Relationship between spontaneous kicking and age of walking attainment in preterm infants with very low birth weight and full-term infants.

Improvements in the management of newborn infants have been associated with a decrease in neonatal mortality among preterm infants with very low birth weight (VLBW) (gestational age of <37 weeks and birth weight of <1,501 g). (1,2) The high risk of long-term developmental sequelae (6%-12% with major neurological disorders and 23%-60% with minor brain dysfunction) in the surviving infants, however, has continued to be of concern. (3,4) Current approaches to identifying and managing preterm infants with neurodevelopmental disorders have emphasized examination and intervention during the early months of life. (5-7)

Spontaneous movements in early infancy have been proposed as related to later motor control. (8,9) Increasing evidence indicates that examination of spontaneous motility may provide a sensitive and reliable indication of an infant's present condition and later neurological condition. (10-12) Among these spontaneous movements, rhythmical kicking has frequently been chosen for research because of its potential role in walking development. This can be seen in the studies of Thelen and colleagues, (13-15) in which kinematic analysis was used to examine longitudinally the leg movements of full-term infants from birth until attainment of walking. Kinematic variables examined included frequency, spatiotemporal organization, interjoint coordination, and interlimb coordination. They observed that spontaneous kicking, which is locomotion-like leg movements of the newborn infant in the supine position, has a spatial and temporal organization similar to that of mature walking. Rhythmical kicking, therefore, has been postulated as the precursory motor pattern that is later incorporated into upright locomotion. (13)

Several researchers (16-24) have used kinematic analysis to examine the kicking development of preterm infants. Some authors (17-19,22,23) aimed to find early kicking differences between full-term infants without known impairments or pathology and preterm infants who sustained severe perinatal brain damage (ie, extensive hemorrhage or ischemia as detected by cranial ultrasound) and later developed major neurological disorders such as cerebral palsy. Certain kicking features have been found among preterm infants with overt cranial sonographic abnormalities. For example, Droit et al (19) documented a longer intra-kick pause and less alternate movements (simultaneous flexion of one leg and extension of the other leg) together with more semi-both-leg movements (simultaneous flexion and nonsimultaneous extension of both legs) in preterm infants with overt neurosonographic abnormalities when they reached full-term age. Heriza, (17) Yokochi et al, (18) and Vaal et al (23) reported a higher correlation of interjoint coordination and a lower variability of spatiotemporal organization in preterm infants with overt cranial sonographic abnormalities from 3 months corrected age onward. Van der Heide et al, (22) however, found no differences between groups in interjoint coordination at 1 and 3 months corrected age. Although not entirely consistent, these studies on kicking have helped detect early neuromotor predictors for major neurodevelopmental disorders in infants born prematurely.

Other studies that contrasted the early kicking of full-term infants without known impairments or pathology and preterm infants who had no overt cranial sonographic abnormalities also showed differences. (16,20,21,24) Heriza (16) documented a lower kick frequency and a longer inter-kick pause in preterm infants who had no overt neurosonographic abnormalities when they approached full-term age. Jeng et al (24) found distinct kicking features at post-term ages in preterm infants who had no overt neurosonographic abnormalities, particularly those born at young gestational ages. The kicking features included a higher kick frequency together with a shorter flexion phase at 4 months corrected age and a higher correlation of interjoint coordination together with a lower variability of interlimb coordination at 2 and 4 months corrected age. Nevertheless, Geerdink et al (20) and Piek and Gasson (21) observed a paradoxical direction of differences in interjoint coordination at post-term ages. Despite the reported early kicking differences for preterm infants without overt cranial sonographic abnormalities, their functional importance remains unclear because no outcomes have been measured.

As the survival rate of preterm infants with VLBW continues to improve, there has been a concomitant reduction of overt cranial sonographic abnormalities and rate of cerebral palsy. (25,26) In the absence of cerebral palsy, however, considerable proportions (20%-30%) of preterm infants with VLBW have been found to exhibit neurobehavioral problems in follow-ups during preschool and school ages. (27-29) Thus, scrutiny of the early neurodevelopmental process of preterm infants without overt cranial sonographic abnormalities may help identify those who will exhibit neurodevelopmental problems in the future. To this end, age of walking attainment is a useful measure of early neurodevelopment because it reflects various degrees of motor delay and has been increasingly used for studies of infants with prematurity. (30-32) Several studies (31,33,34) have shown that preterm infants attain the ability to walk later than full-term infants, after correction for prematurity. Furthermore, failure to walk by 18 months corrected age has been found to be associated with neurodevelopmental disorders. (31,35,36) Thus, identification of early kicking predictors for the age of walking attainment in preterm infants may provide valuable clues for the genesis of neurodevelopmental disorders.

This study extended the findings of Jeng et al (24) by examining the age of walking attainment in preterm infants who had VLBW but no overt neurosonographic abnormalities and full-term infants without known impairments or pathology and by examining the relationship between spontaneous kicking and age of walking attainment in these infants. Infants were prospectively examined for their kicking movement at 2 and 4 months corrected age, and they were monitored for their age of walking attainment until 18 months corrected age.

Method

Subjects

In our study, we enrolled both preterm infants with VLBW and full-term infants without known impairments or pathology who were born at the National Taiwan University Hospital, Taipei, Taiwan. The inclusion criteria for infants with VLBW were: gestational age of less than 37 weeks, birth weight of less than 1,501 g, absence of congenital or chromosomal abnormality, and absence of severe cranial sonographic abnormalities. Severe cranial sonographic abnormalities were considered to be grade III to IV intraventricular hemorrhage or cystic periventricular leukomalacia as examined by serial brain ultrasound using the definitions of Papile et al (37) and de Vries et al. (38) Brain ultrasound was performed on the 1st, 3rd, 7th, and 14th days and then, if necessary, every 2 or 4 weeks until hospital discharge. The inclusion criteria for full-term infants were: gestational age of 38 to 42 weeks, no maternal or perinatal complications, normal status on newborn examination, and birth weight appropriate for gestational age. Informed parental consent was obtained for each infant before participation in the study and prior to hospital discharge.

Twenty-two infants with VLBW and 22 full-term infants were included in this study between June 1997 and December 1998. The 22 infants with VLBW (15 boys [68%] and 7 girls [32%]) had a mean gestational age of 30.1 weeks (SD = 2.5, range = 26-35) and a mean birth weight of 1,180 g (SD = 243, range = 636-1,492). Among the infants with VLBW, 3 (14%) had minor intraventricular hemorrhage (grade I-II) and 9 (41%) had chronic lung disease (respiratory disease that requires oxygen therapy for 28 days or longer (39)). Fifteen of the mothers (68%) had more than 12 years of education, 5 mothers (23%) had 9 to 12 years of education, and 2 mothers (9%) had less than 9 years of education. The 22 full-term infants (8 boys [36%] and 14 girls [64%]) had a mean gestational age of 39.1 weeks (SD = 3.1, range = 38-41) and a mean birth weight of 3,298 g (SD = 219, range = 2,910-3,632). Among the full-term infants, 15 of the mothers (68%) had more than 12 years of education and 7 mothers (32%) had 9 to 12 years of education. Comparison of perinatal and demographic characteristics between groups revealed that the infants with VLBW included more boys ([chi square] = 4.5, df = 1, P = .03) and had lower gestational age (F = 253; df = 1,43; P = .0001) and birth weight (F = 923.6; d = 1,43; P = .0001) than the full-term infants. The groups were comparable in maternal education ([chi square] = 2.9, df = 2, P = .2).

Instrumentation and Testing Procedure

The infants were prospectively examined for kicking movements at 2 and 4 months corrected age and were monitored for their age of walking attainment until 18 months corrected age. During the follow-up period, all infants had regularly scheduled clinic visits for vaccinations at 1, 2, 4, 6, 9, 12, 15, and 18 months corrected age at the pediatric department of the hospital.

During the kicking test, infants were placed in the supine position on an examination table and wore black shorts to reveal the following anatomical landmarks on both sides of the body: mid trunk, greater trochanter, lateral femoral condyle, lateral malleolus, and fifth metatarsal head. Reflective ball-shaped markers with 1- to 2-cm diameter were placed at these landmarks to define the hip, knee, and ankle angles. The infants were tested 1 hour before feeding and were kept in an alert state (state 5, according to the Brazelton Neonatal Behavioral Assessment Scale (40)). Kicking movements were recorded for 5 minutes using 4 synchronized video cameras (2 Peak High Speed Video cameras * and 2 WV CL-350 video cameras ([dagger])). Each camera was operated at 60 Hz and was connected to a videocassette recorder (SVHS AG-1960 and 1970 ([dagger])) and a time code generator (Horita SDR-50[double dagger])). Two cameras were placed on each side of the infants at a distance of 2 m. The cameras were angled at 70 degrees to construct a 3-dimensional analysis of the movements of the same limb with calibration errors of <3 mm. The testing procedure was described in more detail in Jeng et al. (24)

During the first visit to the clinic, when the infants were 2 months corrected age, the parents were asked to prospectively monitor and record their child's age of walking attainment, which was defined as the time the infant began walking 5 successive steps without support. (33) The data were collected within a week of this event and were recorded by corrected age. To reduce observer bias in the determination of age of walking due to parental report alone or other problems such as faulty memory, a research assistant made biweekly telephone calls when the infants were 9 to 18 months corrected age and asked a specific set of questions regarding the child's motor status and age of walking. Parental report of locomotor status has been found to have a high degree of agreement (>95%) with telephone call response. (31) For those infants whose telephone call response did not correspond to the parental report (<5%), the former information was used for analysis.

All infants also were examined for their neurological condition at 18 months corrected age by a physiatrist in the hospital. The physiatrist was masked to the study's purpose and infant groups. The examination included aspects of neurological signs, primitive and pathological reflexes, and motor function. In addition, the information on health state and developmental intervention received during the follow-up period was recorded. The physiatrist made the clinical diagnosis of the infants based on the neurological examination.

Data Acquisition and Reduction

Previous studies by Droit et al (19) and Piek and Carman (41) indicated that bouts of spontaneous kicking usually last only for a few seconds. Therefore, a 20-second segment of recorded movement that best represented continuous kicking movements was selected for the individual infant at each month. The criteria for selection of the segment were: (1) when the behavioral state of the infant was awake but not crying, (2) when continuous kicking was present, and (3) when all joint markers were visible in the camera view. The selected video record was analyzed using the Peak Performance Motion Analysis System (Motus version 3.01 *) with a fourth-order Butterworth filter and a filtering rate at 6 Hz. The 2-dimensional video data set generated from the cameras that recorded the same limb movements was converted into 3-dimensional spatial coordinates. Joint angles were defined as the relative angles between the anatomical landmarks, with 180 degrees regarded as full extension at each joint and 0 degree regarded as full flexion. Angular displacement and velocity of the hip, knee, and ankle joints were calculated from the coordinate data. The obtained angular displacement and velocity data were used to calculate the kinematic variables of kick frequency, spatiotemporal organization, interjoint coordination, and interlimb coordination using the Visual Basic Software (version 5.0) program ([section]).

Kick frequency was measured as the number of kicks by each leg during the 20-second period and was converted to cycles per minute. Spatiotemporal organization included kick amplitude and movement phases. Kick amplitude was measured by the range of hip flexion during a kick cycle, which consisted of 4 movement phases: flexion phase, intra-kick pause, extension phase, and inter-kick pause. The flexion phase was the time from the initiation of hip flexion until the movement ceased; the intra-kick pause was the time from the end of hip flexion until the initiation of hip extension; the extension phase was the time from the initiation of hip extension until the movement ceased; and the inter-kick pause was the time from the end of hip extension until the initiation of the next flexion phase. (15) The initiation of hip flexion was the frame at which continuous hip flexion (angular velocity of <0[degrees] / s for more than 10 frames) was first noticed (angular velocity = -14[degrees] / s). The termination of hip flexion was the frame at which hip flexion stopped, as indicated by the occurrence of minimum absolute value of angular velocity. The initiation of hip extension was the frame at which continuous hip extension (angular velocity of >0[degrees] / s for more than 10 frames) was first noticed (angular velocity = 14[degrees] / s). The termination of hip extension was the frame at which hip extension stopped, as indicated by the occurrence of minimum absolute value of angular velocity.

Interjoint coordination was measured by pair-wise cross-correlations of the hip, knee, and ankle joint angles for the same leg during the kick cycles using Pearson product-moment correlation, and these measurements were transformed to Fisher Z scores. (20) Interlimb coordination was measured by the number of kicks and the percentages of alternate, unilateral, and synchronous kick patterns during the 20-second period. (42) An alternate kick was defined as simultaneous flexion of one leg and extension of the other leg with the flexion phase of the 2 legs overlapping for less than 50% of the movement. A unilateral kick was defined as isolated flexion and extension of one leg when the other leg was in intra-kick pause or inter-kick pause. A synchronous kick was defined as simultaneous flexion (or extension) of both legs during more than 50% of the flexion (or extension) phase. The predominant interlimb coordination pattern also was determined for each infant, as defined by the most frequent type of kick pattern that occurred. The calculations for each kinematic variable have been described in detail elsewhere. (24)

Data Analysis

The perinatal and demographic characteristics were compared between the infants with VLBW and the full-term infants using an analysis of variance (ANOVA) for continuous variables and chi-square tests for categorical variables. The distributions of age of walking attainment for the infants with VLBW and the full-term infants were estimated by the Kaplan-Meier method, a nonparametric survival analysis, and were compared using the log-rank test. (43) During the follow-up of subjects until they reach a pre-specified endpoint (attainment of walking), some subjects may complete the follow-up period of time (18 months corrected age) before the endpoint is reached. For such cases, the survival times were censored; that is, subjects survived to a certain time beyond which their status was unknown. The noncensored survival times were referred to as event times. The Kaplan-Meier method incorporates information from both censored and noncensored data.

The relationship between kicking movements and age of walking attainment for the preterm infants with VLBW and the full-term infants was examined using an ANOVA for repeated measures with a 2 (group) x 2 (age) factorial design. The outcome of walking attainment was stratified into normal versus delayed using data from the full-term infants as the reference. Normal walking attainment was defined as age of walking attainment earlier than 2 standard deviations from the mean age of walking attainment. Delayed walking attainment was considered mild delay or severe delay. Mild delay was defined as age of walking attainment later than 2 standard deviations but earlier than 4 standard deviations from the mean, and severe delay was defined as age of walking attainment later than 4 standard deviations from the mean. (44)

The associations of kicking variables with age of walking attainment also were examined using Cox proportional-hazards univariate regression models in which the outcome of age of walking attainment was a continuous variable. (43) The models provide the maximum likelihood estimates of rate ratio (RR) and the 95% confidence interval (CI) for individual kicking variables. A rate ratio of greater than 1 indicates that the level under consideration may lead to an increased rate of walking attainment as compared with a reference level. A rate ratio of less than 1 indicates that the level under consideration may result in a decreased rate of walking attainment. Multivariate regression analyses of Cox proportional-hazards models were subsequently used to adjust for potentially confounding variables such as prematurity and gestational age. A P value of less than .05 was considered as statistically significant, and the acceptable level for statistical significance in post hoc tests was adjusted to .013 (.05/4). All statistical analyses were performed using the Statistical Analysis Software (SAS) program (version 8.0) ([parallel]).

Results

Age of Walking Attainment

The cumulative probabilities of being unable to walk against time for the infants with VLBW and the full-term infants are presented in Figure 1. Of the full-term infants, 13 (60%) attained walking ability by 12 months of age and 9 (40%) attained walking ability during the age period of 12 to 18 months. Of the infants with VLBW, 9 (40%) attained walking ability by 12 months corrected age and 11 (51%) attained walking ability during the corrected age of 12 to 18 months. Two infants with VLBW (9%) were still unable to walk at 18 months corrected age. Survival analysis of the distributions of age of walking revealed that the infants with VLBW attained walking when they were older (median = 14 months corrected age, minimum = 9.3 months corrected age) than the full-term infants (median = 12 months, range = 9-15 months) ([chi square] = 6.49, P = .01).

[FIGURE 1 OMITTED]

On the basis of the mean and standard deviation of the full-term infants' distribution in age of walking attainment (12.1 [+ or -] 1.4 months), 21 full-term infants (96%) were classified as having normal walking attainment (walking attainment by 15 months of age) and 1 infant (4%) had mild delay in walking attainment (walking attainment within 15-18 months of age). Of the infants with VLBW, 17 (77%) showed normal walking attainment, 3 (14%) had mild delay in walking attainment, and 2 (9%) had severe delay in walking attainment (failure to walk by 18 months corrected age).

Examination of the perinatal history of the infants with VLBW revealed that of the 9 infants with a low gestational age (<30 weeks), 3 (33%) showed delay in walking attainment and 6 (67%) showed normal walking attainment. Of the 3 infants with minor intraventricular hemorrhage, 2 (67%) exhibited delay in walking attainment and 1 (33%) had normal walking attainment. Of the 9 infants with chronic lung disease, 3 (33%) showed delay in walking attainment and 6 (67%) had normal walking attainment.

Neurologic Outcome and Health State

Examination of the neurologic outcome at 18 months corrected age revealed normal neuromotor development in all of the infants with normal walking attainment, motor delay in the 4 infants with mild delay in walking attainment (infants A, D, E, and F), and cerebral palsy with spastic diplegia in the 2 infants with severe delay in walking attainment (infants B and C). Cerebral palsy was diagnosed as the presence of persistently abnormal neurological signs, primitive and pathologic reflexes, and motor dysfunction, whereas motor delay was diagnosed as the presence of motor dysfunction but without abnormal neurological signs. The neurological diagnosis was made by the physiatrist.

Review of the health and developmental intervention records during the follow-up period showed that of the infants with VLBW, 9 (40%) had additional clinic visits (>3 visits) and 4 (18%) had hospitalization (s) because of upper respiratory infection, bronchiolitis, renal tubular acidosis, or hepatitis. Three infants (14%) received developmental interventions (2 infants had developmental counseling, and 1 infant had physical therapy and occupational therapy for 28 sessions). Of the full-term infants, none had additional clinic visits, hospitalization, or developmental intervention during the follow-up period.

Patterns of Kicking Development

Comparison of kicking movements between limbs in all infants revealed differences (all P<.05) but no preference of limb use. The kicking performance of the 2 limbs, therefore, was treated as independent data. Table 1 illustrates the group means of each kicking variable for the infants with mild to severe delay in walking attainment and the infants with normal walking attainment at 2 and 4 months corrected age.

Comparison of kick frequency between groups over age showed an interaction effect (F = 14.1; df = 1,86; P = .0003). Post hoc tests revealed no difference in the kick frequency between groups at 2 months corrected age. The infants with mild to severe delay in walking attainment, however, manifested a higher kick frequency than those with normal walking attainment at 4 months corrected age (P = .0003) (Figs. 2A and 2B). From 2 to 4 months corrected age, the infants with mild to severe delay in walking attainment tended to increase their kick frequency (P = .04), whereas those with normal walking attainment tended to decrease their kick frequency (P = .02).

[FIGURE 2 OMITTED]

With regard to spatiotemporal organization, an age effect was found for the kick amplitude (F = 54.9; df = 1,84; P<.0001); all infants exhibited a larger kick amplitude with increasing age. Effects of group (F = 4; df = 1,84; P = .049) and age (F = 59.7; df = 1,84; P<.0001) were obtained for the flexion phase. Post hoc tests showed no differences in the flexion phase between groups at 2 months corrected age. Nevertheless, the infants with mild to severe delay in walking attainment had a shorter flexion phase than those with normal walking attainment at 4 months corrected age (P = .02) (Figs. 3A and 3B). With increasing age, the infants with mild to severe delay in walking attainment exhibited a stable flexion phase; whereas those with normal walking attainment showed a longer flexion phase (P<.0001). An age effect also was found for the intra-kick pause (F = 18.8; df = 1,83; P<.0001); all infants manifested a longer intra-kick pause over age. No difference was found between groups or across ages for the extension phase and the inter-kick pause.

[FIGURE 3 OMITTED]

Examination of the interjoint coordination revealed effects of age (F = 57.2; df = 1,84; P<.0001) and interaction (F = 5.3; df = 1,84; P = .02) for the hip-knee coordination. The difference of the hip-knee correlation at 2 months corrected age between groups was borderline (P = .06). No difference was observed for the hip-knee correlation at 4 months corrected age between groups (Figs. 4A and 4B). All infants exhibited a decline in the hip-knee correlation with age (both P<.0001). No difference was noted between groups and across ages for the hip-ankle and knee-ankle coordination.

[FIGURE 4 OMITTED]

Analysis of the interlimb coordination showed an age effect (F = 4.9; df = 1,86; P = .03) for the percentage of alternate kicks that all infants exhibited a decrease in the percentage of alternate kicks over age. Furthermore, effects of age (F = 23.5; df = 1,86; P<.0001) and interaction (F = 13; df = 1,86; P = .0005) were found for the percentage of unilateral kicks and synchronous kicks (age: F = 53.5; df = 1,86; P<.0001; interaction: F = 8.7; df = 1,86; P = .004). Post hoc tests revealed that the infants with mild to severe delay in walking attainment exhibited a higher percentage of unilateral kicks (P = .003) together with a lower percentage of synchronous kicks (P = .004) at 2 months corrected age than those with normal walking attainment (Figs. 5A and 5B). Furthermore, the infants with mild to severe delay in walking attainment had a lower percentage of unilateral kicks (P = .03) together with a higher percentage of synchronous kicks (P = .03) at 4 months corrected age than those with normal walking attainment. All infants manifested a decreasing percentage of unilateral kicks together with an increasing percentage of synchronous kicks over age (all P<.01). Examination of the predominant interlimb coordination pattern revealed a skewed distribution of kick patterns in the infants with mild to severe delay in walking attainment (Tab. 2). A high proportion (66%) adopted the unilateral kicks as the predominant pattern at 2 months corrected age, and this was repeated for the synchronous kicks at 4 months corrected age (83%).

[FIGURE 5 OMITTED]

Relationship Between Kicking Movement and Age of Walking Attainment

Cox univariate regression analyses for the relationship between kicking variables and age of walking attainment in all infants showed that a short intra-kick pause, a high hip-knee correlation, a high percentage of unilateral kicks, and a low percentage of synchronous kicks at 2 months corrected age were each associated with a decreased rate of walking attainment (all P<.05) (Tab. 3). Furthermore, a high kick frequency, a short flexion phase, a short intra-kick pause, a low percentage of unilateral kicks, and a high percentage of synchronous kicks at 4 months corrected age also were related to a decreased rate of walking attainment (all P<.05).

Subsequent multivariate regression analyses revealed that the relationship of kicking variables to age of walking attainment remained unchanged when we controlled for prematurity (full-term versus preterm). When adjustment was made for gestational age (per week increase), however, the effect of the flexion phase at 4 months corrected age became nonsignificant (RR = 2.56, 95% CI = 0.91 - 7.16) (Tab. 3).

Discussion

This study used survival analysis (1) to compare the age of walking attainment of preterm infants who had VLBW but no overt cranial sonographic abnormalities with the age of full-term infants and (2) to identify early kicking predictors for age of walking attainment in these infants. By the end of our follow-up, the preterm infants who had VLBW but no overt cranial sonographic abnormalities showed a large variation in the age of walking attainment, as did the full-term infants. Comparison of the distributions of age of walking showed that higher proportions of the infants with VLBW exhibited mild delay and severe delay in walking attainment (14% and 9%) than the full-term infants (4% and 0%). The proportions of severe delay in walking attainment were slightly lower than the data reported by previous investigators (33,35) for preterm (10%) and full-term infants (1.6 %). The results indicate that preterm infants who had VLBW but no overt neurosonographic abnormalities have a greater risk of delayed walking attainment than their full-term counterparts.

The first attempt to identify kicking predictors for age of walking attainment was made by comparing the kicking movements between infants with mild to severe delay in walking attainment (clinically defined walking delay) and those with normal walking attainment. The results showed that (1) a high kick frequency together with a short flexion phase at 4 months corrected age and (2) a low variability of interlimb coordination at 2 and 4 months corrected age were closely related to walking delay. Subsequently, Cox proportional-hazards analyses confirmed the effects of these kicking predictors on age of walking attainment and further indicated that the effect of the flexion phase was confounded by low gestational age. In addition, 2 more predictors were found for delayed walking attainment: a high hip-knee correlation at 2 months corrected age and a short intra-kick pause at 2 and 4 months corrected age. Our findings suggest that kicking frequency, intra-kick pause, hip-knee coordination, and interlimb coordination are important factors related to walking development in prematurity. The identified kicking predictors are useful and relevant measures for day-to-day practice because they are easily observable in home, day care, and clinical settings. Clinicians could use the identified predictors as early markers to detect those infants who may develop walking delay, and they could use these variables as treatment goals when designing early intervention programs. The significance and potential causes of the individual kicking predictors for age of walking attainment are delineated as follows.

A high kick frequency at 4 months corrected age was consistently associated with a decreased rate of walking attainment. From 2 to 4 months corrected age, the infants with normal walking attainment showed a decrease in kick frequency, whereas those with mild to severe delay in walking attainment began with a lower kick frequency and subsequently the frequency increased to higher levels than in infants with normal walking attainment. The different trends in the development of kick frequency among those infants may partly be due to different rates in neural maturation and differences in muscle tension. Van Wulfften and Hopkins (45) observed a broad spectrum of interrelated changes in infant behaviors (eg, learning abilities, motor behavior, social competence) around the age of 2 to 4 months, and they attributed these changes to neural maturation. The changes in motor behaviors are characterized by reorganization from spontaneous to fine-distal and goal-directed movements. In our study, the infants with normal walking attainment may have undergone age-appropriate maturation in neural functions, so that toward 4 months corrected age, they manifested only a few kicks and were paying more attention to manipulation and vocalization. Furthermore, differences in the muscle tension of lower limbs also may influence the kick frequency because higher muscle tension is related to shorter pauses and therefore higher kick frequency. (16) Future studies can examine various aspects of neural functions and muscle tension during this age period to determine their relationship to walking development.

A short intra-kick pause at 2 and 4 months corrected age also may predict a decreased rate of walking attainment. The infants with normal walking attainment tended to kick with a longer duration of intra-kick pause than did those with mild to severe delay in walking attainment throughout the follow-up period. However, the intersubject variations in the intra-kick pause were large (0.24 [+ or -] 0.26 seconds versus 0.71 [+ or -] 1.60 seconds at 2 months; 1.11 [+ or -] 1.62 seconds versus 2.72 [+ or -] 4.14 seconds at 4 months), and the differences between groups did not reach statistical significance. The tendency of the infants with a decreased rate of walking attainment to assume a short intra-kick pause for kicking might be linked to insufficient muscle force in the lower limbs and abdomen for holding the lower limbs in the air. In addition, a high level of stiffness in the passive viscoelastic properties of limb muscles also may result in shorter pauses. (16) Further investigation of muscle force and passive viscoelastic properties of muscles can illuminate their roles in early kicking movement and subsequent walking attainment.

A high hip-knee correlation at 2 months corrected age is closely related to a decreased rate in walking attainment. The unfavorable association between high interjoint correlation and motor outcome has previously been documented. (17,18,23) The development of interjoint coordination of spontaneous limb movements can be characterized by high pair-wise joint correlations at a newborn age, followed by a decrease in the joint correlations. (42,46,47) The release of joints, as indicated by declined joint correlation, from the early obligatory synergism has been considered as an important feature allowing for the emergence of new movement patterns. (42) It has been postulated that the tight interjoint coupling may be linked to an inability to distinguish reciprocal excitation of the antagonist muscles and to poor reciprocal inhibition. (48)

Our data did not support the predictive utility of the hip-ankle and knee-ankle coordination as previously reported. (23) These incongruous results may be due to differences in assessment age because Vaal et al (23) followed up infant kicking movement from 6 to 26 weeks corrected age. They indicated that high knee-ankle and hip-ankle correlations during the age period of 18 to 26 weeks were associated with poor motor outcome. In contrast, we examined infant kicking movement only at 2 and 4 months corrected age, so the follow-up duration in our study might have been too short to detect the predictive value of the hip-ankle and knee-ankle coordination at older ages.

A low variability in interlimb coordination patterns at 2 and 4 months corrected age is associated with a decreased rate of walking attainment. In the development of bilateral movement coordination, all infants exhibited a trend toward more synchronous kicks, together with less unilateral kicks by 4 months corrected age. Infants with normal walking attainment manifested a large variation in the distribution of interlimb coordination patterns. Although most infants adopted unilateral kicks as their predominant pattern at 2 months corrected age and synchronous kicks at 4 months corrected age, some had assumed other forms of coordination as the predominant pattern. Infants with mild to severe delay in walking attainment, however, exhibited a considerably skewed distribution of kicking patterns. An extremely high proportion adopted the unilateral kicking pattern at 2 months corrected age and the synchronous kicking pattern at 4 months corrected age. This obtained relationship between variability of kicking pattern and motor outcome concurs with the data of Vaal et al. (23) It has been proposed that a biological movement system must have stable movement patterns, but at the same time variability or flexibility is necessary for allowing adaptation to environmental changes. (49-51) We speculate that infants with normal walking attainment may have greater flexibility of the lower limbs, so they can accommodate for ongoing changes in the development of leg movements. Further study is warranted to determine what factors--whether intrinsic or extrinsic--constrained the limb actions and thus the capacity for variation in the infants with mild to severe delay in walking attainment.

It is worthwhile finally to highlight the usefulness of survival analysis in the investigation of age of walking attainment and its predictive factors in this study. Survival analysis has been widely used in clinical trials to measure the time to a certain event, such as death, relapse, response, and the development of a given disease when the observation times for some subjects are unknown (ie, censored). (43) In general, survival analysis provides full information regarding the rate of attaining walking ability over the whole follow-up period and allows comparison of such data between groups. The advantage of the Kaplan-Meier method used in our study is that it does not require a parametric assumption of the survival rate and can be readily applied for a small sample size. Nevertheless, when other covariates need to be adjusted for in comparing 2 groups' survival function, the Kaplan-Meier estimation is no longer sufficient. In view of this limitation, we used Cox proportional-hazards models. Under the assumption that the ratio of hazard rate between 2 levels of an independent variable (ie, exposed versus nonexposed) remains constant over variable periods of time, the models provide estimates of rate ratio for each independent variable with adjustment for all the remaining variables. Because Cox proportional-hazards models can incorporate censored information in the analyses, they extract more information from the data than traditional discrete analyses, such as ANOVA for repeated measures. This point was clearly illustrated in the identification of kicking predictors for age of walking attainment in the 2 kinds of analyses performed in this study.

The major limitation of this study was that only a 20-second segment of recorded movement was selected for the individual infant for kinematic analysis. The use of a short segment of movement records for analysis may introduce bias, particularly when kick frequency is a variable of interest. Further research is needed to use a longer section window of kicking data to enhance the generalizability of the findings.

Conclusion

The results of this study showed that preterm infants who had VLBW but no overt neurosonographic abnormalities attained walking ability at older ages than full-term infants without known impairments or pathology after correction for prematurity. Furthermore, several features of spontaneous kicking were found to be associated with a decreased rate of walking attainment in both infants with VLBW and full-term infants. The early kicking predictors included a high hip-knee correlation at 2 months corrected age, a high kick frequency at 4 months corrected age, and a short intra-kick pause together with a low variability in interlimb coordination at 2 and 4 months corrected age. These results suggest that the identified kicking variables may be used as early neuromotor markers for the prediction of age of walking attainment in both preterm and full-term infants.

Our data provide information that can help physical therapists to understand the complex connection between spontaneous kicking movement and the development of initial functions (ie, walking). This connection is of current importance due to the increasing numbers of preterm infants seen clinically, coupled with the overall goal of offering developmentally appropriate intervention as early as possible. Moreover, due to the general lack of data supporting the use of specific interventions in young infants with special needs, future research is necessary to investigate the control mechanisms for these early kicking features, to examine whether intervention of these kicking variables affects walking outcome, and to determine the functional importance of age of walking attainment on long-term outcome.

Table 1
Kicking Movements at 2 and 4 Months Corrected Age for
Infants With Mild to Severe Delay in Walking Attainment
and Infants With Normal Walking Attainment

                                    2 Months Corrected Age

                                    Mild to Severe Delay in
                                    Walking Attainment (n = 6)

Variable                            [bar]X   SD      Range

Kick frequency (cycles/min)         15.1      9.6     0-36
Kick amplitude (([degrees]))        35.7      7.7    23.7-51.6
Phase duration (s)
  Flexion phase                      0.48     0.18    0.28-0.87
  Intra-kick pause                   0.24     0.26    0.02-0.73
  Extension phase                    0.45     0.11    0.28-0.63
  Inter-kick pause                   1.84     1.44    0.02-5.32
Interjoint coordination (Z score)
  Hip-knee coupling                  1.34     0.29    0.91-1.82
  Hip-ankle coupling                 0.28     0.40   -0.18-1.14
  Knee-ankle coupling                0.21     0.44   -0.44-0.97
Interlimb coordination
  Alternate kicks
    No. of kicks/20 s                1.8      1.6     0-4
    % of total kicks                24.0     20.2     0-44.5
  Unilateral kicks
    No. of kicks/20 s                4.0      2.5     2-8
    % of total kicks                62.4     26.9    28.6-100
  Synchronous kicks
    No. of kicks/20 s                1.2      1.0     0-2
    % of total kicks                13.6     11.1     0-28.6

                                    2 Months Corrected Age

                                    Normal Walking Attainment
                                    (n = 38)

Variable                            [bar]X   SD      Range

Kick frequency (cycles/min)         20.9     14.2     3-72
Kick amplitude (([degrees]))        40.0     11.8    12.7-76.6
Phase duration (s)
  Flexion phase                      0.49     0.12    0.31-0.92
  Intra-kick pause                   0.71     1.60    0.71-1.58
  Extension phase                    0.55     0.15    0.24-1.12
  Inter-kick pause                   1.68     2.33    0.03-16.8
Interjoint coordination (Z score)
  Hip-knee coupling                  1.12     0.52   -0.41-2.65
  Hip-ankle coupling                 0.57     0.42   -1.45-1.16
  Knee-ankle coupling                0.42     0.40   -1.16-1.12
Interlimb coordination
  Alternate kicks
    No. of kicks/20 s                2.6      2.2     0-10
    % of total kicks                30.8     20.3     0-100
  Unilateral kicks
    No. of kicks/20 s                2.6      2.5     0-12
    % of total kicks                35.7     28.6     0-100
  Synchronous kicks
    No. of kicks/20 s                3.3      3.4     0-11
    % of total kicks                33.5     22.7     0-75

                                    4 Months Corrected Age

                                    Mild to Severe Delay in
                                    Walking Attainment (n = 6)

Variable                            [bar]X   SD      Range

Kick frequency (cycles/min)         31.1     18.0     6-51
Kick amplitude (([degrees]))        45.0     12.2    33.2-72.6
Phase duration (s)
  Flexion phase                      0.54     0.16    0.38-0.83
  Intra-kick pause                   1.11     1.62    0.02-5.28

  Extension phase                    0.49     0.20    0.31-1.00
  Inter-kick pause                   0.89     1.11    0.04-3.90
Interjoint coordination (Z score)
  Hip-knee coupling                  0.52     0.29    0.10-0.95
  Hip-ankle coupling                 0.56     0.40    0.13-1.56
  Knee-ankle coupling                0.24     0.33   -0.31-0.73
Interlimb coordination
  Alternate kicks
    No. of kicks/20 s                3.8      3.7     0-10
    % of total kicks                26.6     18.4     0-58.8
  Unilateral kicks
    No. of kicks/20 s                0.5      0.8     0-2
    % of total kicks                 4.1      6.0     0-12.5
  Synchronous kicks
    No. of kicks/20 s                7.0      3.4     3-12
    % of total kicks                69.4     19.5    41.2-100

                                    4 Months Corrected Age

                                    Normal Walking Attainment
                                    (n = 38)

Variable                            [bar]X   SD      Range

Kick frequency (cycles/min)         16.5     11.6     0-60
Kick amplitude (([degrees]))        52.1     11.6    23.7-80.9
Phase duration (s)
  Flexion phase                      0.71     0.23    0.32-1.63
  Intra-kick pause                   2.72     4.14    0.02-16.4
  Extension phase                    0.56     0.27    0.25-2.45
  Inter-kick pause                   1.17     1.22    0.02-5.15
Interjoint coordination (Z score)
  Hip-knee coupling                  0.69     0.49   -0.57-1.93
  Hip-ankle coupling                 0.49     0.39   -0.18-1.56
  Knee-ankle coupling                0.27     0.42   -1.02-1.64
Interlimb coordination
  Alternate kicks
    No. of kicks/20 s                1.7      2.1     0-8
    % of total kicks                21.9     21.5     0-72.7
  Unilateral kicks
    No. of kicks/20 s                1.1      1.3     0-5
    % of total kicks                21.3     27.7     0-100
  Synchronous kicks
    No. of kicks/20 s                3.5      2.9     0-13
    % of total kicks                56.7     28.2     0-100

Table 2.
Number and Percentage of Infants With Mild to Severe Delay
in Walking Attainment and Infants With Normal Walking
Attainment Who Adopted Alternate, Unilateral, and Synchronous
Kicks as the Predominant Interlimb Coordination Pattern at
2 and 4 Months Corrected Age

                    2 Months Corrected Age

                    Mild to Severe       Normal Walking
Predominant         Delay in Walking     Attainment
Pattern             Attainment (n = 6)   (n = 38)

Alternate kicks     2 (33)               6 (16)
Unilateral kicks    4 (66)               16 (42)
Synchronous kicks   0 (0)                16 (42)

                    4 Months Corrected Age

                    Mild to Severe       Normal Walking
Predominant         Delay in Walking     Attainment
Pattern             Attainment (n = 6)   (n = 38)

Alternate kicks     1 (17)               7 (18)
Unilateral kicks    0 (0)                6 (16)
Synchronous kicks   5 (83)               25 (66)

Table 3.
Proportional-Hazards Regression Analysis of Individual Kicking
Variables for Age of Walking Attainment (a)

                            2 Months Corrected Age

                                                   Multivariate
                            Univariate Model       Model (b)

Variable                    RR         95% CI      RR         95% CI

Kick frequency [per cycle
  increase in minutes)      1.02       0.99-1.03   1.02       0.99-1.03
Kick amplitude (per
  degree increase)          1.01       0.99-1.03   1.01       0.99-1.03
Phase duration (per
    second increase)
  Flexion phase             1.72       0.36-8.24   1.28       0.26-6.38
  Intra-kick pause          1.21 (c)   1.08-1.36   1.18 (c)   1.04-1.32
  Extension phase           3.55       0.79-16.0   1.71       0.34-8.62
  Inter-kick pause          0.95       0.83-1.10   0.95       0.82-1.10
Interjoint coordination
    (per score increase)
  Hip-knee coupling         0.42 (c)   0.24-0.73   0.48 (c)   0.28-0.83
  Hip-ankle coupling        1.07       0.60-1.90   1.14       0.63-2.09
  Knee-ankle coupling       0.87       0.51-1.50   0.85       0.48-1.50
Interlimb coordination
    (per % increase)
  Alternate kicks           1.01       0.99-1.02   1.01       0.99-1.02
  Unilateral kicks          0.99 (c)   0.98-1.00   0.99 (c)   0.98-1.00
  Synchronous kicks         1.01 (c)   1.00-1.02   1.02 (c)   1.00-1.03

                            4 Months Corrected Age

                                                   Multivariate
                            Univariate Model       Model (b)

Variable                    RR         95% CI      RR         95% CI

Kick frequency [per cycle
  increase in minutes)      0.97 (c)   0.95-0.99   0.97 (c)   0.95-0.99
Kick amplitude (per
  degree increase)          1.01       0.99-1.03   1.01       0.99-1.03
Phase duration (per
    second increase)
  Flexion phase             3.80 (c)   1.45-9.93   2.56       0.91-7.16
  Intra-kick pause          1.10 (c)   1.04-1.17   1.08 (c)   1.02-1.14
  Extension phase           1.78       0.67-4.72   1.55       0.55-4.34
  Inter-kick pause          1.19       0.97-1.44   1.16       0.96-1.41
Interjoint coordination
    (per score increase)
  Hip-knee coupling         1.20       0.69-2.09   1.29       0.75-2.23
  Hip-ankle coupling        1.03       0.59-1.80   1.33       0.72-2.44
  Knee-ankle coupling       1.03       0.58-1.84   1.34       0.70-2.57
Interlimb coordination
    (per % increase)
  Alternate kicks           0.99       0.98-1.01   1.00       0.99-1.01
  Unilateral kicks          1.02 (c)   1.01-1.02   1.01 (c)   1.00-1.02
  Synchronous kicks         0.99 (c)   0.98-1.00   0.99 (c)   0.98-1.00

(a) RR = rate ratio, CI = confidence interval.

(b) RR was adjusted for prematurity (full-term versus
preterm) and gestational age (per week increase).

(c) p<.05.

* Peak Performance Technologies Inc, 7388 S Revere Pkwy, #601, Englewood, CO 80112.

([dagger]) Panasonic Broadcast & Television System Co, 1 Panasonic Way, Secaucus, NJ 07094.

([double dagger]) Horita Co, PO Box 3993, Mission Viejo, CA 92680.

([section]) Microsoft Corp, One Microsoft Way, Redmond, WA 98052-6399.

([parallel]) SAS Institute Inc, PO Box 800, Cary, NC 27511.

References

(1) Palta M, Weinstein MR, McGuinness G, et al. A population study: mortality and morbidity after availability of surfactant therapy. Arch Pediatr Adolesc Med. 1994;148:1295-1301.

(2) Stevenson DK, Wright LL, Lemons JA, et al. Very low birth weight outcomes of the National Institute of Child Health and Human Development Neonatal Research Network, January 1993 through December 1994. Am J Obstet Gynecol. 1998;179:1632-1639.

(3) Ornstein M, Ohlsson A, Edmonds J, et al. Neonatal follow-up of very low birthweight/extremely low birthweight infants to school age: a critical overview. Acta Pediatr Scand. 1991;80:741-748.

(4) Hack M, Flannery DJ, Schluchter M, et al. Outcomes in young adulthood for very-low-birth-weight infants. N Engl J Med. 2002;346: 7149-7157.

(5) Parker SJ, Zahr LK, Cole JG, et al. Outcome after developmental intervention in the neonatal intensive care unit for mothers of preterm infants with low socioeconomic status. J Pediatr. 1992;120:780-785.

(6) Als H, Lawhon G, Duffy FH, et al. Individualized developmental care for the very low-birth-weight preterm infants: medical and neurofunctional effects. JAMA. 1994;272:853-858.

(7) Avon Premature Infant Project. Randomized trial of parental support for families with very preterm children. Arch Dis Child Fetal Neonatal Ed. 1998;79:F4-F11.

(8) Prechtl HFR. Qualitative changes of spontaneous movements in fetus and preterm infants are a marker of neurological dysfunction. Early Hum Dev. 1990;23:151-158.

(9) Piek JP. The influence of preterm birth on early motor development. In; Piek JP, ed. Motor Behavior and Human Skill. Champaign, Ill: Human Kinetics Inc; 1998:233-251.

(10) Ferrari F, Cioni G, Prechtl HFR. Qualitative changes of general movements in preterm infants with brain lesions. Early Hum Dev. 1990;23:193-231.

(11) Albers S, Jorch G. Prognostic significance of spontaneous motility in very immature preterm infants under intensive care treatment. Biol Neonate. 1994;66:182-187.

(12) Bos AF, Martijn A, van Asperen RM, et al. Qualitative assessment of general movements in high-risk preterm infants with chronic lung disease requiring dexamethasone therapy. J Pediatr. 1998;132:300-306.

(13) Thelen E, Bradshaw G, Ward JA. Spontaneous kicking in month-old infants: manifestation of a human central locomotor program. Behav Neural Biol. 1981;32:45-53.

(14) Thelen E, Ridley-Johnson R, Fisher DM. Shifting patterns of bilateral coordination and lateral dominance in the leg movements of young infants. Dev Psychobiol. 1983;16:29-46.

(15) Thelen E, Fisher DM. The organization of spontaneous leg movements in newborn infants. Dev Psychol. 1983;18:760-775.

(16) Heriza CB. Comparison of leg movements in preterm infants at term with healthy full-term infants. Phys Ther. 1988;68:1687-1693.

(17) Heriza CB. Motor development: traditional and contemporary theories. In: Lister MJ, ed. Contemporary, Management of Motor Control Problems. Proceedings of the II STEP Conference. Alexandria, Va: Foundation for Physical Therapy; 1991:99-126.

(18) Yokochi K, Inukai K, Hosoe A, et al. Leg movements in the supine position of infants with spastic diplegia. Dev Med Child Neurol. 1991;33: 903-907.

(19) Droit S, Boldrini A, Cioni G. Rhythmical leg movements in low-risk and brain-damaged preterm infants. Early Hum Dev. 1996;44:201-213.

(20) Geerdink JJ, Hopkins B, Beek WJ, et al. The organization of leg movements in preterm and fun-term infants after term age. Dev Psychobiol. 1996;29:335-351.

(21) Piek JP, Gasson N. Spontaneous kicking in fullterm and preterm infants: are there leg asymmetrics? Hum Mov Sci. 1999;18:377-395.

(22) van der Heide JC, Paolicelli PB, Boldrini A, et al. Kinematic and qualitative analysis of lower-extremity movements in preterm infants with brain lesions. Phys Ther. 1999;79:546-557.

(23) Vaal J, van Soest AJ, Hopkins B, et al. Development of spontaneous leg movements in infants with and without periventricular leukomalacia. Exp Brain Res. 2000;135:94-105.

(24) Jeng SF, Chen LC, Yau KIT. Kinematic analysis of kicking movements in preterm infants with very low birth weight and full-term infants. Phys Ther. 2002;82:148-159.

(25) O'Shea MT, Preisser JS, Klinepeter KL, et al. Trends in mortality and cerebral palsy in a geographically based cohort of very low birth weight neonates between 1982 to 1994. Pediatrics. 1998;101:642-647.

(26) Cooke RWI. Trends in incidence of cranial ultrasound lesions and cerebral palsy in very low birthweight infants 1982-1993. Arch Dis Child Fetal Neonatal Ed. 1999;80:F115-F117.

(27) Marlow N, Roberts L, Cooke R. Outcome at 8 years for children with birth weights of 1250 g or less. Arch Dis Child. 1993;68:286-290.

(28) Whitaker AH, van Rossen R, Feldman JF, et al. Psychiatric outcomes in low birth weight children at age 6 years: relation to neonatal cranial ultrasound abnormalities. Arch Gen Psychiatry. 1997;54:847-856.

(29) Stjernqvist K, Svenningsen NW. Ten year follow-up of children born before 29 gestational weeks: health, cognitive development, behavior and school achievement. Acta Paediatr. 1999;88:557-562.

(30) Amiel-Tison C, Njiokiktjien C, Vaivre-Douret L, et al. Relation of early neuromotor and cranial signs with neuropsychological outcome at 4 years. Brain Dev. 1996;18:280-286.

(31) Jeng SF, Yau KIT, Liao HF, et al. Prognostic factors for walking attainment in very, low-birthweight preterm infants. Early Hum Dev. 2000;59:159-173.

(32) Vohr B, Wright LL, Dusick AM, et al. Neurodevelopmental and functional outcomes of extremely low birth weight infants in the National Institute of Child Health and Human Development Neonatal Research Network, 1993-1994. Pediatrics. 2000;105:1216-1226.

(33) Johnson A, Goddard O, Ashurst H. Is late walking a marker of morbidity? Arch Dis Child. 1990;65:486-488.

(34) de Groot CJ, Hopkins B. An instrument to measure independent walking: are there differences between preterm and fullterm infants? J Child Neurol. 1997;12:37-41.

(35) Chaplais JD, Macfarlane JA. A review of 404 "late walkers." Arch Dis Child, 1984;59:512-516.

(36) Allen MC, Alexander GR. Screening for cerebral palsy in preterm infants: delay criteria for motor milestone attainment. J Perinatol. 1994;14:190-193.

(37) Papile LA, Munsick-Bruno G, Schaefer A. Relationship of cerebral intraventricular hemorrhage and early childhood neurologic handicaps. J Pediatr. 1983;103:273-277.

(38) de Vries LS, Groendaal F, Meiners LC. Ischemic lesions in the preterm brain. In: Rutherford M, ed. MRI of the Neonatal Brain. London, United Kingdom: WB Saunders Co; 2002:155-169.

(39) Skidmore MD, Rivers A, Hack M. Increased risk of cerebral palsy among very low-birthweight infants with chronic lung disease. Dev Med Child Neurol. 1990;32:325-332.

(40) Brazelton TB. Neonatal Behavioral Assessment Scale. Clinics in Developmental Medicine No. 68. Philadelphia, Pa: JB Lippincott Co; 1984.

(41) Piek JP, Carman RC. Developmental profiles of spontaneous movements in infancy. Early Hum Dev. 1994;39:109-126.

(42) Thelen E. Developmental origins of motor coordination: leg movements in human infants. Dev Psychobiol. 1985;18:1-22.

(43) Lee ET. Statistical Methods for Survival Data Analysis. 2nd ed. New York, NY: John Wiley & Sons Inc; 1992.

(44) Grossman HJ. Classification in Mental Retardation. Washington, DC: American Association on Mental Retardation; 1983.

(45) van Wulfften T, Hopkins B. Development of the infant's social competence during early face-to-face interaction: a longitudinal study. In: Prechtl H, ed. Continuity of Neural Functions From Prenatal to Postnatal Life. London, United Kingdom: Spastic International Medical Publications; 1984:198-219.

(46) Touwen B. Neurological Development in Infancy. London, United Kingdom: SIMP and Heinemann; 1976.

(47) von Hofsten C. Developmental change in the organization of prereaching movements. Dev Psychol. 1984;20:378-388.

(48) Vaal J. spontaneous Kicking: On the Organization of Spontaneous Leg Movements in Early Human Development [dissertation]. Amsterdam, the Netherlands: Vrije University; 2001.

(49) Papousek H, Papousek M. Qualitative transitions in integrative processes during the first trimester of human postpartum life. In: Prechtl HFR, ed. Continuity of Neural Functions From Prenatal to Postnatal Life. London, United Kingdom: Spastic International Medical Publications; 1984:220-244.

(50) Beek P, Beek W. Tools for constructing dynamical models of rhythmic movement. Hum Mov Sci. 1988;7:301-342.

(51) Jeng SF, Holt KG, Fetters L, et al. Self-optimization of walking in non-disabled children and children with spastic hemiplegic cerebral palsy. J Motor Behav. 1996;28:15-27.

SF Jeng, PT, ScD, is Associate Professor, School and Graduate Institute of Physical Therapy, College of Medicine, National Taiwan University, and Adjunct Physical Therapist, Department of Rehabilitation and Physical Medicine, National Taiwan University, Hospital, No. 7 Chun-Shan South Rd, Taipei, Taiwan ([e-mail]). Address all correspondence to Dr Jeng.

LC Chen, PT, MS, is a doctoral student, Department of Kinesiology; University of Maryland, College Park, Md.

KI Tsou, MD, is Professor, School of Medicine, College of Medicine, Fu-Jen Catholic University, and Deputy Director of Education and Neonatologist, CardinaI Tien Hospital, Taipei, Taiwan.

WJ Chen, MD, ScD, is Professor, Institute of Epidemiology, College of Public Health, National Taiwan University, and Adjunct Research Fellow, Department of Psychiatry, National Taiwan University Hospital.

HJ Luo, PT, MS, is Physical Therapist, Department of Rehabilitation and Physical Medicine, Kaohsiung Chang-Gung Memorial Hospital, Kaohsiung, Taiwan.

Dr Jeng and Ms Li-Chiou Chen provided concept/idea/research design and data collection. Dr Jeng provided writing and project management. Ms Li-Chiou Chen, Dr Wei Chen and Mr Luo provided data analysis. Dr Jeng and Dr Tsou provided fund procurement. Dr Tsou provided subjects, and Mr Luo provided clerical support. Ms Li-Chiou Chen and Dr Wei Chen provided consultation (including review of manuscript before submission). The authors thank the infants and their parents for their participation in this study; Ms Hsian-Feng Chen and Ms Shiu-Ying Yu for their assistance in data collection; Mr Hong-Nan Chen for his assistance in program development; and Dr Linda Fetters, Dr Tung-Wu Lu, and Ms Hua-Fang Liao for their guidance during the study.

This study was approved by the Institutional Review Committee of the College of Medicine, National Taiwan University.

This work was supported by a grant from the National Health Research Institute (DOH 88-HR-619) of the Department of Health and a grant from the National Science Council (NSC90-2314-B002-307) in Taiwan.

This article was received June 3, 2003, and was accepted August 20, 2003.