Effect of practice on a novel task--walking on a treadmill: preadolescents with and without down syndrome.

People with Down syndrome (DS) exhibit hypotonia, ligamentous laxity, and a reduced capacity to produce muscle force. They also demonstrate more variable movement patterns than their peers with typical development (TD). In a previous study, Ulrich et al (1) showed that preadolescents with DS use significantly higher levels of stiffness and impulse (forcing) compared with preadolescents with TD when walking on a treadmill and use higher impulse but not stiffness during over-ground walking. We propose that children with DS increase stiffness in novel contexts, such as the treadmill, to compensate for their perceived instability largely due to joint laxity and that they will reduce stiffness following practice of the task.

We know that task practice is imperative to acquire the successful performance of a new skill; however, we know little about the effect of practice on the use of dynamic resources to accomplish well-established functional patterns, such as walking, under novel conditions. Almeida et al (2) asked people with DS, 15 to 35 years of age, to practice a 1-degree of freedom reaching task consisting of 1,100 repetitions of elbow flexion over a period of 2 weeks. This practice was sufficient to shift both kinematic and electromyographic (EMG) patterns to achieve similar levels of motor performance to that described in the literature for individuals who are neurologically typical. Furthermore, performance improvements transferred to nontrained distances and to a different starting position for the reaching task.

In a finger-force task, young adults with DS performed 3 practice sessions consisting of 36 practice trials each and generated improvements in force control. (3) The relationship between the total force variance and the sum of the variances in participants with DS became closer to the relationship observed in those with TD. Both studies involved the use of massed practice sessions distributed across time to learn a new skill.

In longer-duration interventions, children with DS have demonstrated improved performance in both standardized measures and kinematic variables. Wang and Ju (4) asked 3- to 6-year-old children with DS to practice jumping for 3 sessions per week for 6 weeks. During each 30-minute session, participants did warm-up exercises and practiced jumping while receiving instructions for vertical and horizontal jumping. Participants' scores for floor walk, beam walk, and horizontal and vertical jumping improved more than the scores of children with TD who did not receive training. In another intervention study, (5) children with DS participated in a stair-climbing intervention program conducted biweekly for 12 weeks during a 3-hour gymnastics class. They showed kinematic changes in both ascent and descent phases. (5)

Taken together, these findings suggest that behavior and skill performance in people with DS can be modified over a period of days to weeks through task-specific practice sessions distributed over time for both novel and more practiced tasks. Ulrich et al (2) showed that preadolescents with DS, who inherently have low tone (velocity-dependent resistance to stretch) with high joint laxity, produce a functional solution to walking in a novel situation (ie, on a motorized treadmill). That is, they adapt their dynamic resources (stiffness and impulse) to fit the task, given their unique capacities and experience. We propose that, given sufficient, task-specific practice distributed over time in a motivating environment, they will show further adaptation, in this case, in the opposite direction (downward) by decreasing stiffness and impulse. Reducing stiffness and impulse would result in more efficient performance of the activity of treadmill walking and demonstrates the ability of preadolescents with DS to adapt their dynamic resources and improve performance with task-specific practice. The purpose of this study was to explore changes in stiffness and impulse values of participants with DS after sufficient, task-specific practice distributed over time in a motivating environment.

Method

Participants

Sixteen preadolescents between the ages of 8 and 10 years participated in the study, 8 with DS (trisomy 21) and 8 with TD. One additional participant with DS was tested and excluded from all analysis because she refused to let go of the handlebar while walking on the treadmill at the fastest speed. We recruited participants with DS from parent support groups in southeast Michigan and northwest Ohio and through flyers. Participants with TD were recruited from local community groups using flyers. We chose preadolescents because we wanted to examine optimal walking patterns. At this age, participants have had significant practice (6-9 years) of this functional activity, and changes in anthropometrics would have been fairly steady for several years.

Procedure

We explained all procedures to parents and participants, each of whom signed a consent form prior to participation. Participants came to our laboratory on 6 occasions. Pretest and posttest data collections took place at visits 2 and 6. The average time between pretest and posttest visits was 12 days, and most participants were within a 7- to 15-day range. There were some participants who, for reasons of extended travel time to the laboratory, had shorter or longer intervals between pretest and posttest visits (range=6-30 days). For all participants, however, practice visits were spread evenly across the interval. Test sessions lasted approximately 1.5 hours, and practice sessions lasted 30 to 45 minutes.

A familiarity session occurred prior to the pretest session. This session was particularly necessary for participants with DS to ensure their comfort with the research team and protocol. On this day, participants practiced walking over ground at their preferred speed. Participants experienced wearing test clothing, the attachment and removal of reflective markers and EMG electrodes, and walking over our GAITRite mat * and wood walkway with markers and cables attached. At pretest and posttest visits, participants walked over ground at their preferred speed and on a treadmill at 40%, 75%, and 110% of their over-ground walking speed. Each practice session consisted of walking at their 75% speed for 12 minutes.

When the participants arrived in the laboratory for pretest and posttest data collections, they changed into a bathing suit. We marked the skin surface of each site requiring a reflective marker with hypoallergenic eyebrow pencil and attached markers (2.5-cm diameter) to the lateral surface on each side of the body (temporomandibular joint, acromion, lateral humeral epicondyle, wrist at the styloid process, greater trochanter, femoral condyle, 10 cm above the lateral malleolus, bony prominence of the heel, and third metatarsophalangeal joint). We cleaned the skin surface using alcohol pads and placed preamplified bipolar EMG electrodes over the muscle bellies of the tibialis anterior, gastrocnemius, quadriceps femoris (rectus femoris), and hamstring (biceps femoris) muscles of the fight lower extremity. We recorded a baseline trial of resting EMG activity while participants sat quietly with their feet supported on a firm surface. For the purposes of this report, EMG results will not be discussed.

Estimates of segmental mass were made using total body mass, body segment lengths, age, and Jensen's regression equations. (6) Body center of mass was calculated from a 10-segment model (2 feet, 2 shanks, 2 thighs, 2 upper arms, head and neck, and trunk). We directly measured weight (Healthometer beam scale ([dagger]) and body segment lengths: standing height (GPM anthropometer ([double dagger])), sitting height (while seated on wooden bench), upper arm (arm extended, acromion-lateral humeral epicondyle), arm (elbow and wrist flexed, radial head-styloid process), thigh (greater trochanter-lateral condyle), shank (while seated with left leg crossed over right knee, tibial condyle to medial malleolus), and foot (heel-tip of longest toe). All body segment measurements were taken on left side. We also administered the balance subtest of the long form of the Bruininks-Oseretsky Test of Motor Proficiency (BOTMP) ([section]) to quantify the balance deficits often referenced in those with DS (4,7) and further describe our participants. Anthropometric measurements and BOTMP scores were obtained by the principal investigator (BDU) or a physical therapist (BAS), both with extensive training and experience with these measures, although formal rater reliability testing was not conducted.

Over-ground walking. For over-ground testing trials, participants walked across the laboratory at their self-selected comfortable speed over a custom-built 6.0-m walkway with a Bertec ([parallel]) forceplate embedded in it. As they walked back to the starting position, participants walked over a 5.3-m GAITRite mat. Participants proceeded to walk multiple times back and forth across the room. In one direction, they walked over the walkway and forceplate and through the calibrated space for data collection; in the other direction, they walked over the GAITRite mat. They had adequate room both to accelerate before encountering the calibrated space or GAITRite mat and to decelerate before turning around.

Participants performed practice trials until they showed consistent walking speeds across trials and appeared comfortable with the protocol (approximately 3-6 practice trials). We then collected data for a minimum of 5 trials in which a trained observer verified that the participants contacted the forceplate with one foot only and maintained their comfortable speed. From the test trials, we used GAITRite software to calculate average walking speed, which we used to adjust the belt speed to each individual's self-selected speed for the treadmill phase of testing.

Operational definition of comfortable speed on the treadmill.

It can be difficult for preadolescents with DS to reliably report their comfortable walking speed when on a treadmill. Therefore, based on our pilot work and the work of others, (8) we operationally defined a comfortable treadmill speed for all participants as 75% of their comfortable speed during over-ground walking, as comfortable speeds on a treadmill are slower than during over-ground walking. (9) Our pilot work demonstrated that self-selected speeds were stable across multiple visits to the laboratory.

Treadmill walking. During treadmill testing trials, participants walked on our treadmill (Parker brand ([#])) at 40%, 75%, and 110% of their over-ground speed. Treadmill trials consisted of a 60-second trial and a 30-second trial at each speed. Order of presentation was from slowest to fastest. Preadolescents with DS are not completely comfortable with the fastest speeds; therefore, we could not randomize speeds without exceeding their level of tolerance. During the 60-second trial, participants walked for 15 seconds with both hands on the treadmill handrail, for 15 seconds with one hand on the handrail, and for 30 seconds without holding onto the treadmill bar. The 30-second trial consisted only of walking without holding on to the treadmill bar.

Practice sessions. Between pretest and posttest visits, participants completed 4 practice sessions (days 2 [after pretest], 3, 4, and 5) consisting of 12 repetitions of the 60-second protocol of treadmill walking at their 75% speed. The first 4 preadolescents tested in each group practiced while wearing street clothes and with bare feet. The second 4 preadolescents also practiced barefoot, but wore bathing suits, reflective markers, and EMG electrodes in an effort to recreate the complexity of the testing situation. With the first group, we noted that the preparation for the test sessions was substantially longer and more involved than preparation for the practice sessions, and we did not want the act of putting on reflective markers, electrodes, and tights only for the test sessions to affect subsequent behavior. We also changed the orientation of the treadmill (facing north, south, east, or west) in the laboratory for each practice session to reduce the monotony of the task. For all preadolescents, we used conversation about their interests, pictures of cartoon characters, verbal and visual encouragement, and so on to motivate them.

Equipment and laboratory setup.

Analog signals from GAITRite mat sensors were sent to our laptop computer at 60 Hz. We used GAITRite software to edit usable steps and to calculate over-ground gait parameters. The GAITRite system measures step width as the perpendicular distance from the center of one foot to the line of progression of the center of the other foot. We used a 6-camera Peak Motus real-time system** to collect reflective marker positions in space at a sampling rate of 60 Hz. Two cameras were placed on tripods on each side of the walkway/ treadmill; another camera was placed in front and one in back of the walkway and were ceiling mounted. Prior to each data collection session, we used a fixed calibration frame and wand to set up a viewing space of 2.9 X 1.4 X 1.85 m. Acceptable summed measurement error for each calibration was set at 0.008 m. In addition, a video camera was placed to the side and slightly forward of the center of the walkway to provide visual confirmation of any unusual patterns in the data.

After over-ground data collection, the GAITRite mat and wooden walkway were removed and the treadmill was moved into the middle of the calibrated space for subsequent data collection. Treadmill speed control allowed precision to 2 decimal places (km/h).

Data reduction and analysis. Raw kinematic data were converted to 3-dimensional (3D) data via the Peak system software and filtered with a second order Butterworth illter at a cutoff frequency of 6 Hz. We then analyzed the data using custom-written software programs written in MATLAB. ([dagger])([dagger]) Initial foot contact and toe-off for each stride were identified from the kinematic data based on the peaks of vertical acceleration of heel markers and the peaks of horizontal accelerations of toe markers. (10) We also determined the 3D location of center of mass and calculated step width and stride length (both absolute and normalized to leg length) on the treadmill.

We used an escapement-driven damped inverted pendulum and spring model (Fig. 1) to derive algorithms to estimate global stiffness and impulse. (1,11) In the inverted pendulum model of walking, the center of mass of the body is the mass of the pendulum as it travels over the stance leg (arm of the pendulum). Following push-off, the mass rises to the top of its are and then falls down the other side due to gravity. The inverted pendulum model applies to the stance phase, and the impulse values are calculated relative to push-off. Stiffness represents a conservation mechanism of soft tissue (storage and return of elastic energy of muscles and tendons). The model captures the global behavior of the system and does not identify the specific sources of stiffness.

[FIGURE 1 OMITTED]

Angular impulse is energy released into the system by the muscles to replace what is lost and to help maintain forward motion of the center of mass. During gait, energy is released into the system through push-off, with higher impulse values representing more force per unit of time. We used custom-written programs and both anthropometric and kinematic data to generate the parameter values. Stride frequency, step width, stride length, and gait speed were converted to dimensionless values (Appendix) in order to account for the effect of differences in body size within and between the 2 groups. For expanded information concerning algorithms used and their derivation, see Ulrich et al. (1)

The mean values for 12 strides for each child for each condition (for all available steps in the event, there were fewer than 12 usable strides for technical reasons) were entered into the analyses. We used SPSS (version 14.0) ([double dagger])([double dagger]) for statistical testing. We set our alpha level of significance at .05 and used Bonferroni adjustment for multiple comparisons for all repeated-measures analyses.

Results

As anticipated, preadolescents with DS were significantly shorter (but not lighter) and had shorter limbs and smaller limb-to-trunk ratios than those with TD. Three one-way (group) multivariate analyses of variance (MANOVAs) were calculated, each with related sets of dependent variables. Set 1 variables were 7 body segment lengths, set 2 variables were 3 body segment ratios, and set 3 variables consisted of height, weight, and body mass index (BMI). Significant main effects resulted for 2 of the 3 analyses (body segment ratios: Wilks lambda=.420, F=5.53, df= 3,12, P=.013; height, weight, and BMI: Wilks lambda=.199, F=16.13, df=3,12, P<.001), and the data for body segment lengths were close to significance (Wilks lambda=.256, F=3.32, df=7,8, P=.057).

Follow-up univariate analyses revealed significant group differences for all limb segments (upper arm: F= 15.51, df=1,14, P=.001; forearm: F=15.40, df=l,14, P=.002; thigh: F=15.24, df=l,14, P=.002; shank: F=21.72, df=l,14, P<.001; and foot: F=9.47, df=1,14, P=.008), for 2 of the 3 body segment ratios (upper extremity/trunk: F=11.50, df=1,14, P=.004; lower extremity/ trunk: F=12.73, df=l,14, P=.003), and for height (F=24.28, df=l,14, P<.001) and BMI (F=6.76, df=1,14, P=.021). Group mean values are shown in Table 1.

In addition, as expected, the group with DS scored significantly lower on the balance assessment than their peers (one-way [group] analysis of variance [ANOVA], F=211.30, df=l,15, P.001). The raw mean score for participants with DS was 6.50, and the mean score for participants with TD was 24.25. During each practice session, participants performed twelve 60-second repetitions of treadmill walking, resulting in an average of 891 (group with DS) and 692 (group with TD) strides per leg per session (3,564 and 2,768 repeated stride cycles per leg overall, respectively).

Stiffness

We used a 2 (group) X 2 (time) X 3 (speed) ANOVA with repeated measures on time for treadmill conditions. We obtained a significant group main effect (F=35.05, df=1,42.53, P< .001), a significant speed main effect (F=54.66, df=2,42.51, P<.001), and a significant time effect (F=24.42, df= 1,41.24, P<.001). There were no significant interaction effects. Follow-up univariate analysis and inspection of means (shown in Fig. 2) revealed that the group with DS had higher stiffness than the group with TD, that stiffness increased with increased treadmill speed, and that stiffness decreased with practice. Participants with DS demonstrated higher stiffness in all conditions at pretest (40% of over-ground walking speed: F=5.98, df= 1,72.32, P=.017; 75% of over-ground walking speed: F=9.74, df=1,73.98, P=.003; 110% of over-ground walking speed: F=12.87, df=1,73.96, P=.001) and in the 40% of over-ground walking speed condition (F=11.33, df= 1,72.32, P=.001) and in the 75% of over-ground walking speed condition (F= 5.30, df=1,73.98, P=.024) at posttest.

[FIGURE 2 OMITTED]

Impulse

We used a 2 (group) X 2 (time) X 3 (speed) ANOVA with repeated measures on time for treadmill conditions. We obtained a significant group main effect (F=52.42, df= 1,42.31, P<.001), a significant speed main effect (F=15.32, df=2,42.29, P<.001), a significant time main effect (F=28.70, df=1,41.40, P<.001) and a significant speed X time interaction (F=4.6, df=2,41.37, P= .016). Group means showed that preadolescents with DS produced higher impulse values than their peers with TD (Figs. 3a and 3b). Figure 3 illustrates the speed X time interaction and shows that, while impulse decreased from pretest to posttest, the amount of decrease varied across speeds, with the fastest speed producing a small change while the moderate and slow speeds decreased more.

[FIGURE 3 OMITTED]

Variability of Stiffness and Impulse Values

We used two 2 (group) X 2 (time) X 3 (speed) ANOVAs with repeated measures on time for treadmill conditions to statistically test our observation that variability of stiffness and impulse values decreased with practice (Figs. 2 and 3). Dependent variables were the stiffness and impulse standard deviations at pretest and posttest for each participant.

For stiffness standard deviations, we obtained a significant group main effect (F=40.16, df=1,73.52, P<.001) and a nearly significant time effect (F=3.95, df=1,73.57, P=.051). There was not a significant speed effect or any interaction effects. Inspection of means revealed that participants with DS were more variable than their peers with TD at pretest and posttest and that both groups tended to decrease variability between pretest and post-test sessions.

In regard to impulse standard deviation, we obtained a significant group main effect (F=15.21, df=1,73.66, P<.001) and a significant time effect (F=4.51, df=l,73.71, P=.037). As with stiffness variability, the speed effect and all interactions were not significant. Inspection of means showed that participants with DS were more variable than their peers with TD at pretest and posttest visits and that both groups decreased variability between pretest and posttest.

Overground Gait Characteristics

We used a 2 (group) X 2 (time) MANOVA with repeated measures on time to compare preadolescents' overground walking characteristics (dependent variables were gait speed, stride frequency, stride length, step width, percentage of stance, percentage of double-support phase, dimensionless speed, dimensionless stride frequency, dimensionless stride length, and dimensionless step width). We obtained a significant group main effect (Wilks lambda=.027, F=18.27, df= 10,5, P=.003). There was not a significant time effect or group X time interaction. Post hoc ANOVA results suggested that the significant group effect was largely due to differences in stride length (F=16.37, df=l,14, P=.001), step width (F=31.54, df= 1,14, P<.001), and gait speed (F= 5.51, df=l,14, P=.034); preadolescents with DS walked slower with shorter strides and wider step widths compared with preadolescents with TD. When these variables were converted to dimensionless values, only step width continued to show a significant difference between the 2 groups (F=88.07, df=1,14, P<.001) (Tab. 2).

Treadmill Gait Characteristics

We used a separate 2 (group) X 2 (time) X 3 (speed) ANOVA with repeated measures on time for each of 3 dependent variables (dimensionless step width, dimensionless stride length, and dimensionless stride frequency).

Step width. We obtained a significant group effect (F=153.73, df=1,41.752, P<.001) and speed effect (F=4.72, df=2,41.74, P=.014), with a significant group X time interaction (F=6.69, df=1,38.61, P=.014) and a trend toward a group X speed interaction (F=2.97, df=2,41.74, P=.062). Inspection of means (Tab. 2) revealed that, although participants with DS demonstrated a larger decrease in step width with practice, especially at the 110% treadmill speed, they continued to produce wider step widths than the group with TD in all conditions. Participants with TD did not show a decrease is step width with practice.

Stride length. We found a significant group effect (F=7.28, df= 1,42.62, P=.010), speed effect (F=85.30, df=2,42.62, P<.001), and time effect (F=31.29, df=1,37.92, P<.001). There were 2 significant interactions: speed X time (F=3.49, df=2,37.89, P=.041) and group X speed X time (F=4.43, df=2,37.89, P=.019). Follow-up inspection of means (Tab. 2) revealed that preadolescents with DS showed less increase in stride length than their peers with TD at the 110% treadmill speed with practice and greater increase in stride length than their peers with TD at the 40% treadmill speed with practice.

Stride frequency. We found a significant group effect (F=19.33, df=1,42.48, P<.001), speed effect (F=49.86, df=2,42.46, P<.001), time effect (F=24.01, df=1,40.34, P<.001), and speed X time interaction (F=3.26, df=2,40.29, P=.049). Follow up inspection of means (Tab. 2) revealed that the group with DS demonstrated higher stride frequency than the group with TD and that all participants decreased their stride frequency more at slower speeds.

Protocols

There was no difference in results between the original and modified protocols. We used a 2 (protocol) X 2 (group) multivariate ANOVA with each participant's change in impulse and stiffness as dependent variables. There were no significant protocol main effects or protocol x group interactions at any speed. Of the 4 participants with DS who showed the largest changes in stiffness and impulse with practice, 2 completed the original protocol and 2 completed the modified protocol.

Discussion

Overall, our results suggest that preadolescents with DS initially increase stiffness and impulse when challenged with a novel task and then reduce these behavioral resources following practice of walking on a treadmill. We found that practice led to stiffness and impulse values more like the pretest values of their peers with TD. Preadolescents with TD also reduced their stiffness and impulse values following practice, although to a lesser extent. We posit that, in our protocol, practice overcame task novelty for both groups, primarily in terms of the unique context of the treadmill, but also in the "unique" laboratory context. Practice allows the performer to explore and settle into more efficient and stable solutions, as demonstrated by lower stiffness and impulse values and less variability at posttest.

Although practice had the effect of enabling preadolescents with DS to generate stiffness and impulse values more similar to those of their peers with TD than before practice, they nevertheless showed unique patterns of change rather than uniform responses across all speeds. Two quite distinctly different points were their significantly higher reduction in impulse at 40% speed and in stiffness at 110% speed. Group similarities and differences in kinematic responses seem to reflect this unique pattern of underlying kinetic responses. Participants with TD increased stride length and decreased step width as treadmill speed increased. Participants with DS also showed these changes in gait parameters, although, even when normalized to leg length, step width remained wider than that of their peers with TD. With practice, participants with DS decreased step width most at the 110% treadmill speed. The larger reduction in stiffness they showed at this same speed may reflect their particular sense, during pretest, of the potential for this fastest speed to cause balance control problems for them.

As we reassured them of their safety and as they practiced being on the treadmill over multiple sessions, their confidence in their ability to retain postural control may have reduced their need for a control strategy optimizing so dominantly on stability, thus reducing stiffness and step width. At the slowest treadmill speed (40%), confidence that they could stabilize their performance was relatively high from the outset, but optimal energy input (impulse) took practice time to discover. With sufficient practice, they learned to put less energy into each step, particularly at the slowest speed, becoming more efficient without modifying stiffness as much as the group with TD.

Along with a decrease in the use of dynamic resources to walk on the treadmill (stiffness and impulse), not surprisingly, variability within each group also generally decreased with practice. Between-group differences persisted over practice time, and participants within each group also showed regression toward their group mean (decreased within-group differences). Although performance consistency improved with practice, the group with DS remained, in most cases, less uniform. This variability reflects the range of solutions to movement problems used by the participants. It is a quantification of different individual behaviors in response to varying levels of both physical and cognitive constraints on performance.

We believe it is important to point out that, within the group with DS, some participants demonstrated stiffness and impulse values close to those of the group with TD, whereas others were distinctly different. Some individual overlap between groups with TD and with DS is expected. (1) However, we looked more closely at the unique nature of this overlap by examining the anthropometric, balance, and age variables that might differentiate those with DS who had stiffness and impulse values most and least like those with TD. Strikingly, the lower-extremity length-to-think length ratio (LE/ trunk ratio) covaried with stiffness and impulse within the group with DS. The 2 preadolescents with DS who demonstrated the highest stiffness and impulse values had the largest LE/trunk ratios (mean LE/trunk ratio=1.46), meaning that preadolescents with DS who are physically proportioned most like their peers with TD (mean LE/trunk ratio=1.44) produce stiffness and impulse values least like them. Conversely, the 2 participants with DS who had the smallest LE/trunk ratios (mean LE/ trunk ratio=1.15) demonstrated stiffness and impulse values closest to those of their peers with TD.

Participants with DS who had the smallest LE/trunk ratios walked slower, while producing longer over-ground and treadmill strides and more narrow treadmill steps, than their peers with DS who had higher LE/trunk ratios. We propose that participants with DS who have shorter legs relative to their trunk dissipate less energy with each stride because their center of mass is relatively lower, leading to lower velocity as the center of mass falls forward prior to foot contact with the surface. Conversely, preadolescents with DS who have longer legs relative to their trunks lose more energy with each stride than their peers; therefore, they put more energy into the system (higher impulse). They also have longer lever arms (limbs) and increase stiffness more in order to stabilize their relatively higher center of mass and the higher forces in their inherently unstable joints.

Beyond exploring this intergroup variability, we also addressed issues of variability at the population level. Our results for this sample of people with DS replicate our previous work with a different sample showing that preadolescents with DS use higher levels of both impulse and stiffness on the treadmill than their peers with TD. (1) Furthermore, as previously demonstrated, participants with DS did not show a significant difference for over-ground stiffness compared with their peers with TD. However, although the means were higher, we did not, as previously, find significantly higher levels of over-ground impulse for participants with DS. We propose that this arises from the normal variability inherent in sampling from a large population.

As Figure 4 illustrates, although sampled from the same population, participants in both groups of the current study tended to walk more slowly and produce less impulse during over-ground walking than the previous sample. Using traditional statistical techniques, about 95% of the population scores should fall within 2 standard deviations of the mean in a normal distribution. Figure 4 illustrates this idea based on both samples and suggests both sets cluster close to each other and overlap but the means fall in sufficiently different locations within the "population" to result in no significant group difference for over-ground impulse values in the current study.

[FIGURE 4 OMITTED]

Preadolescents with DS are less energy efficient than their peers with TD due to their wider step width (see Kuo (12) for a discussion of wider step width providing mediolateral stability at increased energy cost). Other than wider step width, over-ground walking gait parameters, when normalized to body size, are not significantly different between the groups. Our over-ground gait findings for children with TD are consistent with those of Sutherland and colleagues (13) for absolute gait speed, and our results for absolute stride length were roughly twice their results for absolute step length, as might be expected. As for normalized gait speed, our participants with DS and TD walked slightly faster than those of Dusing and Thorpe. (14) It should be noted, however, that slightly different normalization procedures were used (height versus leg length). Our results for gait parameters for preadolescents with DS and TD were consistent with the previous work of Ulrich et al. (1)

By preadolescence, children with DS have had years of practice of stable over-ground ambulation and have maximized their performance in terms of stability and efficiency. This is not the case with walking on a treadmill. The preadolescents in our study did not have previous experience walking on a treadmill, but only those with DS indicated to us that they were less comfortable walking on the treadmill than they were walking over ground. Although participants with DS were, from the outset, able to complete the treadmill task (all except one volunteer), they did so at a higher cost (higher stiffness and impulse values) and with more anxiety than their peers with TD. When we asked them to walk on the treadmill, participants with DS "stiffened up" in an attempt, we argue, to compensate for ligamentous/ joint laxity and balance deficits and thus stabilize their walking in this novel context. With practice, in most cases, preadolescents with DS reduced stiffness and impulse to values closer to those produced by their peers with TD at pretest. We believe extended practice in a supportive and motivating environment is of crucial importance for learning to move efficiently, be it a treadmill walking task, recreational sport, or physical education setting (eg, learning to ice skate).

Brain imaging studies have shown that children and adults with DS have smaller cerebellar volumes than age- and sex-matched controls. (15-17) The cerebellum is involved in balance, coordination, motor control, and motor learning; people with DS are noted to have balance (4,7) and coordination deficits. (3) The correlation between balance and coordination deficits and small cerebellums also has been noted in animal studies using a mouse model of DS. (18,19) Questions, however, remain: Does a smaller cerebellum cause impaired balance? Have people with DS reached their maximum potential; that is, are they limited by the size of their cerebellum? Have they failed to take advantage of the adaptive plasticity of the nervous system and find the optimal solution concerning their inherently different motor system?

Despite poor balance and decreased cerebellar volume, our participants with DS still improved their performance with practice. They were able to explore how their actions changed the performance of the treadmill/person system even in a challenging environment with a moving surface beneath their feet. Note, however, that the task may not have taxed their maximum potential. Although more challenging than the over-ground condition, the context was predictable, unlike the real world in which decisions and feedback comprise a continuum requiring more demanding adaptation. Using the cognitive, cerebellar, or cortex resources available, they were able to take advantage of the feedback generated through repetitive task practice. As Latash wrote, "Practice as repetition of a certain motor task can lead to improvement in motor performance due to two distinct factors ... understanding the explicit instruction but also the relations between subject's actions and movements.... Such comprehension only comes through exploration of the experimental system by actually performing movements." (20(p237))

Motor behavior is influenced by both physical and cognitive factors that contribute to effective practice. Highly skilled individuals, such as professional musicians, practice daily to improve or maintain skilled and efficient performance. These individuals are highly motivated and are able to focus their attention on and detect errors in performance. Due to cognitive limitations, attentional focus during practice is more difficult for people with DS to achieve because they are easily distracted and less able to recognize the "value" of adherence to the goals imposed on them. Because they are less proficient at using conscious mechanisms of learning, they usually require more practice to reach a certain threshold of performance than their peers with TD. As a consequence of experiencing our training protocol, participants with DS generally achieved the level of performance their peers with TD demonstrated in the pretest but did not exceed it. More practice may have enabled further improvements in this group, although boredom with this task may well limit greater effects.

Wishart (21) suggested that the combination of a structured environment, constant maximization of potential, and positive encouragement overcomes the avoidant learning styles adopted by children with DS. This approach influences the serf-efficacy of the child by maintaining motivation and adherence to task. We used conversation about the unique interests of each participant as well as pictures of cartoon characters and animals to engage them in the task. Both verbal and visual encouragement was used to motivate participants. Our goal was to motivate and encourage all participants to actively engage in the treadmill walking task. We also deemed it necessary to provide many repetitions of the task. Thus, over 4 practice sessions, participants stepped, on average, 890 strides per leg. Together, many cycles of perceiving and acting within a motivating and appropriately challenging context allowed them to realize maximum benefit from the practice sessions.

Conclusions

The results of this study suggested that preadolescents with DS are able to significantly improve their walking behavior on the treadmill following active task-specific practice. After initially increasing stiffness and impulse compared with over-ground walking, they modified their use of dynamic resources to achieve levels of stiffness and impulse similar to their peers with TD, while still demonstrating unique gait patterns. Our data in the present study as well as data from previous studies (1,22) show that people with DS demonstrate a wider step width compared with people with TD, even when normalized to leg length, across their lifespan. Although their systems are inherently different, as are the movement solutions that consequently emerge, preadolescents with DS are quite adaptive; they modify and optimize their behavior with adequate task-specific practice.

The clinical implications of this research are: (1) demonstrating the ability of preadolescents with DS to adapt more efficient behavior (decreased stiffness and impulse) with sufficient practice of a novel task, (2) addressing the question of how much and what type of practice is needed to effect behavioral changes, and (3) exploring the sources of inherent differences in available dynamic resources and how they affect the emergence of movement and behavior. Preadolescents with DS experience participation restrictions due to inefficient activity performance and behavior resulting from a combination of body structure characteristics and personal factors. In our case, they show inefficient walking behavior in a novel treadmill walking context as a result of ligamentous laxity, hypotonia, decreased force production, decreased balance, perceived instability, cognitive limitations, and decreased self-efficacy. Following four 12-minute sessions of motivated, task-specific practice, however, participants improved walking behavior efficiency. Practice sessions consisted of massed practice and were evenly distributed across time.

The next question is: Does this adaptive capacity function as well early in life, as walking emerges, and late in life, after many years of walking "wear and tear" on the system? Our current projects are designed to address this question. (22) We have, in progress, studies that address the ability of adults with DS to adapt their movement patterns to different contexts, including the treadmill.

Appendix.

Formulas for Normalization

[FORMULA NOT REPRODUCIBLE IN ASCII]

where v (velocity), g=gravity, [l.sub.STRIDE] (stride length), [f.sub.STRIDE] (stride frequency), and [w.sub.STEP] (step width) are converted gait variables and [l.sub.o] is leg length (ie, the sum of thigh length and shank length).

Dr Smith, Dr Kubo, Dr Holt, and Dr Ulrich provided concept/idea/research design. Dr Smith and Dr Ulrich provided writing and project management. Dr Smith, Dr Kubo, Dr Black, and Dr Ulrich provided data collection. Dr Smith and Dr Kubo provided data analysis. Dr Holt and Dr Ulrich provided fund procurement. Dr Ulrich provided facilities/ equipment.

The authors thank the preadolescents and their families who participated in this study as well as the Down syndrome support groups of Lansing, Toledo (Down Syndrome Association of Greater Toledo [DSAGT]), Detroit (Families Exploring Down Syndrome [FEDS]), Brighton, and Ann Arbor.

All procedures were approved by the University of Michigan Institutional Review Board.

Poster presentations of this research were given at Dynamic Walking 2006; May 6-8, 2006; Ann Arbor, Mich; and at the Annual Meeting of the North American Society for the Psychology of Sport and Physical Activity; June 1-3, 2006; Denver, Colo. An oral communication of this research was presented at the 8th Biannual Meeting of the North American Federation for Adapted Physical Activity; October 12-14, 2006; Ann Arbor, Mich.

This research was supported by National Institutes of Health grant HD42728 awarded to Dr Ulrich. Dr Smith was supported by grant H424C010067 from the US Office of Special Education and Rehabilitative Services awarded to Dr Ulrich and R Angulo Barroso.

This article was received September 25, 2006, and was accepted February 28, 2007.

DOI: 10.2522/ptj.20060289

References

(1) Ulrich BD, Haehl V, Buzzi UH, et al. Modeling dynamic resource utilization in populations with unique constraints: preadolescents with and without Down syndrome. Hum Mov Sci. 2004;23: 133-156.

(2) Almeida GL, Corcos DM, Latash ML. Practice and transfer effects during fast single-joint elbow movements in individuals with Down syndrome. Phys Ther. 1994;74: 1000-1012.

(3) Latash M, Kang N, Patterson D. Finger coordination in persons with Down syndrome: atypical patterns of coordination and the effects of practice. Exp Brain Res. 2002;146:345-355.

(4) Wang WY, Ju YH. Promoting balance and jumping skills in children with Down syndrome. Percept Mot Skills. 2002;94: 443-448.

(5) Lafferty ME. A stair-walking intervention strategy for children with Down's syndrome. Journal of Bodywork and Movement Therapies. 2005;9(1):65-74.

(6) Jensen RK. Body segment mass, radius and radius of gyration proportions of children. J Biomech. 1986;19:359-368.

(7) Connolly BH, Michael BT. Performance of retarded children, with and without Down syndrome, on the Bruininks-Oseretsky Test of Motor Proficiency [erratum appears in Phys Ther 1986;66:1781]. Phys Ther. 1986;66:344-348.

(8) Jeng SF, Holt KG, Fetters L, Certo C. Self-optimization of walking in nondisabled children and children with spastic hemiplegic cerebral palsy. J Mot Behav. 1996; 28:15-27.

(9) Alton F, Baldey L, Caplan S, Morrissey LC. A kinematic comparison of overground and treadmill walking. Clin Biomech. 1998;13:434-440.

(10) Hreljac A, Marshall RN. Algorithms to determine event timing during normal walking using kinematic data. J Biomech. 2000; 33:783-786.

(11) Holt KG, Fonseca ST, LaFiandra ME. The dynamics of gait in children with spastic hemipleglc cerebral palsy: theoretical and clinical implications. Hum Mov Sci. 2000; 19:375-405.

(12) Kuo AD. Stabilization of lateral motion in passive dynamic walking. International Journal of Robotics Research. 1999;18: 917-930.

(13) Sutherland DH, Olshen R, Cooper L, Woo SL. The development of mature gait. J Bone Joint Surg Am. 1980;62:336-353.

(14) Dusing SC, Thorpe DE. A normative sample of temporal and spatial gait parameters in children using the GAITRite electronic walkway. Gait & Posture. 2007;25: 135-139.

(15) Aylward EH, Habbak R, Warren AC, et al. Cerebellar volume in adults with Down syndrome. Arch Neurol. 1997;54:209-212.

(16) Pinter JD, Eliez S, Schmitt JE, et al. Neuroanatomy of Down's syndrome: a high-resolution MRI study. Am J Psychiatry. 2001;158:1659-1665.

(17) Raz N, Torres IJ, Briggs SD, et al. Selective neuroanatomic abnormalities in Down's syndrome and their cognitive correlates: evidence from MRI morphometry. Neurology. 1995;45:356-366.

(18) Costa AC, Walsh K, Davisson MT. Motor dysfunction in a mouse model for Down syndrome. Physiol Behav. 1999;68: 211-220.

(19) Hyde LA, Crnic LS, Pollock A, Bickford PC. Motor learning in Ts65Dn mice: a model for Down syndrome. Dev Psychobiol. 2001;38:33-45.

(20) Latash ML. Motor control in Down syndrome: the role of adaptation and practice. Journal of Developmental and Physical Disabilities. 1992;4:227-261.

(21) Wishart JG. The development of learning difficulties in children with Down's syndrome. J Intellect Disabil Res. 1993;37: 389-403.

(22) Ulrich BD. Dynamic systems and pediatric disorder: Down syndrome. Programs and Abstracts: Proceedings of the III STEP Conference--Linking Movement Science and Intervention. Alexandria, Va: American Physical Therapy Association; 2005.

* CIR Systems Inc, 60 Garlor Dr, Havertown, PA 19083.

([dagger]) Healthometer, Precision Weighing Balances, 10 Peabody St, Bradford, MA 01835.

([double dagger]) Siber Hegner and Co, Wiesenstr 8, PO Box 888, Zurich, Switzerland 8034.

([section]) American Guidance Service, 4201 Woodland Rd, Circle Pines, MN 55014.

([parallel]) Bertec Corp, 6171 Huntley Rd, Columbus, OH 43229.

([#]) LET Medical Systems Corp, 5755 NW 151 St, Miami Lakes, FL 33014.

** Peak Performance Technologies, 7388 S Revere Pkwy # 901, Centennial, CO 80112

([dagger])([dagger]) The MathWorks Inc, 3 Apple Hill Dr, Natick, MA 01760.

([double dagger])([double dagger]) SPSS Inc, 233 S Wacker D., Chicago, IL 60606.

BA Smith, PT, DPT, is a kinesiology PhD student in the Division of Kinesiology, University of Michigan, Ann Arbor, Mich. Address all correspondence to Dr Smith at: [e-mail].

M Kubo, PT, ScD, is a postdoctoral research associate, Department of Physical Therapy, University of Delaware, Newark, Del.

DP Black, PhD, is a postdoctoral research associate, Division of Kinesiology, University of Michigan.

KG Holt, PT, PhD, is Associate Professor, Department of Physical Therapy and Athletic Training, Sargent College of Health and Rehabilitation Sciences, Boston University, Boston, Mass.

BD Ulrich, PhD, is Professor and Dean, Division of Kinesiology, University of Michigan.

[Smith BA, Kubo M, Black DP, et al. Effect of practice on a novel task--walking on a treadmill: preadolescents with and without Down syndrome. Phys Ther. 2007;87:766-777.]

Table 1.
Anthropometric Data (Mean and SD [in Parentheses]) of Preadolescents
With Down Syndrome (DS) and With Typical Development (TD)

Variable                            DS               TD

Lengths (cm)
  Head and neck                     23.54 (4.38)     25.59 (2.04)
  Trunk                             44.27 (3.73)     46.30 (3.43)
  Thigh (a)                         28.56 (1.80)     33.81 (3.36)
  Shank (a)                         26.75 (2.61)     32.89 (2.66)
  Foot (a)                          19.40 (1.68)     21.92 (1.60)
  Upper arm (a)                     20.66 (1.40)     25.06 (2.83)
  Forearm (a)                       17.97 (1.85)     21.34 (1.57)
Upper-extremity/trunk ratio (a)      0.88 (0.06)      1.00 (0.08)
Lower extremity/trunk ratio (a)      1.26 (0.13)      1.44 (0.06)
Foot/height ratio                    0.16 (0.01)      0.16 (0.01)
Weight (kg)                         29.14 (6.79)     32.07 (3.94)
Height (m) (a)                       1.22 (0.06)      1.39 (0.08)
Body mass index (kg/m2) (a)         19.41 (2.61)     16.70 (1.37)

(a) P < .05.

Table 2.
Descriptive Characteristics of Over-ground (OVG) and Treadmill
Gait of Preadolescents With Down Syndrome (DS) and With Typical
Development (TD) (a)

                                                   Dimensionless
                                                      Stride
                         Stride Length (m)            Length

Speed    Day         DS             TD             DS       TD

40%      Pretest     0.43 (0.06)    0.56 (0.10)    0.76     0.83
         Posttest    0.46 (0.06)    0.71 (0.13)    0.82     1.07

75%      Pretest     0.68 (0.09)    0.89 (0.09)    1.22     1.31
         Posttest    0.71 (0.05)    0.91 (0.13)    1.28     1.37

110%     Pretest     0.83 (0.09)    1.11 (0.11)    1.54     1.68
         Posttest    0.90 (0.07)    1.14 (0.11)    1.61     1.72

OVG      Pretest     0.91 (0.15)    1.12 (0.14)    1.63     1.67
         Posttest    0.84 (0.13)    1.11 (0.12)    1.62     1.71

                                                   Dimensionless
                           Step Width (m)           Step Width

Speed    Day         DS             TD             DS       TD

40%      Pretest     0.15 (0.02)    0.08 (0.03)    0.27     0.13
         Posttest    0.15 (.02)     0.09 (0.05)    0.26     0.13

75%      Pretest     0.14 (0.02)    0.08 (0.03)    0.25     0.12
         Posttest    0.13 (0.02)    0.09 (0.03)    0.22     0.13

110%     Pretest     0.13 (0.02)    0.07 (0.02)    0.23     0.11
         Posttest    0.11 (0.02)    0.09 (0.03)    0.18     0.13

OVG      Pretest     0.12 (0.04)    0.08 (0.03)    0.22     0.11
         Posttest    0.11 (0.06)    0.06 (0.04)    0.18     0.09

                                                   Dimensionless
                               Stride                 Stride
                           Frequency (Hz)            Frequency

Speed    Day         DS             TD             DS       TD

40%      Pretest     1.02 (0.21)    0.81 (0.10)    0.24     0.21
         Posttest    0.89 (0.09)    0.62 (0.05)    0.21     0.16

75%      Pretest     1.17 (0.15)    0.93 (0.07)    0.28     0.25
         Posttest    1.07 (0.07)    0.89 (0.06)    0.25     0.23

110%     Pretest     1.29 (0.12)    1.1 (0.06)     0.31     0.28
         Posttest    1.21 (0.09)    1.0 (0.08)     0.29     0.27

OVG      Pretest     1.09 (0.15)    0.97 (0.07)    0.26     0.25
         Posttest    1.04 (0.20)    0.98 (0.07)    0.25     0.26

                                                   Dimensionless
                                Gait                   Gait
                            Speed (m/s)                Speed

Speed    Day         DS             TD             DS       TD

40%      Pretest     0.40 (0.03)    0.43 (0.06)
         Posttest

75%      Pretest     0.75 (0.06)    0.82 (0.10)
         Posttest

110%     Pretest     1.09 (0.09)    1.2 (0.15)
         Posttest

OVG      Pretest     1.01 (0.10)    1.08 (0.14)    0.43     0.42
         Posttest    0.95 (0.15)    1.13 (0.18)    0.41     0.44

(a) Values are mean (SD). See Appendix for formulas used to
convert absolute values to dimensionless values.