Growth and Health in Children With Moderate-to-Severe Cerebral Palsy
BACKGROUND. Children with cerebral palsy frequently grow poorly. The purpose of this study was to describe observed growth patterns and their relationship to health and social participation in a representative sample of children with moderate-severe cerebral palsy.
METHODS. In a 6-site, multicentered, region-based cross-sectional study, multiple sources were used to identify children with moderate or severe cerebral palsy. There were 273 children enrolled, 58% male, 71% white, with Gross Motor Function Classification System levels III (22%), IV (25%), or V (53%). Anthropometric measures included: weight, knee height, upper arm length, midupper arm muscle area, triceps skinfold, and subscapular skinfold. Intraobserver and interobserver reliability was established. Health care use (days in bed, days in hospital, and visits to doctor or emergency department) and social participation (days missed of school or of usual activities for child and family) over the preceding 4 weeks were measured by questionnaire. Growth curves were developed and z scores calculated for each of the 6 measures. Cluster analysis methodology was then used to create 3 distinct groups of subjects based on average z scores across the 6 measures chosen to provide an overview of growth.
RESULTS. Gender-specific growth curves with 10th, 25th, 50th, 75th, and 90th percentiles for each of the 6 measurements were created. Cluster analyses identified 3 clusters of subjects based on their average z scores for these measures. The subjects with the best growth had fewest days of health care use and fewest days of social participation missed, and the subjects with the worst growth had the most days of health care use and most days of participation missed.
CONCLUSIONS. Growth patterns in children with cerebral palsy were associated with their overall health and social participation. The role of these cerebral palsy-specific growth curves in clinical decision-making will require further study.
Growth is a fundamental and integral marker of health and well-being in children. Normal growth is an indicator of health, whereas abnormal growth may indicate illness, malnutrition, or something awry in the child's environment. Cerebral palsy (CP) is a common neurologic condition that originates in early childhood but affects individuals throughout their life span. Children with CP are known to grow poorly compared with their peers, but it is unclear whether this poor growth is “normal” for the population or a marker of some secondary condition that requires further evaluation and treatment. The basic clinical questions are: (1) does the observed “poor” growth negatively impact health and well-being of children with CP; and (2) if growth is improved, are health and well-being also improved? This investigation addresses the first of these questions.
Growth assessment requires reliable measures and comparison reference data. Reliable alternative measures of growth for children with CP have been widely adopted.1–3 However, appropriate reference growth curves for these children have not been clearly established. The objectives of this study are as follows: (1) to describe growth status in a region-based sample of children with moderate or severe CP enrolled in the North American Growth in Cerebral Palsy Project (NAGCPP), (2) to develop growth curves and calculate z scores, and (3) to correlate growth with markers of health and social participation.
The NAGCPP is a multicenter study designed to investigate growth, physical development, and nutritional status in children with CP according to an established protocol.4 The University of Virginia, Duke University, the University of North Carolina, the Children's Hospital of Philadelphia, the University of Rochester, McMaster University, and the University of British Columbia were the participating sites, with Duke and University of North Carolina operating as a single site.
The details of subject identification, recruitment, and enrollment were fully described previously.4 The subjects were a sample drawn from the 6 region-based sites, using multiple sources for identification: clinic samples, parent organizations, local United Cerebral Palsy Associations, school systems, public service announcements, physical therapists, local physicians, equipment vendors, and newspaper advertisements. Each of the sites defined a geographical region with a population of ∼500 000 people. All of the children with CP5 between the ages of 2 and 18 years in each region were eligible for recruitment. All of the subjects of moderate or severe impairment defined by Gross Motor Function Classification System levels III, IV, or V,6 who had clinically diagnosed CP, were included. Medical history was reviewed. Children with a history of genetic, metabolic, or neurodegenerative disease or other medical illnesses known to influence growth were excluded. However, children were not excluded on the basis of prematurity or low birth weight. Informed consent was obtained from the parent or legal guardian, and assent was obtained when appropriate. The institutional review boards of each participating site approved the study.
Subjects traveled to each study site to participate. At a single observation, trained observers performed a detailed anthropometric assessment of each child using standard techniques.1,7 Duplicate measures were obtained, and the average was used for analyses. Measures included knee height (KH), upper arm length (UAL), weight, midupper arm circumference, subscapular skinfold thickness (SUB), triceps skinfold thickness (TRI), and calculated arm muscle area (AMA). Duplicate measures were also used to calculate intraobserver and interobserver reliability, and these are reported in Table 1. Reliability was comparable with other published reports in children with CP.1 Sexual maturity ratings (Tanner stages) and severity of CP were determined after observers were trained for these assessments.8 We have reported previously on sexual maturity ratings in this population.9 Severity of CP was assessed using the Gross Motor Function Classification System (GMFCS).6 The GMFCS categorizes severity into 5 levels (I through V) based on gross motor function, predominantly independent mobility. This study enrolled subjects who were levels III, IV, and V, the more severe end of the spectrum. Older children who are level III can ambulate independently with a walker (household ambulators) but often use wheelchairs in the community; those at level IV cannot walk independently but can achieve independent mobility in a motorized wheelchair; and those at level V have no independent mobility but, rather, are transported.
Information regarding demographics, medical history, functional abilities, health care use, and social participation was collected by interview questionnaire. Questions regarding health and participation were adapted from the National Health Interview Survey of the Centers for Disease Control and Prevention (CDC)10 and referred to the 4 weeks preceding the measurements. These questions asked how many days over the previous 4 weeks were spent with the child at the emergency department, at the doctor's office, overnight in the hospital, home from school, home in bed, home from usual activities, or with a family member home and missing usual activities. No attempt was made to validate the caregiver's recall. The original questions from National Health Interview Survey asked similar questions but over the preceding 12 months. Other analyses of the NAGCPP population regarding general health status, nutritional status, bone mineral status, fracture rate, and health and participation have been published elsewhere.4,11–13
Cross-sectional data analysis was performed by the second author (M.C.) at the coordinating center using GAUSS 5.014 and SAS 9.1 (SAS Institute, Cary, NC)15 software. The development of growth curves used methods that were similar to those used for the current CDC growth charts for healthy U.S. children.16,17 These CP-specific curves were used to calculate CP-based z scores or SD scores for further analysis. To relate body size and health, K-means cluster analysis18 was used to group children according to their CP z scores for each of 6 measures of growth and body composition. These 6 measures were chosen a priori as a way to describe each child's growth in a comprehensive fashion using KH (lower extremity linear growth), UAL (upper extremity linear growth), weight (body mass), TRI (extremity fat stores), SUB (truncal fat stores), and midupper AMA (skeletal muscle mass). The number of clusters (3) was chosen empirically as a way to discriminate the sample, and the cluster analysis program assigned subjects to the appropriate cluster based on the z scores on the 6 measures. Thus, the technique created a single 3-level composite categorical variable of 6 separate continuous variables with each subject assigned to only 1 of the 3 clusters.
The population is described in Table 2 and consisted of 273 children, with 58% male and 71% white subjects. Gender-specific growth curves were constructed for all of the anthropometric data, and 4 illustrative curves are reproduced in the figures. Because a direct measure of recumbent length or stature could not be obtained in these children, KH was used as a proxy for linear growth. Figures 1 and 2 show the KH curves for boys and girls with CP, superimposed on a comparison KH curve generated for healthy children. KH estimates for the healthy children were based on converting the CDC height charts16 to KH estimates using the formulas in Chumlea et al.19 Figures 3 and 4 show the weight curves for boys and girls with CP superimposed on the weight curves for healthy children. All of these growth curves include the 10th, 25th, 50th, 75th, and 90th percentiles.
The cluster analysis discriminated 3 groups of children based on their CP z scores for each of 6 anthropometric assessments: KH, UAL, weight, TRI, SUB, and midupper AMA. Each child is represented in only 1 group. Figure 5 depicts these 3 groups, with the y-axis representing the average CP z score for each of the 6 measures. For discussion, these groups will be referred to as: Z−, the “smaller” group (bottom line in Fig 5; on average 1 SD below the mean); Zφ, the “middle” group (middle line in Fig 5; on average at about the mean); and Z+, the “larger” group (top line in Fig 5; on average ∼1 SD above the mean). Of these children, 90 (35%) were in the Z− group, 106 (42%) were in the Zφ group, and 58 (23%) were in the Z+ group.
These 3 groups were then correlated with markers of health and social participation over the preceding 4 weeks (Fig 6 and Table 3). The x-axis in Fig 6 and the numbers in Table 3 represent number of days, so higher numbers indicate more days with health care use and more days missed of school or usual activities. These analyses demonstrate that the Z+ group had fewest days of health care and fewest days missed of usual activities, and the Z− group had the most days of health care and most activities days missed, with the Zφ group falling in between. Although these results did not reach statistical significance (at the P < .05 level; see Table 3), their order, relative to each other, was consistent.
The 3 groups from cluster analysis did correlate with functional severity. Table 4 displays GMFCS and feeding tube status by group. There are significant (P < .001) differences among the groups with respect to the GMFCS classifications, with the Zφ and Z+ groups tending to have fewer GMFCS level V children (most severely impaired) and more GMFCS level 3 children (least severely impaired) than the Z− group. Negative binomial regression models were fit to the health and participation measures to assess the effect of group membership on these measures, adjusting for age, gender, and GMFCS/feeding tube status. The results of these analyses are displayed in Fig 7. These 2 figures display the estimated ratio of the health and participation measures from the Z− group, relative to the Zφ and Z+ groups, respectively, adjusting for age, gender, and GMFCS/feeding tube status. As in the unadjusted analyses, children in the Z− group consistently had greater health care use and lower social participation than children in the Zφ and Z+ groups. Moreover, for 2 of the health and participation measures (days missed at school and days of usual activity missed by family members) in the adjusted analyses (see Fig 7B), these differences were statistically significant (P < .05).
In pediatric practice, “poor” or subnormal growth is equated with poor health. Poor growth is determined by careful measurement (anthropometry) and comparison of the results to appropriate reference standards. The anthropometric assessment of children with CP has been difficult because of difficulty acquiring reliable measurements of stature (height or length) and lack of appropriate reference data for comparison.2 Although the measurement problem has been overcome through the use of alternative measures (eg, UAL, lower leg length, and skinfold thickness),1,3 the reference standard problem remains. The use of general population standards for a group of nonambulatory, severely impaired children is questionable.20
Most reference standards were established by measuring a representative cross-section of the healthy population and creating growth charts.16 Such growth charts are intended to encompass individual variability because of genetic potential. The primary measures used are stature (height or length), body mass (weight), and often some measure of body proportion (body mass index [BMI] or weight for height). An important feature of most growth charts is that they are intended to be descriptive of a population (“how they grew”) and are used to monitor normal growth and screen for abnormal growth. They are not intended to be prescriptive for health (“how they should grow”).21 Growth charts help clinicians determine the body size (stature and weight) and relative proportions (BMI) of an individual child compared with reference data from healthy children. This is usually described in terms of the child's growth percentile (50th percentile being “average”) or in SDs (z score) from the mean. However, a child at the 10th percentile weight for stature, for example, is not necessarily less healthy than a child at the 50th percentile, although clearly thinner. Although clinicians often consider the 50th percentile weight for a particular height as “ideal” (implying ideal for health), it is simply no more than a statistical average for the population. Relatively few data exist that actually link patterns of growth to clinically meaningful health indicators.
Diagnosis-specific growth curves have been developed for other health conditions, such as Down syndrome and Turner syndrome,22–24 conditions in which malnutrition is uncommon and the genetic abnormality directly influences stature. Having a reference growth curve for children with CP may be helpful to clinicians and beneficial to children. However, any representative sample of children with moderate or severe CP is likely to include many children with differing degrees of acute and chronic malnutrition and possibly growth hormone deficiency.25,26 This limits the usefulness of a descriptive reference growth chart for clinical management. What would be useful for clinical practice is a “prescriptive” growth curve, with statistical and clinically significant links between body size and proportions and health and social participation outcomes.
We developed CP-specific growth curves for measures of linear growth, body mass, and body composition. These growth curves, illustrated in the sample figures, are similar in shape to other growth curves, both in typical children16 and in special populations.23,24 Of note is a steady increase in growth throughout childhood followed by a plateau in adolescence. The usual adolescent growth spurt seems blunted in the CP curve percentiles compared with healthy children. On average, the children with moderate or severe CP were smaller, thinner, and lighter than their age- and gender-related peers without CP. The differences are significant and become more pronounced as children age. For example, whereas the CP KH percentiles overlap with the percentiles for typical children early in childhood, none of our subjects was above the 5th percentile for their typical peers by adolescence. These data are consistent with previous reports that show an apparent worsening or “falling off” of growth in children with CP over time when compared with typical children.25,27–29
These CP-specific growth curves could be published for clinical use, as other diagnosis-specific reference curves have been,23,24,30 and our group debated this issue pointedly. However, because the representative sample of children with CP used to develop the curves likely included children with confounding secondary conditions of acute and chronic malnutrition and growth hormone deficiency, we decided against the idea. We were concerned that clinicians might use these growth curves as “prescriptive” for the population. Therefore, we decided to embark on further analyses to link physical growth to health and participation outcomes. There is clear precedent for this notion in child health research, most notably in research related to obesity and cystic fibrosis.31–33 In addition, research in adults supports a relationship between underweight and excess mortality.34
To link physical growth with markers of health and participation, cluster analyses were chosen to extract the most information possible from the anthropometric data collected. Although we considered using simpler measures of proportion, such as weight for height or BMI, these children were so varied in body proportions that overall patterns of growth and body composition would be misrepresented or not captured entirely. Our intent was to take a broad “snapshot” of physical growth. Not unexpectedly, and as illustrated in Table 4, growth correlated with neurologic severity and the presence of a feeding gastrostomy. A large proportion of the smallest growth group (Z−) was made up of the most severely impaired children (GMFCS level V), particularly those without a gastrostomy. However, every level of severity was represented in each of the anthropometric groups. Thus, whereas neurologic severity accounted for a portion of the variance in outcomes, it was only when we controlled for severity (see Fig 7) that 2 of the analyses reached statistical significance.
The markers of health that we chose were modified from the National Health Information Survey by the CDC.10 Although based on questionnaire responses from parents, they address the construct of health and well-being through the reported use of health services and reported participation in school and usual activities. The results of our analyses were consistent in that the best growth correlated with the least use of health care services and fewest days missed of usual activities, even when neurologic severity was controlled. The consistency of these results suggests that a larger sample size would likely demonstrate more statistically significant results. However, health and participation are influenced by many different factors, only some of which may be modifiable, and even a larger cross-sectional sample would not confirm cause and effect relationships. Understanding the impact of growth on health and participation will require clinical trials in which growth is modified (eg, nutritionally and/or hormonally) and outcomes are carefully measured. Nevertheless, we feel that the current data are suggestive and worthy of further investigation.
What do these results mean, and where do we go from here? These data and analyses represent the next step in clarifying the relationship between growth and health in children with CP. However, these data must be viewed with caution. Although “better” health and “bigger” growth correlated strongly with one another, the direction of causality is unclear. Children may be healthier because they are bigger, or they could be bigger because they are in better health. Alternatively, because the health and participation questions refer to the 4 weeks before measurement, it is conceivable that children who had been ill (and missed school, etc) subsequently lost weight and were smaller. Further studies, such as those recently published by Sullivan et al,35,36 are essential to corroborate these findings and to evaluate whether specific interventions that improve growth result in improved health and social participation. The clinical goal is to optimize the health and well-being of children with CP and their families through appropriate management growth and nutrition.37
Children with moderate or severe CP have poor growth compared with typical children. We developed growth curves for children with moderate or severe CP and correlated growth with markers of health and social participation. Bigger children with CP had better health and social participation than similar smaller children. Further studies are needed to corroborate these findings and to evaluate whether specific interventions can improve growth, as well as health and social participation. Determining the potential role of these growth curves in clinical decision-making will require further study.
This study was conducted as part of the North American Growth in Cerebral Palsy Project. It was supported by the Kluge Research Fund and the Children's Hospital Committee of the University of Virginia, the Genentech Foundation for Growth and Development, the National Center for Medical Rehabilitation Research (grants 5 R01 HD35739-04, 1 F32 HD08615-01A1, 1 R24 HD39631-01, and K24-HD041504-01), and the University of Virginia General Clinical Research Center (M01RR00847). In addition, this study was supported by the University of North Carolina General Clinical Research Center (M01 RR00046) and the National Institute of Arthritis, Musculoskeletal and Skin Diseases (K24 AR02132); the General Clinical Research Center (M01 RR00240) and the Nutrition Center at the Children's Hospital of Philadelphia; the Jones-Guerrero Fund at Duke University, the Children's Hospital Fund at the University of British Columbia, and the General Clinical Research Center at the University of Rochester (M01 RR00044).
We gratefully acknowledge the help and support of the children and their families. We also give special thanks to Vivienne Spauls, Jillian Bumler, Teresa Olsen, and Candra Gerrick.
- Accepted March 31, 2006.
- Address correspondence to Richard D. Stevenson, MD, Department of Pediatrics, University of Virginia School of Medicine, Kluge Children's Rehabilitation Center and Research Institute, 2270 Ivy Rd, Charlottesville, VA 22903. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Cameron N. The methods of auxological anthropology. In: Falkner F, Tanner JM, eds. Human Growth: A Comprehensive Treatise. 2nd ed. New York, NY: Plenum Press;1986
- ↵Herman-Giddens ME, Slora EJ, Wasserman RC, et al. Secondary sexual characteristics and menses in young girls seen in office practice: a study from the pediatric research in office settings network. Pediatrics.1997;99 :505– 512
- ↵Worley G, Houlihan CM, Herman-Giddens ME, et al. Secondary sexual characteristics in children with cerebral palsy and moderate to severe motor impairment: a cross-sectional survey. Pediatrics.2002;110 :897– 902
- ↵National Health Interview Survey. National Center for Health Statistics 1997. Available at: www.cdc.gov/nchs/about/major/nhis/hisdesgn.htm. Accessed August 1, 2004
- Henderson RC, Lark RK, Gurka MJ, et al. Bone density and metabolism in children and adolescents with moderate to severe cerebral palsy. Pediatrics.2002;110 (1). Available at: www.pediatrics.org/cgi/content/full/110/1/e5
- ↵Stevenson RD, Conaway M, Barrington JW, Cuthill SL, Worley G, Henderson RC. Fracture rate in children with cerebral palsy. Pediatr Rehab.2006; In press
- ↵Gauss Mathematical and Statistical System [computer program]. Version 5.0. Maple Valley, Washington: Aptech Systems, Inc; 2002
- ↵The SAS System [computer program]. Version 9.1. Cary, North Carolina: SAS Institute Inc; 2004
- ↵Ogden CL, Kuczmarski RJ, Flegal KM, et al. Centers for Disease Control and Prevention 2000 growth charts for the United States: improvements to the 1977 National Center for Health Statistics version. Pediatrics.2002;109 ;45– 60
- ↵Johnson R, Wichern J. Applied Multivariat Statistical Analyses. 3rd ed. Upper Saddle River, NJ: Prentice Hall;1992
- ↵Grummer-Strawn LM, Garza C, Johnson CL. Childhood growth charts. Pediatrics.2002;109 :141– 142
- ↵Styles ME, Cole TJ, Dennis J, Preece MA. New cross sectional stature, weight, and head circumference references for Down's syndrome in the UK and Republic of Ireland. Arch Dis Child.2002;87 :104– 108
- ↵Myrelid A, Gustafsson J, Ollars B, Anneren G. Growth charts for Down's syndrome from birth to 18 years of age. Arch Dis Child.2002;87 :97– 103
- ↵Rempel GR, Colwell SO, Nelson RP. Growth in children with cerebral palsy fed via gastrostomy. Pediatrics.1988;82 :857– 862
- ↵Krebs NF, Jacobson MS, American Academy of Pediatrics Committee on Nutrition. Prevention of pediatric overweight and obesity. Pediatrics.2003;112 :424– 430
- ↵Sharma R, Florea VG, Bolger AP, et al. Wasting as an independent predictor of mortality in patients with cystic fibrosis [see comment]. Thorax.2001;56 :746– 750
- Copyright © 2006 by the American Academy of Pediatrics