, Wu YW, Strauss DJ, Shavelle RM, Reynolds RJ. Change in ambulatory ability of adolescents and young adults with cerebral palsy. Dev Med Child Neurol 2007; 49: 64753. 5. Rosenbaum PL, Walter SD, Hanna SE, et al. Prognosis for gross motor function in cerebral palsy. JAMA 2002; 288: 135763. 6. Russell D, Rosenbaum P, Cadman D, Gowland C, Hardy S, Jarvis S. The gross motor function measure: a means to evaluate the effects of physical therapy. Dev Med Child Neurol 1989; 31: 34152. 7. Russell DL, Rosenbaum PL, Avery LM, Lane M. Gross Motor Function Measure (GMFM-66 & GMFM-88) Users Manual. Clinics in Developmental Medicine No. 159. London: Mac Keith Press, 2002. 8. Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi B. Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol 1997; 39: 21423. 9. 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: 295302. 10. Beckung E, Carlsson G, Carlsdotter S, Uvebrant P. The natural history of gross motor development in children with cerebral palsy aged 1 to 15 years. Dev Med Child Neurol 2007; 49: 7516.

66 Developmental Medicine & Child Neurology 2011, 53: 6167

11. Oefnger D, Bagley A, Rogers S, et al. Outcome tools used for ambulatory cerebral palsy: responsiveness and minimum clinically important differences. Dev Med Child Neurol 2008; 50: 91825. 12. Voorman JM, Dallmeijer AJ, Knol DL, Lankhosrt GL, Becher JG. Prospective longitudinal study of gross motor function in children with cerebral palsy. Arch Phys Med Rehabil 2007; 88: 8716. 13. Van Eck M, Dallmeijer AJ, Voorman JM, Becher JG. Longitudinal study of motor performance and its relation to motor capacity in children with cerebral palsy. Dev Med Child Neurol 2009; 51: 30310. 14. Parkes J, Dolk H, Hill N. Children and Young People with Cerebral Palsy in Northern Ireland Birth years 19771997. A comprehensive report from The Northern Ireland Cerebral Palsy Register. Belfast: Queens University Belfast, 2005. 15. Parkes J, Kerr C, McDowell BC, Cosgrove AP. Recruitment bias in a population-based study of children with cerebral palsy. Paediatrics 2006; 118: 161622. 16. Kerr C, Parkes J, Stevenson M, Cosgrove AP, McDowell BC. Energy efciency in gait, activity, participation and health status in children with cerebral palsy. Dev Med Child Neurol 2008; 50: 20410. 17. Haley SM, Coster WJ, Ludlow LH, Haltiwanger J, Andrellos P. Pediatric Evaluation of Disability Inventory (PEDI). Development, Standardization and Administration Manual. Boston: Boston University, 1992.

18. Mackie PC, Jessen EC, Jarvis SN. The lifestyle assessment questionnaire: an instrument to measure the impact of disability on the lives of children with cerebral palsy and their families. Child Care Health Dev 1998; 24: 47386. 19. Carstairs V, Morris R. Deprivation and Health in Scotland. Aberdeen: Aberdeen University Press, 1991. 20. Landis RJ, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: 15974. 21. Andersson C, Mattsson E. Adults with cerebral palsy: a survey describing problems, needs, and resources, with special emphasis on locomotion. Dev Med Child Neurol 2001; 43: 7682. 22. Jahnsen R, Villien L, Egeland T, Stanghelle JK, Holm I. Locomotion skills in adults with cerebral palsy. Clin Rehabil 2004; 18: 30916. 23. Opheim A, Jahnsen R, Olsson E, Stanghelle JK. Walking function, pain, and fatigue in adults with cerebral palsy: a 7year follow-up study. Dev Med Child Neurol 2009; 51: 381 8. 24. Schwartz MH, Koop SE, Bourke JL, Baker R. A nondimensional normalization scheme for oxygen utilization data. Gait Posture 2006; 24: 1422. 25. Ostensjo S, Brogen Carlberg E, Vollestad NK. Motor impairments in young children with cerebral palsy: relationship to gross motor function and everyday activities. Dev Med Child Neurol 2004; 51: 924. 26. Morris C, Kurinczuk JJ, Fitzpatrick R, Rosenbaum PL. Do the abilities of children with cerebral palsy explain their activ-

ities and participation? Dev Med Child Neurol 2006; 48: 954 61. 27. Voorman JM, Dallmeijer AJ, Schuengel C, Knol DL, Lankhorst GJ, Becher JG. Activities and participation of 9- to 13year old children with cerebral palsy. Clin Rehabil 2006; 20: 93748. 28. Holsbeeke L, Ketelaar M, Schoemaker MM, Gorter JW. Capacity, capability, and performance: different constructs or three of a kind? Arch Phys Med Rehabil 2009; 90: 84955. 29. Wood E, Rosenbaum P. The gross motor function classication system for cerebral palsy: a study of reliability and stability over time. Dev Med Child Neurol 2000; 42: 2926. 30. Palisano RJ, Cameron D, Rosenbaum PL, Walter SD, Russel D. Stability of the gross motor function classication system. Dev Med Child Neurol 2006; 48: 4248. 31. Imms C, Carlin J, Eliasson AC. Stability of care-giver reported manual ability and gross motor function classications of cerebral palsy. Dev Med Child Neurol 2010; 52: 1539. 32. Palisano RJ, Rosenbaum P, Bartlett D, Livingston MH. Content validity of the extended and revised gross motor function classication system. Dev Med Child Neurol 2008; 50: 74450. 33. Twisk JWR. Applied Longitudinal Data Analysis for Epidemiology, a Practical Guide. Cambridge: Cambridge University Press, 2003.

Energy Efficiency, Activity, and Participation in CP Claire Kerr et al. 67

20
GMFM-66 difference score

E E E E E EE E E E E E

10

E E E

-10

E E EE E E E E E E E E E E E E E E E E E E E EE EE E E E E E EE E E EE EE E E E E E E E EE E E E E EE EE E E E E E E E E E E E EE E E EEE E E E E E EE E E EE E E E E E E E E E E E E E E E E

-20 5 10

15

age at first assessment (y)

Table S1: Multiple linear regression and bootstrap analyses for those measurement tools exhibiting significant agerelated changes. Statistically significant models are shown in bold. Measurement tool Net oxygen cost of walking (n=74) Significantly contributing variables Baseline score Age Sex CP type Surgery GMFCS level II GMFCS level III IQ Seizures Perception Concentration Regression analysis Coefficient 0.834 0.001 0.001 0.002 0.046 0.011 0.051 0.042 0.001 0.009 0.042 95% CI 0.6980.970 0.002 to 0.005 0.020 to 0.023 0.023 to 0.027 0.081 to 0.010 0014 to 0.037 0.0050.096 0.0130.071 0.030 to 0.033 0.045 to 0.026 0.071 to 0.013 Regression adjusted r=0.865 GMFM-66 (n=102) Baseline score Age 0.848 0.724 0.6841.012 1.149 to 0.011 0.051 0.042 0.001 0.009 0.042 Bootstrap analysis Coefficient 0.834 0.001 0.001 0.002 0.046 95% CI 0.5251.144 0.002 to 0.004 0.017 to 0.020 0.025 to 0.029 0.081 to 0.010 0.014 to 0.037 0.0002 to 0.102 0.008 to 0.076 0.035 to 0.038 0.033 to 0.014 0.074 to 0.010 Bootstrap adjusted r=0.865 0.848 0.724 0.6891.007 1.087 to

0.300 Sex CP type Surgery GMFCS level II GMFCS level III IQ Seizures Perception Concentration 0.524 0.253 0.453 3.019 7.104 1.066 3.316 5.281 2.142 2.082 to 3.130 3.000 to 3.507 3.417 to 4.323 6.378 to 0.342 13.461 to 0.748 4.611 to 2.478 1.066 6.869 to 0.236 3.316 1.2149.350 1.382 to 5.665 5.281 2.142 0.524 0.253 0.453 3.018 7.104

0.361 2.058 to 3.105 2.477 to 2.984 3.285 to 4.191 6.474 to 0.438 14.429 to 0.220 4.708 to 2.576 6.361 to 0.271 1.1909.374 1.876 to 6.159

Regression adjusted r=0.822 PEDI CAS mobility (n=117) Baseline score Age Sex CP type Surgery GMFCS level II GMFCS level III 0.508 0.232 1.903 0.447 4.202 3.277 8.316 0.3300.686 0.422 to 0.885 5.614 to 1.808 4.981 to 4.088 1.145 to 9.549 8.063 to 1.509 15.973 to 0.659

Bootstrap adjusted r=0.822 0.508 0.231 1.903 0.447 4.202 3.277 8.316 0.3110.704 0.355 to 0.818 5.603 to 1.797 5.085 to 4.192 2.000 to 10.404 8.559 to 2.004 17.684 to 1.053

IQ Seizures Perception Concentration

1.969 0.712 0.108 0.531

7.007 to 3.069 5.858 to 4.434 5.513 to 5.730 5.424 to 4.362

1.969 0.712 0.108 0.531

7.303 to 3.366 6.913 to 5.489 4.673 to 4.890 5.614 to 4.551

Regression adjusted r=0.532 PEDI CAS social function (n=117) Baseline score Age Sex CP type Surgery GMFCS level II GMFCS level III IQ Seizures Perception Concentration 0.524 0.073 2.648 2.608 6.935 3.482 3.130 7.345 2.558 3.584 5.632 0.3910.657 0.576 to 0.430 5.949 to 0.654 1.456 to 6.673 2.06311.807 7.607 to 0.643 8.992 to 2.731 11.837 to 2.853 2.017 to 7.132 8.582 to 1.414 10.082 to 1.182 Regression adjusted r=0.539

Bootstrap adjusted r=0.532 0.524 0.073 2.648 2.608 6.935 3.482 3.130 7.345 2.558 3.584 5.632 0.2080.840 0.641 to 0.495 6.007 to 0.711 1.538 to 6.755 0.50213.368 7.408 to 0.444 8.956 to 2.695 12.239 to 2.451 3.135 to 8.250 9.946 to 2.779 10.533 to 0.731 Bootstrap adjusted r=0.539

CI, confidence interval; GMFCS, Gross Motor Function Classification System; GMFM-66, Gross Motor Function Measure66; PEDI, Paediatric Evaluation of Disability Inventory; CAS, caregiver assistance score.

Figure S1: Relationship between GMFM-66 difference score (visit 2 visit 1) and age. Regression line with 95% CIs is shown.