Abstract
Background A vicious circle of decreased physical fitness, early fatigue, and low physical activity levels (PAL) is thought to affect children with cerebral palsy (CP). However, the relationship of changes in physical fitness to changes in PAL and fatigue is unclear.
Objective The objective of this study was to investigate the associations among changes in physical fitness, walking-related PAL, and fatigue in children with CP.
Design This study was a secondary analysis of a randomized controlled trial with measurements at baseline, 6 months (after the intervention period), and 12 months.
Methods Twenty-four children with bilateral spastic CP and 22 with unilateral spastic CP, aged 7 to 13 years, all walking, participated in this study. Physical fitness was measured by aerobic capacity, anaerobic threshold, anaerobic capacity, and isometric and functional muscle strength. Walking-related PAL was measured using an ankle-worn activity monitor for 1 week. Fatigue was determined with the Pediatric Quality of Life (PedsQL) Multidimensional Fatigue Scale. Longitudinal associations were analyzed by random coefficient regression analysis.
Results In children with bilateral CP, all fitness parameters showed a positive, significant association with walking-related PAL, whereas no associations between physical fitness and walking-related PAL were seen in children with unilateral CP. No clinically relevant association between physical fitness and fatigue was found.
Limitations Although random coefficient regression analysis can be used to investigate longitudinal associations between parameters, a causal relationship cannot be determined. The actual direction of the association between physical fitness and walking-related PAL, therefore, remains inconclusive.
Conclusions Children with bilateral spastic CP might benefit from improved physical fitness to increase their PAL or vice versa, although this is not the case in children with unilateral CP. There appears to be no relationship between physical fitness and self-reported fatigue in children with CP. Interventions aimed at improving PAL may be differently targeted in children with either bilateral or unilateral CP.
It has been suggested that children with a physical disability may be trapped in a vicious circle of low physical fitness, early fatigue in daily activities, and inactivity, resulting in deconditioning and a further decrease in physical activity.1 The physical activity level (PAL) tends to deteriorate during the transition to adulthood, especially in people with a physical disability.2 From this perspective, establishing a healthy and active lifestyle during childhood is even more important for individuals with a disability, who are at higher risk for developing secondary conditions such as cardiovascular disease, diabetes, and obesity.1
Cerebral palsy (CP) is the most common cause of physical disability in childhood3 and is associated with low physical fitness,4 decreased PAL,5 and general fatigue.6 The health-related components of physical fitness are defined by the American College of Sports Medicine (ACSM) as cardiorespiratory (aerobic) fitness, body composition, muscle strength and endurance, and flexibility.7 Anaerobic fitness is not a separate fitness component according to this definition. However, anaerobic fitness is, next to aerobic fitness and muscle strength, an important determinant for physical activity and exercise in children who have short, intermittent activity patterns.8,9 Cerebral palsy10 is defined as “a group of disorders of the development of movement and posture causing activity limitation that are attributed to non-progressive disturbances that occurred in the developing fetal or infant brain.”11 Due to motor impairments, children with CP have higher energy requirements during certain activities such as walking.12 Also, a lower anaerobic threshold in CP4 implies that children with CP walk close to or above their anaerobic threshold,13 inducing early fatigue. The high energy requirements and the lower anaerobic threshold might cause a lower PAL as a compensatory mechanism to prevent fatigue, aggravating the above-mentioned vicious circle.14
Recently, the focus of intervention programs to break this vicious circle of low physical fitness, early fatigue in daily activities, and inactivity, resulting in deconditioning and a further decrease in physical activity, has been on improving physical fitness and increasing PAL by fitness training and stimulating an active lifestyle.15,16 It has been reported that physical fitness can be improved in children with CP,16,17 but it is not yet clear whether increased physical fitness levels lead to higher activity levels.16 A recent multicomponent physical activity intervention including both fitness training and a lifestyle intervention showed no effect on physical fitness and walking-related PAL.18 A large interindividual variation in changes in both physical fitness and walking-related PAL was found in the intervention and control groups. These findings indicate that some participants improved and others deteriorated, despite their group allocation, and that physical fitness and walking-related PAL were influenced by factors other than the intervention. A secondary analysis on these data provide insight into the relationship between changes in physical fitness and walking-related physical activity, indicating whether improved physical fitness is related to higher PAL levels regardless of interventions.
Understanding the longitudinal relationship between changes in physical fitness and PAL may contribute to the development of effective programs for improving both physical fitness and PAL in children with CP. Although cross-sectional correlations may be influenced by the large interindividual variability of physical fitness and PAL in people with CP, longitudinal linear regression analyses provide information on intraindividual changes over time. This approach has an advantage over reporting group mean effects, in which individual responses are averaged out in a heterogeneous group such as people with CP.19 This longitudinal relationship provides insight into whether a change in physical fitness is related to a change in walking-related PAL, which is in contrast to cross-sectional relationships, which do not preclude an association between changes in components.20 Previous studies investigating the association between physical fitness and PAL using a cross-sectional design showed no associations between PAL and peak oxygen uptake (V̇o2peak) in children14 and adults21,22 with CP. However, physical strain, defined as the oxygen consumption (V̇o2) during walking expressed as a percentage of V̇o2peak, did show an association with PAL.14,22 Oxygen consumption during walking is elevated in children with CP and is closely related to the severity of the motor disorder.23 Orthoses, orthopedic surgery, or spasticity treatment have the potential to reduce V̇o2 during walking.24–26 In addition, increasing the V̇o2peak while the V̇o2 during walking remains constant will reduce physical strain and may concomitantly improve walking-related PAL. Consequently, it is anticipated that changes in V̇o2peak are related to changes in walking-related PAL.
Our hypothesis is that if physical fitness changes in children with CP, this change will be related to a change in PAL and to a change in fatigue. To investigate this hypothesis, we performed a secondary analysis of the above-mentioned intervention study.18 Using a longitudinal design, the objective of this study was to investigate the association between changes (either improvements or decreases, regardless of group allocation) in physical fitness, walking-related PAL, and fatigue in children with CP.
Method
Participants
This study included 24 children with bilateral CP and 22 children with unilateral CP who were recruited in special schools or through physical therapy practices from August 2009 to February 2011. Measurements were performed at baseline, at 6 months (after the intervention period) (X̅=31 weeks, SD=2.4), and at 12-month follow-up (X̅=52 weeks, SD=3.4) and were part of a randomized controlled trial evaluating the effects of a physical activity stimulation program.15 The intervention program consisted of physical fitness training focusing on gross motor activities, anaerobic fitness and muscle strength, counseling, and home-based physical therapy, and the control group continued their regular physical therapy. Inclusion criteria were: (1) children with spastic CP; (2) 7 to 13 years of age; and (3) Gross Motor Function Classification System (GMFCS) level I, II (walking without aids), or III (walking with aids). Exclusion criteria were: (1) contraindications for maximal exercise, (2) history of botulinum toxin injections or serial casting (<3 months), and (3) surgery (<6 months). All participants were aged above 12 years, and all participants' parents signed an informed consent form. Participants' characteristics are shown in Table 1.
Participants' Characteristicsa
Procedure
All participants visited the outpatient unit of VU University Medical Centre, Amsterdam, the Netherlands. Measurements were taken at baseline, at 6 months, and after 1 year (all 3 either in the morning or in the afternoon) and consisted of body composition assessments, physical fitness measurement, and calibration of the StepWatch Activity Monitor (Cyma Corp, Seattle, Washington). After this session, the children wore the StepWatch Activity Monitor for 1 week during waking hours, excluding bathing time and swimming. The children were asked to fill in a questionnaire on previously experienced fatigue prior to each measurement session. All data were collected by 2 trained researchers. All measurements were performed by the same researcher (A.B.). Data analysis was performed by the primary author (A.B.) (for body composition and fitness measurements) or by the second author (L.vW.) (for fatigue and physical activity measures).
All children agreed to participate in all measurements, except for 2 children who were unable to perform measurements at 12 months for logistical reasons (eg, availability of laboratories or participants). At all measurement sessions, there were missing data for a variety of reasons, including incomplete registrations (walking-related PAL), questionnaires that were not completed (fatigue), refusal to wear the mask, and lack of motivation or equipment problems (physical fitness). All children classified as GMFCS level III completed all anaerobic fitness and strength measurements, except for one child who was not able to complete the functional strength test at 12-month follow-up. Out of the 8 children classified as GMFCS level III, 3 children at the baseline measurement, 4 children at the 6-month measurement, and 2 children at the 12-month measurement were not able to complete the V̇o2peak test due to peripheral restrictions (eg, severely impaired selectivity [not able to cycle at 60 rpm with increasing load]).
Measurements
Body composition.
Body height and weight were measured on an electronic scale (DGI 250D, Kern version 3.3 10/2004, Kern & Sohn GmbH, Balingen-Frommern, Germany), which enabled calculation of body mass index (BMI). Consecutively, skinfold measurements were performed at the subscapular site and suprailiac site of the nondominant arm (bilateral) or nonaffected side (unilateral) using a Holtain skinfold caliper (accuracy 0.2 mm, ProCare BV, Groningen, the Netherlands), providing a summed score of skinfold thickness.27
Physical fitness.
Children performed a maximal aerobic exercise test on a cycle ergometer (Corival V2, Lode BV, Groningen, the Netherlands) until exhaustion. The exercise test started with a 3-minute warm-up phase, followed by a 4- to 5-minute steady-state phase at submaximal exercise, a 1-minute rest, and finally a maximal phase starting with the load of the submaximal phase being incremented every minute based on height and GMFCS level.28 Heart rate was measured using a heart rate monitor, and gas was analyzed with a gas analysis system (Quark CPET, COSMED Srl, Rome, Italy) with the corresponding software (PFT CPET Suite, version 9.1b). Prior to testing, the flow sensor was calibrated with a 3-L syringe, and the oxygen and carbon dioxide concentration sensors were calibrated with ambient air and a reference gas of a known mixture. Data were included if at least 2 out of the following 3 criteria for achieving maximal exercise were met: (1) heart rate >180 bpm, (2) respiratory exchange ratio >1.00, and (3) patient-reported exhaustion was present.29 Peak oxygen uptake was defined as the highest V̇o2 over 30 seconds. Anaerobic threshold was determined by the V-slope method by 2 independent raters.30 Test-retest reliability of V̇o2peak assessments in children with CP using this protocol was excellent (intraclass correlation coefficient [ICC] of .94 and standard error of measurement [SEM] of 2.06 mL·kg−1·min−1).28
A 20-second Wingate test on the leg cycle ergometer was performed, a full-out sprint test against a constant workload, with the mean power output over 20 seconds (P20mean) representing anaerobic capacity. Wingate software (Wingate Software V1, Lode BV, Groningen, the Netherlands) was used to apply the workload and to measure P20mean. Test-retest reliability of the 20-second Wingate test showed a high ICC of .99 and an SEM of 0.219 W·kg−1 for P20mean in children with CP.31
Isometric muscle strength of the knee extensors and the hip abductors was measured by use of the “make test” with handheld dynamometry (MicroFet, Biometrics, Almere, the Netherlands) at the nondominant leg by taking the average over 3 measurements, preceded by a practice trial.32 The child's limb was fixed by the assessor, and the child pushed for 3 seconds with maximal force against the dynamometer. Peak force (in newtons) and the moment arm (in meters) were measured.32 This procedure's feasibility and good intersession reliability were shown in children with CP, with SEMs of 11.3% (knee flexors) and 16.6% (hip abductors) and ICCs >.82 for both muscle groups.32
Functional strength was measured with the lateral step-up test (with both the dominant and nondominant legs) and the sit-to-stand test, where the number of repetitions over 30 seconds was determined.33 The number of repetitions was summed over the 3 tests, resulting in a total score for functional muscle strength.33
Walking-related physical activity.
Walking-related PAL was determined by measuring walking activity with the biaxial StepWatch Activity Monitor 3.0. This device was worn at the ankle of the dominant leg and measures accelerations of the leg in the frontal-sagittal plane per time interval.34 The psychometric properties of this device have been shown to be good in children who are developing typically,35 and by adjusting the sensitivity settings, the StepWatch can accurately record strides in children with CP.36 Calibration was carried out with each child walking on an oval 50-m track, while strides were counted manually and concurrently registered by the StepWatch device and sensitivity settings were adjusted until StepWatch recordings and manual counting agreed 95%. StepWatch calibration resulted in a mean accuracy of 99.8% (SD=3.4%). Mean values per minute were stored, providing average strides·min−1. Walking-related PAL was expressed as total strides·day−1 and minutes at high stride rates (>30 strides·min−1 [SR30]), as used in previous studies,36,37 on an average weekday scaled by 4/5 school day and 1/5 weekend day. At least 3 school days and 1 weekend day were required to provide reliable data.38 A minimum registration duration of 10 hours on school days and 8 hours on weekend days was required for days to be included, as recommended in literature.39 Days were excluded if more than 3 hours of data were missing within the time interval awake. The time interval awake was registered by the parent or child, or both, in a diary.
Fatigue.
Experienced fatigue was assessed with the Pediatric Quality of Life Inventory (PedsQL) Multidimensional Fatigue Scale domain “general fatigue” that was completed by the child. This domain encompasses 6 items with a 5-point response, where 0=“never a problem” and 4=“almost always a problem.” An example of an item included in this domain is “I feel too tired to do things that I like to do.” The items are reversely scored and linearly transformed to a 0 to 100 scale, with higher scores indicating less experienced fatigue.6,40 Scale internal consistency reliability was found to be good in children with CP (α=.79) and their parents (α=.91).6
Data Analysis
Participant characteristics were compared between bilateral and unilateral CP with a Student t test, a chi-square test (for sex), or a Fisherman exact test (for GMFCS scores). The Student t test with adjustment for unequal variances was used for height and weight. Distribution of the data was checked using inspection of mean values, standard deviations, and ranges and using visual inspection of the histogram and normal Q-Q plot of the residuals of the mixed model.
As the data were normally distributed, a parametric test was applied. In longitudinal analysis, repeated measurements over time are performed within the same participants. These repeated measurements for each participant are dependent on each other. The statistical method should take into account this within-subject correlation. The random coefficient regression analysis adjusts for the within-subject correlation in longitudinal data.20 Therefore, we performed a random coefficient regression analysis with a random intercept to determine the longitudinal association of the fitness parameters with walking-related PAL and fatigue.20
The random coefficient regression analysis is a mixed-model analysis with fixed effects and a random intercept. Walking-related physical activity or fatigue was used as the dependent variable, and fitness components were used as independent variables. A random coefficient did not improve the model; therefore, only a random intercept was included in the model. With this analysis, all data from all participants (N=46) with at least 2 out of the 3 repeated measurements as well as data of children with missing data at one occasion were included in the analysis. This statistical method handles missing data as missing at random. Each dependent variable (strides per day, SR>30, or fatigue) was examined in a separate model. Also, the independent variables (V̇o2peak, anaerobic threshold, P20mean, knee extension, hip abduction, and functional strength) were included in separate models. In each regression model, we checked effect modification by entering anatomical involvement (as a dichotomous variable: unilateral or bilateral CP), sex, or GMFCS score (as dummy variables) as a covariate and as an interaction term with the independent fitness variable. Finally, age or height was included as a covariate in the models to check for confounding (when regression coefficients of the fitness parameter changed >10%). Each model included 1 independent variable, 1 interaction term, and 2 covariates maximum. Statistical analyses were performed using IBM SPSS Statistics, version 20 (IBM Corp, Armonk, New York). The level of significance was set at P<.05.
Role of the Funding Source
This study was supported by a grant from The Netherlands Organisation for Health Research and Development (ZonMw) and the Phelps Foundation for Spastics.
Results
Random Coefficient Regression Analysis
Analysis revealed a significant interaction effect of bilateral or unilateral CP (indicating different associations for bilateral CP than for unilateral CP) for all analyzed associations between physical fitness and walking-related PAL, except for anaerobic threshold, isometric knee extension strength, and hip abduction strength. There were no interaction effects of localization for the associations between the fitness parameters and fatigue and no significant interaction effects with any of the investigated associations for sex and GMFCS scores. Therefore, descriptive statistics and associations are presented for the whole group and separately for children with bilateral and unilateral CP. Descriptive statistics of physical fitness, walking-related PAL, and fatigue and the number of cases per variable are presented in Table 2. Table 3 shows the regression coefficients of the longitudinal associations. For children with bilateral CP, significant positive associations with walking-related PAL (both parameters) were found for all fitness parameters. For example, 1 mL·kg−1·min−1 V̇o2peak translated to a change of 98 strides per day (Tab. 3). No associations of fitness parameters with walking-related PAL were found for children with unilateral CP. Functional muscle strength was significantly and positively associated with fatigue in children with unilateral involvement, whereas all other fitness parameters were not associated with fatigue in all children.
Descriptive Statistics for Physical Activity, Fatigue, and Physical Fitnessa
Associations of Physical Fitness With Physical Activity and Fatiguea
Discussion
This was the first study to investigate the longitudinal relationship of physical fitness to walking-related PAL and experienced fatigue in children with CP. Our results showed a significant positive association between all fitness parameters and walking-related PAL in children with bilateral CP. For children with unilateral CP, no association was found between physical fitness and walking-related PAL. Functional muscle strength was significantly related to fatigue.
The fitness parameter showing the strongest association to walking-related PAL in children with bilateral CP was V̇o2peak. A recent study showed that children with CP were able to increase their V̇o2peak by 7 mL·kg−1·min−1 compared with a control group,41 which corresponds to a clinically relevant increase of 686 strides per day (16% of our group mean) (Tab. 2). However, determining the association between changes in physical fitness and walking-related PAL parameters does not necessarily ascertain causality, as the association also might be the reverse. The longitudinal association between these measures indicates not only that improving V̇o2peak has the potential to increase walking-related PAL but also that a higher walking-related PAL might lead to an improved V̇o2peak in children with CP.
In contrast to previous cross-sectional studies where no association between V̇o2peak and PAL was found in adults and a relatively small sample of children with CP,14,21,22 physical fitness showed a longitudinal association to walking-related PAL in children with bilateral CP in the current study. In addition, an intervention study showed an improvement in physical fitness with only a positive trend toward increased PAL, although V̇o2peak for aerobic capacity was not included.16 Although there also might be a cross-sectional component in the interpretation of the regression coefficient, our present longitudinal results strongly suggest that a change in physical fitness is related to a change in walking-related PAL in children with bilateral CP.
Our results show that the association between physical fitness and walking-related PAL for children with bilateral CP differs from that for children with unilateral CP. The present results showed that physical fitness and walking-related PAL were more severely reduced in children with bilateral CP than in children with unilateral CP, which is in agreement with previous studies.42,43 Decreased muscle strength and muscle volume on the affected side of the body in unilateral CP influences the aerobic and anaerobic maximal capacity that can be achieved and, therefore, results in a lower but less pronounced decreased aerobic and anaerobic capacity than is present in bilateral CP.4,44 An explanation for the higher walking-related PAL may be the lower oxygen cost of walking in children with unilateral CP, resulting in lower physical strain compared with children with bilateral CP.23 Therefore, walking-related PAL in children with unilateral CP might be less influenced by lower physical fitness because the physical strain is lower. Also, other factors, such as cognitive, behavioral, and environmental factors, might have a greater impact on PAL in children with unilateral CP.45 These results indicate that, for children with unilateral CP, a change in physical fitness does not necessarily lead to a change in walking-related PAL.
The association between changes in anaerobic capacity and muscle strength and a change in PAL in children with bilateral CP confirms that anaerobic capacity and muscle strength also contribute to higher walking-related PAL. Earlier findings showed that decreases in anaerobic exercise responses are more strongly related to the severity of CP, in contrast to the aerobic exercise responses, indicating that aerobic exercise responses are determined to a greater extent by other factors, such as the amount of physical exercise.4 This finding supports the stronger association we found between aerobic fitness and walking-related PAL. It should be noted, however, that the large decrease in anaerobic capacity compared with peers (−55%) and the short, intermittent activity patterns that characterize physical activity of children indicate that this physical fitness component also remains important in enhancing physical fitness and walking-related PAL.4,8 In addition, anaerobic training can contribute to improvement in both aerobic and anaerobic fitness in children with CP.17
The anaerobic threshold, which was related to the amount of strides per day in children with bilateral CP, might be a restricting factor in walking-related PAL, as the anaerobic threshold appears at a lower exercise intensity compared with reference values.4 The anaerobic threshold was found to appear at 19.4 mL·kg−1·min−1,4 which is at the same level as the measured average V̇o2 during walking, 19.7 mL·kg−1·min−1 (both values measured in children with CP classified as GMFCS level I, a combined sample with unilateral and bilateral involvement).13 As a result, children with CP walk at an intensity at or above the anaerobic threshold, requiring anaerobic glycolysis, which hampers walking for longer periods.13 If the anaerobic threshold is improved, it enables performance of activities at a higher absolute exercise intensity.
In children with unilateral CP, we uncovered a significant association between functional muscle strength and fatigue. This association was not found for children with bilateral CP or with any of the other fitness parameters. The ability to perform one additional repetition might have resulted in the higher score on the fatigue scale (less fatigue) of 0.32. In the knowledge that a previous study showed children and adolescents with CP (bilateral and unilateral) were able to improve their functional muscle strength by 20% as a result of training, this score would correspond to an improvement in fatigue of 3.4 points (<5%).17 With a group mean score of 72 in unilateral CP, compared with a normative score of 85 in peers reported in a previous study,40 we believe that this 3.4-point increase does not seem clinically relevant. An explanation for the lack of a relationship between physical fitness and fatigue may lie in the actual construct that the multidimensional fatigue scale measures.6,46,47 The PedSQL measures general fatigue, whereas peripheral muscle fatigue might be the restricting factor for walking-related PAL. Another explanation may lie in our assumption that the child experiences early fatigue during walking, whereas the child might already have reduced the walking-related PAL to prevent fatigue. Interestingly, an earlier study showed that parents gave a more severe rating to the fatigue experienced by their children than the children themselves,6 indicating that children do not experience high levels of fatigue. Future research should investigate whether fatigue rated by the parent does relate to walking-related PAL.
Limitations
Although random coefficient analysis can be used to investigate longitudinal associations between parameters, a causal relationship cannot be determined. The actual direction of the association between physical fitness and walking-related PAL, therefore, remains inconclusive. Nevertheless, it appears that focusing on either factor contributes to improvements in the other factor. In this study, we performed multiple tests because we were interested in the associations between each fitness component and physical activity and fatigue separately, as each fitness component usually requires a different intervention. If we account for multiple testing by setting the P value at .01, most associations remain significant (with the exception of anaerobic threshold and knee muscle strength with strides per day) and leave the main conclusions unaltered. Another limitation was that muscle strength of only 2 muscle groups was measured; however, the significant associations between strength and walking-related PAL in children with bilateral involvement indicates that muscle strength may be a limiting factor for physical activity in this population. An activity-specific mode of exercise testing (running) would be preferred in children who are able to walk. However, a cycle ergometer test was performed because a cycle ergometer was considered to be more suitable for children who have disturbances in balance and for children who are dependent on assistive devices for walking.
It can be concluded that changes in physical fitness are related to changes in walking-related PAL in children with bilateral CP. In contrast, no association was found between physical fitness and walking-related PAL in children with unilateral CP. Functional muscle strength was significantly (but without clinical relevance, as we hypothesized) related to fatigue. Although children with bilateral spastic CP might benefit from better physical fitness and, consequently, an increased walking-related PAL (or vice versa), this is not the case in children with unilateral CP. The role of physical fitness in reducing fatigue remains unclear in all children with CP. Interventions aimed at improving walking-related PAL should be differently targeted in children with bilateral CP compared with children with unilateral CP.
Footnotes
All authors provided concept/idea/research design, writing, and project management. Dr Balemans and Dr van Wely provided data collection. Dr Balemans and Dr Dallmeijer provided data analysis. Dr Becher and Dr Dallmeijer provided fund procurement. Dr van Wely and Dr Becher provided participants and facilities/equipment. Dr van Wely and Dr Dallmeijer provided consultation (including review of manuscript before submission). The authors thank all children and their parents for participating in this study.
This study was approved by the Institutional Ethics Committee of VU University Medical Centre, Amsterdam, the Netherlands.
This study was supported by a grant from The Netherlands Organisation for Health Research and Development (ZonMw) and the Phelps Foundation for Spastics.
- Received June 24, 2014.
- Accepted January 28, 2015.
- © 2015 American Physical Therapy Association