A Multivariate Model of Determinants of Motor Change for Children With Cerebral Palsy.

Cerebral palsy refers to the neuromuscular deficit caused by a nonprogressive defect or lesion in single or multiple locations in the immature brain resulting in impaired motor function and sensory integrity.[1] The pathophysiological events may occur during the prenatal, intrapartum, perinatal, or early postnatal period. Cerebral palsy is the most common condition seen by pediatric physical therapists,[2] and it poses a challenge to practitioners due to the large variation in prognosis for motor function of children with this diagnosis.[3] In the context of goals and outcomes identified by the family, part of physical therapy intervention is aimed at assisting a child to perform tasks in a variety of environmental settings.[4] Maintaining a position against gravity and moving from one place to another are also a focus of intervention because these abilities contribute to activities of daily living and play.[5] Motor ability--the capacity to perform a movement--is also associated with employability, economic status, and social integration.[6]

Determining factors that influence outcomes is an area of physical therapy research. The purpose of this article is to describe the development and plans for testing of a multivariate model of determinants of motor change for children with cerebral palsy. A second purpose is to describe structural equation modeling as a method for testing multivariate models relevant to physical therapy.

Structural Equation Modeling

Development of our model was guided by structural equation modeling, a method for analysis of theory-based models in which relationships among many components of a system over time can be taken into account.[7,8] Structural equation modeling has been used extensively in the social and behavioral sciences for the past 2 decades and more recently in nursing research.[9-11] Structural equation modeling has been used in occupational therapy research,[12] but, to our knowledge, it has not been used in physical therapy.

Investigators interested in testing multivariate models using structural equation modeling should develop both a structural model and a measurement model. Briefly, a structural model is the hypothetical relationship among unobservable multidimensional components referred to as "constructs."[7,8] A measurement model details the multiple measures that will be used to estimate the constructs. Proponents of structural equation modeling advise researchers to use theory, prior research, and reasoning to first develop structural models.[7,8]

Our model was developed over a period of several years through an iterative process of identifying and evaluating compatible theoretical frameworks, appraising the literature, and, in the absence of evidence, generating hypothetical relationships. Constructs of a structural model are described as exogenous if they are thought to be independent of any other constructs in the model and as endogenous if they are viewed as being influenced by other constructs in the model.[7,8] The constructs of our model are child characteristics (primary impairments, secondary impairments, and personality), family ecology, and health care services. Child characteristics related to the primary impairments and personality characteristics are exogenous; the remaining constructs in our model are endogenous. By convention, the constructs of a structural model are depicted as ellipses, and presumed causal paths are represented by straight, single-headed arrows. The relative placement of the constructs in a figure and the direction of arrows provide an indication of the temporal relationships specified by the model. This constellation of ellipses and arrows is referred to as the structural model. Figure 1 presents our full structural model of determinants of motor change in children with cerebral palsy.

[Figure 1 ILLUSTRATION OMITTED]

Using structural equation modeling, once the structural model is developed, the measurement model is designed. This process involves specifying the indicators for each construct and estimating the reliability of measurements relating to each indicator. The constructs of the model cannot be measured; however, inferences can be made about a multidimensional construct through the measurement of several indicators. Indicators are variables that are used to quantify elements of an attribute, phenomenon, or event. Ideal indicators are measurements that: (1) are measured along a continuum (ie, not dichotomous or categorical), (2) are able to have sufficient intersubject variability, (3) are normally distributed, and (4) have a linear relationship to the dependent variable.[7] Each of the indicators within a construct should be scaled in the same direction.[7] For example, each variable of the construct of secondary impairment should be scaled such that higher values of each indicator are associated with greater degrees of secondary impairment, and vice versa. Potential indicators for our measurement model are listed in the Table.

Table. Constructs and Possible Indicators of the Model of Determinants of Motor Change for Children With Cerebral Palsy

Construct               Possible Indicators

Primary impairments
Motor                   Aberrations in muscle tone(a)
                        Type of motor disorder
                        Distribution of motor disorder
                        Primitive reflexes
                        Age of accquisition of motor
                        abilities

Sensory                 Visual impairment
                        Sensorimotor integration (vision,
                        somatosensation, vestibular)

Cognitive               Cognitive ability
                        Epilepsy

Secondary impairments   Muscle hypoextensibility,
                        Joint contracture
                        Skeletal malalignment
                        Impaired force production
                        Impaired endurance

Inherent child          Temperament
characteristics         Motivation
                        Risk taking

Family ecology          Family resources
                        Quality of home environment
                        Family support
                        Parental expectations
                        Family function

Health care services    Availability
                        Accessibility
                        Intervention options
                        Method of service delivery
                        Satisfaction with care

(a) Muscle tone is defined as "the force with which a muscle resists being lengthened, that is, its stiffness."[21(p577)]

Once both the structural and measurement models have been developed, data from all of the selected indicators are collected on a sample--preferably a very large sample--of people with the condition of interest in the order and timing defined by the structural model. For example, data on characteristics relating to the primary impairments and child personality characteristics might be collected at the time of the initial assessment. Motor abilities would also be measured at that time. Data on secondary impairments, family ecology (eg, family resources, support, expectations, and function), and health care services, for example, might be collected 6 months later, and basic motor abilities might be re-evaluated again 1 year after the initial assessment.

Analysis is conducted in several steps. Standardized regression coefficients are calculated for each of the pathways in the model. These standardized coefficients quantify the strength of the relationship between the corresponding constructs, in the context of the whole model. The magnitude of each coefficient is interpreted similarly to a correlation coefficient. The magnitudes of the coefficients, therefore, assist with interpretation of paths that have different levels of predictive or explanatory power. In addition to estimation of the strength of effect of each of the specified paths, most structural equation modeling statistical packages will provide a description of problem areas in the model relating to negative error variances (ie, error variances with negative values), standardized coefficients exceeding 1.0, colinearity, the pattern of standardized residuals, and misspecification of causal paths.[7] This information can be used to modify the model so that the specified relationships more closely match the data. A final step in the analysis involves testing the "goodness of fit" between the data collected and the model as specified.[7,8] A variety of chi-square procedures are available to test the match between information collected and the structural and measurement models.[7,8]

A Model of Determinants of Motor Change for Children With Cerebral Palsy

In this section, we describe the phases of development of the structural model. The description of this process of development of the relationships among the constructs is complemented by Figures 1 through 4. Within each phase, we review evidence to support the constructs in the structural model and to identify potential indicators of the measurement model. Because the literature on motor change among children with cerebral palsy is limited, we also review related literature on motor development of infants without motor impairment as well as determinants of multiple developmental outcomes among children who are at risk for developmental delays.

[Figures 2-4 ILLUSTRATION OMITTED]

Phase 1: Model Guided by the Disablement Process

We used the process of disablement, defined by the National Center for Medical Rehabilitation Research,[13] as an initial guide to model development. More recently, we used the aspects of human functioning described by the International Classification of Impairments, Activities, and Participation (ICIDH-2)[14] in model development (Fig. 2). The ICIDH-2 incorporates biological and social perspectives of disablement to represent the impact of health conditions on an individual's life, including participation in society. The focus of our model is on the relationships among impairment, activity, and participation. In our model, the constructs of primary impairments and secondary impairments are directly related to the impairment dimension of the ICIDH-2 framework. Impairment occurs within a body system. In our model, we propose that neuromuscular, musculoskeletal, cardiopulmonary, sensory, and cognitive impairments are determinants of change in motor ability of children with cerebral palsy. Unlike the World Health Organization (WHO) framework,[14] we divide impairments into 2 categories: primary and secondary. A primary impairment is one that we conceptualize as being present at the onset of a condition (and, therefore, as being an exogenous construct), and a secondary impairment is one we view as arising over time as a result of a primary impairment (and, therefore, as being an endogenous construct influenced by the primary impairment). Secondary impairments have also been referred to as secondary conditions[15] and are a focus of physical therapists in tertiary prevention.[16]

The onset of cerebral palsy frequently occurs long before the diagnosis, complicating attempts to differentiate primary and secondary impairments. For example, although co-contraction has been identified as a primary impairment,[17] recent work on the development of postural adjustment in sitting suggests that co-contraction might develop as a secondary compensatory mechanism to inadequate postural control.[18] Ideally, identification of primary impairments would occur at the time the condition starts, but difficulties associated with early diagnosis often make this impossible. From a pragmatic viewpoint, we contend it might be useful to identify those impairments observed at the time of diagnosis as primary and those evolving at some point after the diagnosis as secondary. In addition, although motor impairments comprise key primary impairments, comorbidities associated with the central nervous system lesion such as sensory deficits, cognitive impairments, and epilepsy might, in our opinion, also influence future motor function.

We hypothesize that primary impairments influence motor abilities both directly and through a causal path associated with secondary impairments. Primary and secondary impairments are, in our opinion, relatively major and moderate influences on motor change, as indicated by the different thicknesses of the arrows. Change in basic motor abilities over time is, for us, the outcome of interest in the proposed model.

Basic motor abilities can be measured using the Gross Motor Function Measure (GMFM).[19,20] The GMFM provides reliable and valid criterion-referenced measurements that measure the motor function of children with cerebral palsy over time.[19] Reliability of GMFM scores was initially demonstrated with trained physical therapist raters on a sample of 12 children with cerebral palsy representing a spectrum of ages and severity of involvement. Intraclass correlation coefficients were .99 for both interrater reliability and test-retest reliability within a 2-week period.[20] Validation of the responsiveness of the GMFM was established by demonstrating that change in GMFM scores over 4 to 6 months among samples of children with cerebral palsy (n = 111), children recovering from acute brain injury (n=25), and children under 5 years of age with no known physical disabilities (n=34) were correlated with judgments of change by parents (r=.54), physical therapists (r=.65), and masked evaluators (r =.82).[19] Furthermore an interaction between age and severity of involvement was detected for the amount of change in the group of children with cerebral palsy; younger children who were judged to have mild involvement demonstrated greater change than older children who were judged to have moderate or severe involvement.[19] For children with no known motor disabilities, motor change was greater among those under 3 years of age compared with those aged 3 years and older. Finally, among children judged to be responsive, children who were recovering from brain injury demonstrated greater change than children with cerebral palsy.[19] Unpublished research by Palisano and colleagues suggests that GMFM scores of children with cerebral palsy plateau by 6 or 7 years of age. The age group of interest in testing our proposed model and in evaluating the research evidence, therefore, involves children less than 7 years of age because of the use of the GMFM as a measure of basic motor ability.

Basic motor ability, as defined in our model, is related to activity. The WHO defines activity as the performance of a task or action by an individual.[14] For children with cerebral palsy, this might refer to activities of daily living and play. Participation is defined by the WHO as an individual's involvement in life situations[14]; attendance at preschool or school are examples of life situations experienced by children. As indicated by the dashed ovals and arrows in Figures 1 through 4, the constructs of activity and participation are not part of the proposed model but are included to provide a perspective on how motor ability relates to activity and participation.

Evidence Supporting the Construct of Primary Impairments

Motor impairment. The movement disorder of children with cerebral palsy includes one or more of the following impairments: delay in movement onset, poor timing of force generation, poor force production, inability to maintain antigravity postural control, decreased speed of movement, and increased co-contraction.[17] The relationship between these impairments and the acquisition of basic motor abilities has not been investigated. The focus of most research on children with cerebral palsy has been on aberrations in muscle tone, type and distribution of the motor disorder, primitive reflexes, and prediction of walking from age of acquisition of basic motor abilities.

In most of the cited literature, the concept of muscle tone is captured by the following definition: "the force with which a muscle resists being lengthened, that is, its stiffness."[21(p577)] As such, muscle tone covers a spectrum from low to high. Distribution refers to the area of involvement in the body. For example, hemiplegia is ipsilateral involvement of the upper and lower extremities, diplegia is bilateral lower-extremity involvement, and quadriplegia is involvement of all extremities and the trunk.[22]

Spasticity has been attributed as a major contributor to movement dysfunction among children with cerebral palsy.[22] Spasticity is defined as a velocity-dependent increase in the monosynaptic reflex.[23] It is differentiated from muscle tone in that it covers only the high end of the spectrum of tone and is tested by elongating muscle groups rapidly.[23] Although spasticity does contribute to movement problems, recent experience with ablative procedures such as selective dorsal rhizotomy reveals that muscle weakness[24] and other motor control deficits relating to poor use and timing of muscle activity[25] persist following surgery.

Type and distribution of motor involvement are among the main descriptors and prognostic indicators of motor ability relating to cerebral palsy. Most of the investigations in the literature reviewed in this paragraph focus on children with spastic cerebral palsy and relate to some form of walking as the motor outcome of interest. Investigations of the distribution of motor involvement have provided more consistent results regarding eventual prognosis for motor function than the type of motor disorder. For example, children with hemiplegia--in the absence of major comorbidities--appear to acquire motor abilities such as independent sitting at the expected (or slightly later than expected) time,[26] and they walk independently during the preschool years.[27-30] Among those with bilateral involvement, children with diplegia have a much better prognosis for walking than children with quadriplegia.[26-28,30-32] Children with dyskinetic involvement--particularly those with athetosis--have been reported to have more varied outcomes than children with dystonia, whereas children with ataxia have been reported to have a good prognosis for walking.[29] The lack of clear specification of outcomes using the prognostic indicators of type and distribution of the motor disorder is due, in part, to the lack of independence of categorization between the two variables, the lack of stability of classification as the children mature, and the lack of a standardized definition of type and distribution.[30]

Several investigators[27-30,32] have examined the relationship between primitive reflexes and motor ability. The presence of primitive reflexes beyond 18 to 24 months of age has been found to be associated with a lower probability of ambulation[27-30,32] and, therefore, could be associated with the acquisition of other motor abilities. The assumption that persistent primitive reflexes cause delayed motor development has been an integral part of past physical therapy interventions (eg, Bobath technique[22]). In contrast to the view that persistent reflexes impede motor development, Fetters[33] has proposed that primitive reflexes are species-specific movement patterns that may be adapted to accomplish functional movements in the absence of alternative patterns. We concur with other authors[26,34] who have speculated that the persistence of primitive reflexes may be markers indicating the severity of involvement rather than the cause of limitations in motor function.

Investigations of prognostic factors in cerebral palsy have also focused on the age of acquisition of motor abilities as a proxy for severity of involvement. An earlier age of acquisition of independent sitting--and specifically sitting independently by 2 years of age--has been reported to be associated with a greater probability of later success in both pre-walking and walking abilities.[26,27,29-31] Between 90% and 100% of children who sit independently by the age of 2 years will later walk independently.[29-31] In contrast, the acquisition of independent sitting between 2 and 3 years has been reported to be associated with a 50% probability of walking.[29] Although crawling on hands and knees has been identified as a predictor of ambulation in a univariate analysis,[35] the ability to sit independently--and not the ability to crawl--was found to predict walking at 6 years in a multivariate analysis.[32]

Sensory impairments. Children with cerebral palsy who have visual impairments have a poorer prognosis for ambulation than children with cerebral palsy who do not have visual impairment.[30] Furthermore, children with complete blindness are less likely to walk than those with milder visual impairments.[32] When multivariate analyses were conducted, however, the effect of visual impairment did not remain significant.[32] Specifically, once the effects associated with variables such as distribution of involvement, presence of Moro and asymmetric tonic neck reflexes, and epilepsy were accounted for statistically, visual impairment did not remain a significant predictor. Thus, although infants with no physical impairments who were blind from birth have been identified as having delayed self-initiated movement and locomotion,[36,37] the contribution of visual impairment alone to the acquisition of motor abilities among children with cerebral palsy has not been demonstrated. The covariation of visual impairment and other comorbidities among children with cerebral palsy makes it difficult to differentiate the effects of any one variable.

Complex visual-somatosensory[38,39] and visual-vestibular[40] interactions contribute to the development of posture in infants who are developing typically. Nashner and colleagues[41] investigated the postural control of a group of children with mild involvement due to cerebral palsy during external perturbations and self-initiated arm movements under different conflict conditions of visual, vestibular, and somatosensory feedback. They reported

that children with spastic hemiplegia had abnormalities in muscle coordination but exhibited no difficulties with sensory conflicts. Conversely, children with ataxia had difficulties under sensory conflict conditions but coordinated their muscles typically. These findings suggest that children with various types of cerebral palsy differ in their ability to effectively use combinations of visual, vestibular, and somatosensory information to maintain postural control. However, the role of sensory information has not been investigated with respect to acquisition of motor abilities.

Cognitive impairments. The contribution of cognitive ability to motor prognosis is unclear. Although cognitive ability has not been reported to be a primary determinant of ambulation,[42] other researchers have noted the poor outcomes for children with microcephaly[32] and very low scores on intelligence tests,[30] indicating that there may be a threshold effect. Specifically, there may be a relationship between cognition and motor ability only among children with impairments in cognition. A difference in age of walking has been reported among 336 children with varying degrees of mental retardation who lived in residential care facilities.[43] All children with mental retardation described as trainable walked before the age of 4 years. In contrast, all children who walked only after the age of 7 years had profound mental retardation. We believe that in order to have an influence on the attainment of basic motor abilities such as walking, the degree of cognitive impairment may have to be severe. For children with severe or profound mental retardation who do not have a motor disability, the primary concern might not be the ability to move, but rather safety and performance of goal-directed actions while walking.

Epilepsy has been reported to be associated with intelligence level among children with cerebral palsy.[27] Children with both cerebral palsy and epilepsy have been reported to have a poorer prognosis for ambulation than children with only cerebral palsy.[28,30,32,42] Trahan and Marcoux[32] have described and association between epilepsy and the ability to ambulate while accounting for other factors such as the type of cerebral palsy, the inability to sit, and the presence of primitive reflexes.

Evidence Supporting the Construct of Secondary Impairments

Impaired range of motion, joint mobility, and skeletal alignment. Individuals with chronic spasticity have been observed to develop secondary muscle hypoextensibility as a result of a decrease in the number of sarcomeres,[44] impaired muscle growth,[45] and stiffennig of the parallel elastic structures.[46] The reduction of movement associated with muscle hypoextensibility may contribute to joint contractures through changes in the joint capsule and other connective tissues.[46] Furthermore, interactions among musculoskeletal growth, postural asymmetries, and reduced amount and variability of movement patterns may contribute to deformation of bone.[47]

Muscle contractures appear to contribute to the variations in motor behavior of children with spastic cerebral palsy. For example, equinus gait may sometimes be attributable to a contracture in the triceps surae muscle, to excessive contraction of the triceps surae, muscle, or to both a contracture and excessive contractions.[48] Joint contractures may also interface with the acquisition of motor abilities. Children with severe spastic diplegia can walk with crutches with hip flexion contractures of up to 40 degrees.[42] Knee flexion contractures of even a mild degree, however, are reported to be incompatible with the maintenance of walking.[42] Altered body mechanics associated with gait abnormalities have been seen as precursors to long-term degenerative changes such as overstretching of the patellar tendon, patella alta, chondromalacia patellae,[49] and arthritis.[50] Furthermore, adults with cerebral palsy have reported pain from musculoskeletal impairments such as repetitive motion disorders, arthritis, and bursitis.[51]

Impaired force production. Children with cerebral palsy exhibit reduced muscle force compared with children without motor impairments.[52,53] Although poor force production has been identified as a primary movement disorder in some cases of cerebral palsy,[17] secondary changes in muscle properties associated with the muscle fibers[46] and length-tension relationship[54] may also contribute to what appear to be deficits in force production. The mechanism of impaired force production among children with cerebral palsy has not been clarified to date.

Muscle force of the knee extensors has been found to be related to motor functions such as walking, running, and jumping in children with cerebral palsy; greater muscle force was associated with greater ability levels.[55] Training designed to enhance force production may reduce the degree of crouch in children with spastic diplegia[56] and may contribute to improvement in gross motor abilities of children with cerebral palsy who have minimal motor impairments.[57]

Impaired aerobic capacity and endurance. Secondary impairments may also be associated with cardiovascular fitness and physiological efficiency. The Physiological Cost Index (PCI)[58]--also known as the Energy Cost Index[59]--has been used to demonstrate poorer efficiency among children with cerebral palsy.[60] The PCI is easily administered in a clinical or community facility with a free walking space of at least 25 m, a portable heart rate monitor, and a stopwatch. It is calculated by dividing the difference in heart rate (in beats per minute) between steady-state walking and resting conditions by the speed of walking (in meters per minute); final measurements are expressed as beats per meter. The PCI values were 48% higher than normal in a group of adolescents with cerebral palsy with minimal motor impairments.[55] The structural changes in muscle of children with spasticity might contribute to metabolic inefficiency.[61] Olney[62] has speculated that increased stiffness contributes to a high energy cost of movement in children with spastic cerebral palsy. For example, according to this theory, muscle with very stiff parallel elastic structures may require more energy input to obtain the same degree of active range of motion as muscles with normal parallel elastic structures. Over the longer term, a reduced level of efficiency may evolve. Poor cardiovascular fitness in adults with chronic disabilities has been associated with poor endurance.[51] Reduced endurance in childhood, in our opinion, may manifest as reduced opportunities for independence, including interaction with other children through play and recreaction.

Phase 2: Incorporation of Aspects of the Child Unrelated to the Disability

Physical therapists usually consider aspects of a child's primary and secondary musculoskeletal and cardiopulmonary impairments through the examination, evaluation, diagnosis, prognosis, and intervention phases of care.[16] Less frequently, however, are the child's personality characteristics incorporated into the process. In our proposed model, personality characteristics are defined as the individual's set of customary ways of reacting or behaving.[63] These inherent characteristics are attributes that are conceptualized as being largely independent of the impairments; that is, they are associated with children and not necessarily with cerebral palsy. They are generally (although not always) stable within individuals over time while also being variable across individuals. These characteristics are considered to be exogenous, and we view them as influencing change in motor abilities through their effect on secondary impairments (Fig. 3). For example, we believe an active child can make greater gains in motor ability than a passive child, in part due to the effect of physical activity on range of motion, force production, and endurance. Those personal attributes that we believe enhance motor outcomes, that is, those factors that contribute to an individual's resiliency and adaptability, we believe should be of interest to therapists.[64]

Evidence Supporting the Construct of Child Personality Characteristics

Temperament. The role of temperament in developmental outcome was addressed by Werner[64] in her longitudinal study of the outcomes of infants exposed to perinatal trauma and chronic poverty on the island of Kauai. Although she did not investigate motor outcomes, the main "resiliency factors" that she identified as being related to optimal psychosocial outcomes were attributes related to temperament that elicited positive responses from caregivers and other adults. Specifically, children with optimal outcomes were described as active, affectionate, cuddly, good-natured, and easy to deal with. Any relationship between temperament and motor development, either among children developing typically or among children with cerebral palsy has not been documented.

Motivation. Thelen and Smith[65] have suggested that motivation is a key determinant of developmental change. They suggested that infants come into the world with a rich set of adaptive predispositions that serve to motivate them to seek nourishment, contact, and warmth; in essence, to seek out pleasure and avoid discomfort. As with other personal characteristics, motivation to explore the environment varies among infants and young children. In our opinion, young children who actively explore their environments--and take pleasure from exploration--may acquire motor skills more readily than children who are content to explore visually.

A special form of motivation is "mastery motivation," which is defined as "a psychological force that originates without the need for extrinsic reward and leads an infant or young child to attempt to master tasks for the intrinsic feeling of efficacy rather than because of recurrent reward."[66(p6)] "Moderately challenging tasks" are associated with mastery motivation. Implicit in the definition is that the child attempts the task independently, in a focused and persistent manner, in order to solve a problem or master a skill or task that is at least moderately challenging. Farther along this spectrum are the children who may regularly exhibit risk-taking behaviors. Risk-taking behaviors are motor behaviors that have an element of physical danger.[67] Infants who attempt to assume an erect position and take steps would be viewed as taking risks if they do not have the postural control to maintain stable upright stance. Cintas[67] reported that among children aged 12 to 24 months, those with higher levels of motor development demonstrated increased risk-taking behavior. Two competing theories are possible. First, these results may have occurred because children with higher motor skill levels take greater risks. Alternatively, children who take greater risks may develop higher levels of motor skills.

Phase 3: Incorporation of Theories of General Systems and Human Ecology

The ICIDH-2 framework describes human functioning as occurring in the context of environmental factors, which are defined as external influences.[14] Systems theory also suggests that change occurs in a specific context as a result of the interaction of many elements in both the person and environment.[68] Based on this perspective, we believe physical therapists should examine how the musculoskeletal, neuromuscular, cardiopulmonary, sensory, cognitive, and emotional systems interact to either enhance or restrict motor behavior within a specific context. Furthermore, therapists are encouraged to consider psychosocial aspects of the child (eg, in phase 2), as well as physical and social aspects of the environment in which the child lives. This perspective is compatible with Bronfenbrenner's[69] conceptualization of the important role of the family in shaping child development and with Sameroff and Chandler's[70] theory of the transactional relationships among family functioning and child

health and developmental outcomes. Scrutton and Rosenbaum[26] stated that, although neuromuscular and perhaps cognitive impairments have a major influence on the locomotor development of children with cerebral palsy, other factors, including the child's physical and social environment, might influence how movements develop. Bower and McLellan[71] have also suggested that expectations and experiences might influence the acquisition of motor abilities among children with cerebral palsy.

The construct of family ecology encompasses both family demographics (eg, parent age, education, and income) and the dynamics of family function (eg, coping with stress) within the home. We view family ecology as an endogenous construct (Fig. 4) because the way a family functions is partially a result of personality characteristics of each child in the family. In our model, only the characteristics of the child with cerebral palsy are modeled. We believe parents' interaction, expectations, resources, and supports (eg, within the family, from the extended family, from the community) might influence the acquisition of motor abilities among children with cerebral palsy through the variations in the experiences and opportunities a family is able to provide.

Evidence Supporting the Construct of Family Ecology

Family demographics. Demographics potentially influence the family environment. However, variables such as marital status,[72] parental education,[73] socioeconomic status,[72] and birth order[73] have been found to be unrelated to motor development in the population at large. In contrast, children growing up in poverty have been described as being exposed to a "double jeopardy": they are more likely to have global developmental problems, and the manifestation of these problems tends to be more severe.[74] Although poverty has been found to influence cognitive function more than motor function among children at risk for a variety of adverse developmental outcomes,[75] recent evidence suggests a relationship between poverty and motor development. Specifically, poverty was found to have a greater impact than in utero cocaine exposure on motor development.[76] Poverty may limit the resources a family has to provide opportunities for a variety of play and exploration experiences, thus impeding optimal motor development. Alternatively, the demographic characteristic of low income level may be a marker of the dynamics of family function if families with low incomes have poorer family function than families with high incomes. In support of the idea that income might be a marker of family function, families living in poverty are more likely to report increased stress, diminished social support, and depression as compared with families who do not live in poverty.[74] Similarly, adolescent mothers have been found to be less responsive,[77,78] less involved with their children,[77] and less likely to provide appropriate play materials than older mothers.[78] They are also more likely to have infants who attain lower scores on measures of cognitive development while also demonstrating a trend toward lower scores on motor function.[77,78] Thus, the impact of some of the nonsignificant demographic variables on the acquisition of motor abilities may operate through processes associated with family dynamics. Specifically, socioeconomic status and maternal age might not directly influence the acquisition of motor abilities. Rather, families (ie, resources, social supports, and family functioning) might be important determinants of motor outcome.

Dynamics of family functioning. A supportive parent-infant relationship has been shown to promote early development.[79] Qualities of the home environment such as parental responsivity, parental involvement, and provision of learning materials provide an indication of the support that a family offers to general development and exploration.[77,78] A detailed review of caregiving practices and parental expectations suggests that a parent's expectations of the infant's performance influence the acquisition of early motor abilities.[80] Aggressive infant handling used by many African-American caregivers includes tossing in the air,[81,82] bouncing vigorously,[82] and supporting infants in sitting and standing when awake.[81,83] These handling practices stimulate postural adjustments and may contribute to an accelerated rate of motor development observed in these caregivers' infants.[81,84,85] In contrast, infants of Asian-American parents are frequently calmed and nursed in a supine position instead of prone and upright positions.[86] These infants also display little spontaneous movement early in life and exhibit generalized low muscle tone.[87]

The support a family is able to provide their infant or young child is also dependent on social support beyond the family unit. Sources of social support include relatives, friends, neighbors, churches, schools, and parents' place of work. The number of sources of social support available to a family has been associated with the number of games parents played with their children and the magnitude of cognitive gains made over a 1-year period among families with children participating in early intervention programs.[88] The role of social support in the acquisition of motor abilities has not been investigated. Parents with a range of social supports, however, play a great variety of games with their children, which suggests to us that the parents may provide greater opportunities for their children to develop. Thus, family dynamics, in our view, determine the support families are able to provide their developing infants and young children. This support is mediated by complex relationships among the parents' perceived stresses, the extent of their personal supports, and the coping strategies they use to deal with the stresses in the context of societal supports.[89]

Phase 4: Incorporation of Health Care Services in the Context of the Philosophy of Family-Centered Care

Health care services relate to the availability and accessibility of care, intervention options, methods of service delivery, and satisfaction with care. We believe this construct is endogenous (Fig. 1, full model). Specifically, we hypothesize that the type and extent of the child's primary impairments have a moderate influence on the services provided by rehabilitation practitioners to the child with cerebral palsy and his or her family. To a lesser extent, we contend that the child's personality characteristics might contribute to decisions for care. In our view, the nature of secondary impairments affects services, which, in turn, should have an impact on the secondary impairments over time. In Figure 1, bidirectional arrows have been placed between the constructs of health care services and family ecology, in keeping with the philosophy of family-centered care.[90] A guiding principle of family-centered care is that decisions about goals, intervention approaches, and outcomes are made collaboratively with families.[90] Correspondingly, the effect of services on change in basic motor abilities is modeled through family ecology.

Although outside the main structural model we offer, we contend that activity and participation might also be influenced through pathways other than basic motor abilities. For example, family support--either directly or through interaction with health care services--may influence both activity and participation. These alternate pathways might be particularly influential for children with severe neurological impairments. Specifically, these children might be able to participate and experience success and enjoyment through assistive technologies that do not rely on motor abilities.

Support for Incorporation of the Construct of Health Care Services

Reduced payment for health care and constraints in public funding for early intervention and school-related therapy services have created challenges for those who provide services to children with cerebral palsy and their families. Service providers are increasingly being challenged not only to offer predictors of outcomes to families and payers but also to demonstrate effectiveness and evidence that they are providing high-quality care with desirable cost/benefit ratios.[91] In the United States, the transition from traditional fee-for-service plans to managed health care plans appears to have influenced the availability and accessibility of services for children with chronic conditions such as cerebral palsy. In a survey of 33 parents of children with special health care needs, respondents reported overall satisfaction with services available in their children's health care plans.[92] Specifically, between 64% and 89% of the parents reported good to excellent satisfaction for 8 out of 12 items (3 on access, 4 on coordination and information, and 1 on time for approval of services). Access to a care coordinator and coordination among health care professionals, however, were areas of dissatisfaction. Forty-eight percent of parents who were surveyed reported that one of more services (including physical therapy) were not covered by their health plan and frequently were too expensive to cover privately.

Limited information exists about the efficacy of physical therapy services for children with cerebral palsy. In a 10-year-old review of research, Piper[3] concluded that in order to be efficacious in promoting motor milestone attainment, physical therapy must be provided a minimum of 2 times per week. In contrast, Parette and associates[93] reviewed studies that addressed the intensity of occupational therapy and physical therapy for infants and young children with cerebral palsy and concluded that there is a lack of consensus regarding intensity of therapy. Their review, however, was published 9 years ago. The state of Oregon recently developed guidelines for outpatient pediatric services that emphasize periodic and episodic use of physical therapy intervention based on specific functional limitations.[94] This represents a departure from providing ongoing rehabilitation services to children with developmental disabilities based on their diagnoses or type and severity of their impairments.

The shift from child-centered approaches to care to family-centered approaches to care has expanded the role of the physical therapist to include interventions other than individual direct treatment.[95] Family-centered care includes addressing the concerns and goals identified by children and families, providing intervention in the settings in which children normally live and play, and determining outcomes that are important for the child's function at home, at school, and in the community.[96] Family-centered care is consistent with interventions identified in the Guide to Physical Therapist Practice[16]: coordination of care, communication, and patient instruction. Alternatives to direct individual therapy such as large-group programs, small-group programs, and consultation also have been advocated by one of us (RJP).[97] The effectiveness of these interventions for children with cerebral palsy has not been investigated.

Child and family satisfaction with care are outcomes identified in the Guide to Physical Therapist Practice.[16] Satisfaction with care, in our opinion, is an indicator of the success of interactions among health care providers, the child, and family that are directed at improving the child's motor function. Cramer and Tucker[98] have termed this interaction between health care provider and consumer as a "patient-provider partnership." King and colleagues[99] reviewed research that examined the relationship between interpersonal aspects of care and client satisfaction. Information exchange, respectful and supportive care, and partnership and enabling were positively related to client satisfaction. Assumptions about the relationship between child and family satisfaction with care and change in motor function have not been studied.[100]

An Approach to Examining the Model

In this article, we describe the development of a structural model of determinants of change through the application of theoretical frameworks. We contend that 5 constructs--primary impairments, secondary impairments, child personality characteristics, family ecology, and health care services--interact to explain and predict the acquisition of basic motor abilities among children with cerebral palsy. The next step in clarifying determinants of motor change for children with cerebral palsy is to select variables to measure for each construct; we contend that the selected variables should be important to service providers and the clients they serve, feasible to collect, and have many of the characteristics described earlier in this article.[7,8] Once the indicators are selected, the measurement model can be specified.

Due to the paucity of information in the research literature, selection of the most important variables for each construct in our model will be examined by use of a 2-stage consensus process involving physical therapists who work with preschool and early school-aged children with cerebral palsy in the Ontario Association for Children's Rehabilitation Services (OACRS) centers. During the first stage, the nominal group process[101] will be used to generate ideas and to come to consensus within each participating center on items relating to each of the constructs. In the second stage, a mailed survey to the same participants will be used to establish consensus across centers about items that are important and feasible to test in clinical practice. The top-ranked variables in each of the constructs will be selected for further investigation. Thus, indicators believed to be important in influencing motor outcomes will be included in the measurement model. Indicators not supported through the consensus process will be discarded unless strong research evidence exists or a theoretical argument can be made to include them in the measurement model.

Before collecting longitudinal data on the selected variables, tools that yield reliable and valid measurements will need to be identified or developed. Measurements of secondary musculoskeletal and cardiopulmonary impairments fall within the scope of traditional physical therapy; however, not all measurements have reliability when applied to children with cerebral palsy. For example, goniometric measurements of range of motion on this group of children have not been reproducible.[102,103] Strategies to optimize interrater reliability for all measures will need to be developed. Although not explicitly considered by physical therapists in the past, measurement of variables such as a child's temperament and motivation, family function, and perception of processes of care, in addition to measures of primary impairments, might provide useful information for identifying important prognostic variables and thus planning specific therapeutic programs for children with different prognostic attributes.

When all of the variables have been identified and the measurement model has been specified, a plan for longitudinal data collection on a large and representative sample can be implemented. This step will be facilitated through the OACRS. This association consists of 19 children's rehabilitation centers that serve most of the children with a diagnosis of cerebral palsy in the province of Ontario, Canada. Although formal sample size calculations are possible, general guidelines suggest a minimum of 200 subjects[104]; however, larger sample sizes are preferred.[105]

Relevance of Model Testing in Physical Therapy

The process of planning physical therapy services for patients and clients is complicated when individual attributes or environmental conditions influence outcomes. We contend that multiple influences are more likely to affect outcomes among people with multisystem and chronic diseases and disabilities than among people with single-system problems of short duration. As is true for all areas of health care, these prognostic factors, or determinants, must be clarified before effective interventions can be discovered for specific subgroups of clients.[106]

Although randomized controlled trials are considered the "gold standard" for determining the effectiveness of interventions, the multiple factors influencing outcomes for children with cerebral palsy are difficult to incorporate into an experimental study.[107] We believe that a research project using an observational design based on a model has the potential to identify important prognostic factors relating to differences among children with cerebral palsy, to examine these children's environments, and to determine differences due to interventions. Observational designs are nonexperimental; that is, investigators do not manipulate an independent variable, but instead make observations about events and interactions in a natural setting.[108] Cause-and-effect relationships can be examined, but not substantiated, using observational designs through prospective data collection, multivariate analyses of potentially influential variables, and careful interpretation of results in the context of eliminating plausible alternative explanations, Furthermore, according to Gordis,[106] observational designs have the potential to yield more generalizable results than experimental approaches. Although we have elaborated on the specification and testing of a model of determinants of motor change for children with cerebral palsy, we believe that this type of observational approach is suitable--and preferable--for other complex and chronic diagnostic or functional groups.

The approach we have outlined permits testing of a complementary set of theories specified by a model. By using structural equation modeling, investigators are provided with information about problem areas that can suggest modification of the model to improve the overall "goodness of fit." Thus, although Hayduk[7] strongly advised investigators to start with a theoretical framework, he encouraged us to use the pattern of data to guide model revision. For example, detailed feedback from the statistical output (as described earlier in this article) might suggest that several of the indicators of the secondary impairments are highly correlated (ie, have a high value for colinearity). In such a case, reduction of the number of indicators to include only the most influential indicators could lead to a simpler and more parsimonious model. Furthermore, an investigator can compare the "goodness of fit" of competing models to determine the model that best explains or predicts the phenomenon or outcome of interest.[7,8] For example, we could test the overall fit of our model as shown in Figure 2 (guided only by the disablement process) with the full model in Figure 1 (guided by a combination of compatible theories). Other models can be tested with different configurations of presumed causal paths. For example, we might want to compare a model specifying a path from child personality directly with motor change and compare the "goodness of fit" with the model specifying the path through secondary impairments.

Implications for Practice

We believe that decisions about intervention options and methods of service delivery for children with cerebral palsy, and those with other complex and chronic conditions, should logically occur in the context of knowledge about the relationships among the primary and secondary impairments, personality characteristics, environmental supports and barriers, and the outcome of interest (which, in our case, is change in basic motor abilities). Such decisions can be facilitated through the application of research results to individual decision making. If standardized regression coefficients are used to describe the relationship among constructs specified by a model, the relative magnitude of one potentially causal path to another can be interpreted in the context of the whole model. The data from this aspect of testing can be useful in predicting outcomes for individuals who were not a part of the sample used to test the model's ability to explain the outcome of interest. Clinicians can use the estimates of the strength of the relationship among constructs to generalize beyond the specific situation of a study to make predictions about what should happen in similar situations.[109] If factors that contribute to change in basic motor abilities are identified and if these factors are found to be amenable to manipulation, intervention to optimize the identified determinants might improve long-term outcomes. As an example, in our model, if the standardized regression coefficient between the constructs of secondary impairments and change in basic motor abilities is large, this finding would suggest to us that prevention of secondary impairments needs to be a major focus of intervention aimed at enhancing motor abilities.

Speaking more broadly, research results from model testing provide physical therapists with information to assist with the determination of prognosis, establishment of realistic and attainable goals, and direction for interventions. As illustrated by the model we offer of motor change in children with cerebral palsy, practitioners may have support for discussions with parents about their child's prognosis for motor function, a concern uppermost on the minds of parents at the time of diagnosis. Crothers and Paine[27] stressed the importance of communicating openly and honestly with parents about their primary concern and emphasized that overoptimism is dangerous and that unjustified pessimism is worse. Clearly, such improved communication will be attainable only through knowledge obtained from a large representative sample of children with cerebral palsy and their families, with consideration given to the multiple complex relationships among influential variables. Furthermore, the knowledge gained through testing of our model will assist families and health care practitioners in establishing goals and outcomes of intervention.

References

[1] Scherzer AL, Tscharnuter I. Early Diagnosis and Therapy in Cerebral Palsy: A Primer on Infant Developmental Problems. 2nd ed. New York, NY: Marcel Dekker Inc; 1990.

[2] Wilson JM. Cerebral palsy. In: Campbell SK, ed. Pediatric Neurologic Physical Therapy. 2nd ed. New York, NY: Churchill Livingstone Inc; 1991:301-360.

[3] Piper MC. Efficacy of physical therapy: rate of motor development in children with cerebral palsy. Pediatric Physical Therapy. 1990;2:126-130.

[4] Harris SR. Functional abilities in context. In: Lister MJ, ed. Contemporary Management of Motor Control Problems. Proceedings of the II Step Conference. Alexandria. Va: Foundation for Physical Therapy; 1991:253-259.

[5] Campbell SK. The child's development of functional movement. In: Campbell SK, ed. Physical Therapy for Children. Philadelphia, Pa: WB Saunders Co; 1994:3-37.

[6] Klapper ZS, Birch HG. The relation of childhood characteristics to outcome in young adults with cerebral palsy. Dev Med Child Neurol. 1966;8:645-656.

[7] Hayduk LA. Structural Equation Modeling With LISREL: Essentials and Advances. Baltimore, Md: The Johns Hopkins University Press; 1987.

[8] Hoyle RH. Structural Equation Modeling: Concepts, Issues, and Applications. Newbury Park, Calif: Sage Publications Inc; 1995.

[9] Campbell JC, Soeken KL. Women's responses to battering: a test of the model. Res Nurs Health. 1999;22:49-58.

[10] Riegel BJ, Dracup KA, Glaser D. A longitudinal causal model of cardiac invalidism following myocardial infarction. Nurs Res. 1998;47:285-292.

[11] Ford-Gilboe M. Family strengths, motivation, and resources as predictors of health promotion behavior in single-parent and two-parent families. Res Nurs Health. 1997;20:205-217.

[12] Mulligan S. Application of structural equation modeling in occupational therapy research. Am J Occup Ther. 1998;52:829-834.

[13] Research Plan for the National Center for Medical Rehabilitation Research. Bethesda, Md: National Institutes of Health; 1993. NIH Publication No. 93-3509.

[14] ICIDH-2: International Classification of Functioning and Disability. Beta-2 draft, Short Version. Geneva, Switzerland: World Health Organization; 1999.

[15] Lollar DJ, ed. Preventing Secondary Conditions Associated With Spina Bifida or Cerebral Palsy. Washington, DC: Spina Bifida Association of America; 1994.

[16] Guide to Physical Therapist Practice. Phys Ther. 1997;77:1163-1650.

[17] Campbell SK. Central nervous system dysfunction in children. In: Campbell SK, ed. Pediatric Neurologic Physical Therapy. New York, NY: Churchill Livingstone Inc; 1991:1-17.

[18] Hadders-Algra M, Brogren E, Forssberg H. Postural adjustments during sitting at preschool age: presence of a transient toddling phase. Dev Med Child Neurol. 1998;40:436-447.

[19] Russell DJ, Rosenbaum PL, Cadman DT, et al. The Gross Motor Function Measure: a means to evaluate the effects of physical therapy. Dev Med Child Neurol. 1989;31:341-352.

[20] Russell D, Rosenbaum P, Gowland C, et al. Gross Motor Function Measure Manual. 2nd ed. Hamilton, Ontario, Canada: Gross Motor Measures Group, McMaster University; 1993.

[21] Gordon J, Ghez C. Muscle receptors and spinal reflexes: the stretch reflex. In: Kandel ER, Schwarz JH, Jessell JM, eds. Principles of Neural Science. 3rd ed. New York, NY: Elsevier; 1991.

[22] Bobath B, Bobath K. Motor Development in the Different Types of Cerebral Palsy. London, England: William Heinemann Medical Books Ltd; 1975.

[23] Lance JW. Symposium synopsis. In: Feldman RG, Young RR, Koella WP, eds. Spasticity: Disordered Motor Control. Chicago, Ill: Year Book Medical Publishers; 1980:485-494.

[24] Lin JP. Dorsal rhizotomy and physical therapy [editorial]. Dev Med Child Neurol. 1998;40:219.

[25] Giuliani CA. Dorsal rhizotomy for children with cerebral palsy: support for concepts of motor control. Phys Ther. 1991;71:248-259.

[26] Scrutton D, Rosenbaum P. Locomotor development in children with cerebral palsy. In: Connolly KJ, Forssberg H, eds. Neurophysiology and Neuropsychology of Motor Development. London, England: MacKeith Press; 1997:101-123.

[27] Crothers B, Paine RS. The Natural History of Cerebral Palsy. Oxford, England: MacKeith Press; 1959.

[28] Bleck EE. Locomotor prognosis in cerebral palsy. Dev Med Child Neurol. 1975;17:18-25.

[29] Molnar GE, Gordon SU. Cerebral palsy: predictive value of selected clinical signs for early prognostication of motor function. Arch Phys Med Rehabil. 1976;57:153-158.

[30] Watt JM, Robertson CM, Grace MG. Early prognosis for ambulation of neonatal intensive care survivors with cerebral palsy. Dev Med Child Neurol. 1989;31:766-773.

[31] Campos Da Paz AC, Burnett SM, Braga LW. Walking prognosis in cerebral palsy: a 22-year retrospective analysis. Dev Med Child Neurol. 1994;36:130-134.

[32] Trahan J, Marcoux S. Factors associated with the inability of children with cerebral palsy to walk at six years: a retrospective study. Dev Med Child Neurol. 1994;36:787-795.

[33] Fetters L. Cerebral palsy: contemporary treatment concepts. In: Lister M, ed. Contemporary Management of Motor Control Problems: Proceedings of the II STEP Conference. Alexandria, Va: Foundation for Physical Therapy; 1991:219-224.

[34] Sala DA, Grant AD. Prognosis for ambulation in cerebral palsy. Dev Med Child Neurol. 1995;37:1020-1026.

[35] Badell-Ribera A. Cerebral palsy: postural-locomotor prognosis in spastic diplegia. Arch Phys Med Rehabil. 1985;66:614-619.

[36] Adelson E, Fraiberg S. Gross motor development in infants blind from birth. Child Dev. 1974;45:114-126.

[37] Bigelow AE. Locomotion and search behavior in blind infants. Infant Behav Dev. 1992; 15:179-189.

[38] Lee DN, Aronson E. Visual proprioceptive control of standing in human infants. Perception Psychophysics. 1974;15:529-532.

[39] Butterworth G, Hicks L. Visual proprioception and postural stability in infancy: a developmental study. Perception. 1977;6:255-262.

[40] Jouen F. Early visual-vestibular interactions and postural development. In: Bloch H, Bertenthal BI, eds. Sensory-Motor Organizations and Development in Infancy and Early Childhood. Dordrecht, the Netherlands: Kluwer; 1990:199-216.

[41] Nashner LM, Shumway-Cook A, Marin O. Stance posture control in select groups of children with cerebral palsy: deficits in sensory organization and muscular coordination. Exp Brain Res. 1983;49:393-409.

[42] Beals RK. Spastic paraplegia and diplegia: an evaluation of nonsurgical and surgical factors influencing the prognosis for ambulation. J Bone Joint Surg Am. 1966;48:827-846.

[43] Donoghue EC, Kirman BH, Bullmore GH, et al. Some factors affecting age of walking in a mentally retarded population. Dev Med Child Neurol. 1970;12:781-792.

[44] Tardieu C, Tardieu G, Colbeau-Justin P, et al. Trophic muscle regulation in children with congenital cerebral lesions. J Neurol Sci. 1979;42:357-364.

[45] Ziv I, Blackburn N, Rang M, Koreska J. Muscle growth in normal and spastic mice. Dev Med Child Neurol. 1984;26:94-99.

[46] Hufschmidt A, Mauritz KH. Chronic transformation of muscle in spasticity: a peripheral contribution to increased tone. J Neurol Neurosurg Psychiatry. 1985;48:676-685.

[47] Bax MC, Brown JK. Contractures and their therapy [editorial]. Dev Med Child Neurol. 1985;27:423- 424.

[48] Tardieu C, Lespargot A, Tabary C, Bret MD. Toe-walking in children with cerebral palsy: contributions of contracture and excessive contraction of triceps surae muscle. Phys Ther. 1989;69:656-662.

[49] Sutherland DH, Cooper L. The pathomechanics of progressive crouch gait in spastic diplegia. Orthop Clin North Am. 1978;9:143-154.

[50] Sutherland DH, Davids JR. Common gait abnormalities of the knee in cerebral palsy. Clin Orthop. 1993;288:139-147.

[51] Campbell SK. Therapy programs for children that last a lifetime. Physical & Occupational Therapy in Pediatrics. 1997;17(1):1-15.

[52] Olney SJ, MacPhail HE, Hedden DM, Boyce WF. Work and power in hemiplegic cerebral palsy gait. Phys Ther. 1990;70:431-438.

[53] Wiley ME, Damiano DL. Lower-extremity strength profiles in spastic cerebral palsy. Dev Med Child Neurol. 1998;40:100-107.

[54] Gossman MR, Sahrmann SA, Rose SJ. Review of length-associated changes in muscle: experimental evidence and clinical implications. Phys Ther. 1982;62:1799-1808.

[55] Kramer JF, MacPhail HE. Relationships among measures of walking efficiency, gross motor ability, and isokinetic strength in adolescents with cerebral palsy. Pediatric Physical Therapy. 1994;6:3-8.

[56] Damiano DL, Kelly LE, Vaughn CL. Effects of quadriceps femoris muscle strengthening on crouch gait in children with spastic diplegia. Phys Ther. 1995;75:658-671.

[57] MacPhail HE, Kramer JF. Effect of isokinetic strength-training on functional ability and walking efficiency in adolescents with cerebral palsy. Dev Med Child Neurol. 1995;37:763-775.

[58] Butler P, Engelbrecht M, Major RE, et al. Physiological cost index of walking for normal children and its use as an indicator of physical handicap. Dev Med Child Neurol. 1984;26:607-612.

[59] Rose J, Medeiros JM, Parker R. Energy cost index as an estimate of energy expenditure of cerebral-palsied children during assisted ambulation. Dev Med Child Neurol. 1985;27:485-490.

[60] Rose J, Gamble JG, Medeiros JM, et al. Energy cost of walking in normal children and in those with cerebral palsy: comparison of heart rate and oxygen uptake. J Pediatr Orthop. 1989;9:276-279.

[61] Rose J, McGill KC. The motor unit in cerebral palsy. Dev Med Child Neurol. 1998;40:270-277.

[62] Olney SJ. Efficacy of physical therapy in improving mechanical and metabolic efficiency of movement in cerebral palsy. Pediatric Physical Therapy. 1990;2:145-154.

[63] LeFrancois GR. The Lifespan. 3rd ed. Belmont, Calif: Wadsworth Inc; 1990.

[64] Werner EE. Vulnerable but invincible: high-risk children from birth to adulthood. Acta Paediatr Suppl. 1997;422:103-105.

[65] Thelen E, Smith LB. A Dynamic Systems Approach to the Development of Cognition and Action. Cambridge, Mass: The MIT Press; 1994.

[66] Morgan GA, MacTurk RH, Hrncir EJ. Mastery motivation: overview, definitions, and conceptual issues. In: MacTurk RH, Morgan GA, eds. Mastery Motivation: Origins, Conceptualizations, and Applications. Norwood, NJ: Ablex Publishing Corp; 1995:1-18.

[67] Cintas HL. The relationship of motor skill level and risk-taking during exploration in toddlers. Pediatric Physical Therapy. 1992;4:165-170.

[68] von Bertalanffy L. General System Theory. New York, NY: Brazillier; 1968.

[69] Bronfenbrenner U. The Ecology of Human Development: Experiments by Nature and Design. Cambridge, Mass: Harvard University Press; 1979.

[70] Sameroff A, Chandler M. Reproductive risk and the continuum of caretaking casualty. In: Horowitz F, ed. Review of Child Development Research; vol 4. Chicago, Ill: University of Chicago Press; 1975:187-244.

[71] Bower E, McLellan DL. Assessing motor skill acquisition in four centres for the treatment of children with cerebral palsy. Dev Med Child Neurol. 1994;36:902-909.

[72] Rantakallio P, von Wendt L, Makinen H. Influence of social background on psychomotor development in the first year of life and its correlation with later intellectual capacity: a prospective cohort study. Early Hum Dev. 1985;11:141-148.

[73] Bayley N. Comparisons of mental and motor test scores for ages 1-15 months by sex, birth order, race, geographical location and education of the parents. Child Dev. 1965;36:379-411.

[74] Parker S, Greer S, Zuckerman B. Double jeopardy: the impact of poverty on early childhood development. Pediatr Clin North Am. 1988;35:1227-1240.

[75] Aylward GP. Environmental influences on the developmental outcome of children at risk. Infants Young Child. 1990;2:1-9.

[76] Fetters L, Tronick EZ. Neuromotor development of cocaine-exposed and control infants from birth through 15 months: poor and poorer performance. Pediatrics. 1996;98:938-943.

[77] Coll C, Vohr B, Hoffman J, Oh W. Maternal and environmental factors affecting development outcome of infants of adolescent mothers. Dev Behav Pediatr. 1986;7:230-236.

[78] Carlson D, Labarba R, Sclafani J, Bowers C. Cognitive and motor development in infants of adolescent mothers: a longitudinal analysis. Int J Behav Dev. 1985;9:1-13.

[79] Palisano RJ, Chiarello LA, Haley SM. Factors related to mother-infant interaction in infants with motor delays. Pediatric Physical Therapy. 1993;5:55-60.

[80] Cintas HL. Cross-cultural similarities and differences in development and the impact of parental expectations on motor behavior. Pediatric Physical Therapy. 1995;7:103-111.

[81] Hopkins B, Westra T. Maternal handling and motor development: an intracultural study. Genet Soc Gen Psychol Monogr. 1988;114:377-408.

[82] Brazelton, TB, Koslowski B, Tronick E. Neonatal behavior among urban Zambians and Americans. J Am Acad Child Psychiatry. 1976;15:97-107.

[83] Konner MJ. Aspects of the developmental ethology of a foraging people. In: Blurton-Jones NG, ed. Ethological Studies of Child Behavior. Cambridge, England: Cambridge University Press; 1972:285-304.

[84] Super CM. Environmental effects on motor development: the case of "African infant precocity." Dev Med Child Neurol. 1976;18:561-567.

[85] Rosser PL, Randolph SM. Black American infants: the Howard University normative study. In: Nugent, JK, Lester BM, Brazelton TB, eds. The Cultural Context of Infancy: Biology, Culture, and Infant Development. Norwood, NJ: Ablex Publishing Corp; 1989:133-165.

[86] Lin CC, Fu VR. A comparison of child-rearing practices among Chinese, immigrant Chinese, and Caucasian-American parents. Child Dev. 1990;61:429-433.

[87] Nugent JK. Cross-cultural studies of child development: implications for clinicians. Zero to Three: National Center for Clinical Infant Programs. 1994;15:1-8.

[88] Dunst CJ, Trivette CM, Cross AH. Mediating influences of social support: personal, family, and child outcomes. Am J Ment Defic. 1986;90:403-417.

[89] Sparling JW, Kolobe THA, Ezzelle L. Family-centered intervention. In: Campbell SK, ed. Physical Therapy for Children. Philadelphia, Pa: WB Saunders Co; 1994:823-846.

[90] Law M, ed. Family-Centered Assessment and Intervention in Pediatric Rehabilitation. Birmingham, NY: The Haworth Press Inc; 1998.

[91] Ellwood PM. Outcomes management: a technology of patient experience. N Engl J Med. 1988;318:1549-1556.

[92] O'Neil M, Farel A, Palisano RJ. Parents perspectives of managed care: implications for pediatric physical therapy. Pediatric Physical Therapy. 1999; 11:24-32.

[93] Parette HP, Hendricks MD, Rock SL. Efficacy of therapeutic intervention intensity with infants and young children with cerebral palsy Infants Young Child. 1991;4:1-19.

[94] Coolman R, Foran W, Lee J. Oregon Guidelines for Medically-based Outpatient Physical Therapy and Occupational Therapy for Children With Special Needs in the Managed Care Environment. Eugene, Ore: Oregon Health Sciences University; 1998.

[95] Kolobe TH. Working with families of children with disabilities. Pediatric Physical Therapy. 1992;4:57-63.

[96] Palisano RJ. Pediatric physical therapy: an individual perspective. Pediatric Physical Therapy. 1994;6:140-141.

[97] Palisano RJ. Comparison of two methods of service delivery for students with learning disabilities. Physical & Occupational Therapy in Pediatrics. 1988;9(3):79-100.

[98] Cramer D, Tucker SM. The consumers' role in quality: partnering for quality outcomes. J Nurs Care Qual. 1995;9:54-60.

[99] King G, King S, Rosenbaum P. Interpersonal aspects of caregiving and client outcomes: a review of literature. Ambulatory Child Health. 1996;2:151-160.

[100] Campbell SK. Quantifying the effects of interventions for movement disorders resulting from cerebral palsy. J Child Neurol. 1996; 11(suppl 1):S61-S70.

[101] Delbecq AL, Van de Ven AH, Gustafson DH. Group Techniques for Program Planning: A Guide to Nominal Group and Delphi Processes. Middleton, Wis: Green Briar Press; 1986.

[102] Ashton BB, Pickles B, Roll JW. Reliability of goniometric measurements of hip motion in spastic cerebral palsy. Dev Med Child Neurol. 1978;20:87-94.

[103] Harris SR, Smith LH, Krukowski L. Goniometric reliability for a child with spastic quadriplegia. J Pediatr Orthop. 1985;5:348-351.

[104] Boomsma A. The robustness of maximum likelihood estimation in structural equation models. In: Cuttance P, Ecob R, eds. Structural Modeling by Example: Applications in Educational, Sociological, and Behavioral Research. New York, NY: Cambridge University Press; 1987:160-188.

[105] Bollen KA. Structural Equations with Latent Variables. New York, NY: John Wiley & Sons Inc; 1989.

[106] Gordis L. Epidemiology. Philadelphia, Pa: WB Saunders Co; 1996.

[107] Bower E, McLellan DL. Evaluating therapy in cerebral palsy. Child Care Health Dev. 1994;20:409-419.

[108] Last JM. A Dictionary of Epidemiology. 3rd ed. New York, NY: Oxford University Press; 1995.

[109] Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Practice. East Norwalk, Conn: Appleton and Lange; 1993.

DJ Bartlett, PT, PhD, is Assistant Professor, School of Physical Therapy, Faculty of Health Sciences, The University of Western Ontario, 1588 Elborn College, London, Ontario, Canada N6G 1H1 ([e-mail]). She is also an Associate Member, CanChild, Centre for Childhood Disability Research, McMaster University, Hamilton, Ontario, Canada. Address all correspondence to Dr Bartlett at the first address.

RJ Palisano, PT, ScD, is Professor, Department of Rehabilitation Sciences, MCP Hahnemann University, Philadelphia, Pa. He is also Co-investigator, CanChild, Centre for Childhood Disability Research, McMaster University.

Dr Bartlett and Dr Palisano provided concept/research design and writing. The authors acknowledge the Ontario Motor Growth Study Group at CanChild for providing a forum for the development of this work.

Dr Bartlett was supported by a Postdoctoral Fellowship from the Medical Research Council of Canada during the early phases of this work.