A Systematic Review of the Effectiveness of Treadmill Training and Body Weight Support in Pediatric Rehabilitation

Abstract

Background and Purpose: 

Given the extensive literature on body weight–supported treadmill training (BWSTT) in adult rehabilitation, a systematic review was undertaken to explore the strength, quality, and conclusiveness of evidence supporting use of treadmill training and body weight support in those with pediatric motor disabilities. A secondary goal was to ascertain whether protocol guidelines for BWSTT are available to guide pediatric physical therapy practice.

Methods: 

The database search included MEDLINE, EMBASE, CINAHL Plus, PEDro, Cochrane Library databases, and ERIC from January 1, 1980 to May 31, 2008 for articles that included treadmill training and body weight support for individuals under 21 years of age, with or at risk for a motor disability. We identified 277 unique articles from which 29 met all inclusion criteria.

Results: 

Efficacy of treadmill training in accelerating walking development in Down syndrome has been well demonstrated. Evidence supporting efficacy or effectiveness of BWSTT in pediatric practice for improving gait impairments and level of activity and participation in those with cerebral palsy, spinal cord injury, and other central nervous system disorders remains insufficient, although many studies noted positive effects.

Discussion and Conclusion: 

The original evidence demonstrates efficacy of BWSTT in children with Down syndrome, but large-scale controlled trials are needed to support the use of BWSTT in other pediatric subgroups. Increased use of randomized designs, studies with treadmill training–only groups, and dosage studies are needed before practice guidelines can be formulated. Neural changes in response to training warrant exploration, especially given the capacity for change in developing nervous systems.

INTRODUCTION

Research demonstrating that coordinated stepping movements could be elicited in spinalized cats leads ultimately to the incredible discovery that this was also possible in humans with complete spinal cord injury (SCI).¹ To facilitate step training, weight-support systems for treadmills were developed that drastically reduced the postural requirements and hence the amount of physical assistance needed to safely participate in ambulation training. As a result, rehabilitation practices were transformed for those with SCI² and for those post-stroke,³ with multiple studies focusing on therapist- or machine-assisted step training in these populations.

However, the evidence supporting this rapidly expanding clinical approach may not be as strong as some presume. A recent Cochrane review of stroke presented the evidence from 15 randomized trials of treadmill training and body weight support in the treatment of walking limitations after stroke in which no statistically significant differences between treadmill training, regardless of body weight support, and overground training for improving walking speed or dependence were found.⁴ The data from individual trials indicated that task-specific practice, not the treadmill per se, was the active element in producing functional improvements in gait. Lam et al⁵ performed a systematic review of the efficacy of gait rehabilitation strategies for those with SCI and echoed this conclusion. Lower level evidence (nonrandomized studies) did show some support for body weight–supported treadmill training (BWSTT) in chronic SCI by showing that these programs did improve aspects of functioning. However, their review revealed that there was strong support for comparable outcomes from BWSTT and overground walking practice in subacute SCI when intensity was equivalent. The Cochrane review of the stroke population also recommended further investigation because some individual studies suggested that treadmill training with body weight support was superior to treadmill training alone. The addition of body weight support makes repetitive training far more feasible for a broader range of clients and allows more flexibility in terms of optimizing speed and training kinematic patterns for those with weakness or other impairments limiting their gait function by increasing safety and decreasing the physical work necessary by one or more therapists.

Widespread clinical and research interest in locomotor training in adult neurological rehabilitation has now infiltrated pediatric physical therapy practice for children with delayed or reduced gait function. Pediatric research on the effectiveness or efficacy of locomotor-based training devices and protocols has lagged behind their clinical incorporation, with practitioners relying primarily on evidence and guidelines from the adult literature because that is what was available to them. A review of this topic was published in 2006.⁶ Of the 15 studies uncovered by the literature search, seven were abstracts and one appeared in a physical therapy news magazine, leaving only seven studies of BWSTT in children with motor disabilities that were published as research reports in peer-reviewed scientific journals. The authors concluded that with the limited and relatively low level of evidence, the current research does not support the effectiveness of treadmill training. Because BWSTT provides opportunities for physical training of lower extremity strength and endurance and repetitive task (and/or speed)-specific training of stepping, both of which are consistent with principles in the exercise physiology and motor control literature, respectively, and because research and clinical interest in this approach has increased substantially in recent years, we thought that it was important to revisit this topic. Therefore, the goal of this article was to substantially enhance the previous search to include the rapidly accumulating scientific literature on this topic and to investigate the evidence supporting treadmill training with and without body weight support across broader diagnostic categories in pediatric neurorehabilitation. A secondary goal was to ascertain whether sufficient protocol guidelines are available to guide practice in certain pediatric subgroups. We also expanded the scope compared with the previous pediatric review to include infants, children, and young adults (younger than 21 years of age) with medical diagnoses in which a motor disability was a consistent and/or prominent feature.

METHODS

Inclusion and Exclusion Criteria

As stated above, the search aimed to include all studies that investigated outcomes of treadmill training and body weight–supported gait training, used separately or in combination. We chose not to eliminate studies in which other treatments were administered or permitted at the same time. The review was restricted to those studies with the primary goals of improving lower extremity motor functioning including increasing step counts, rate, or coordination on the treadmill; increasing overground gait speed, symmetry, and coordination; decreasing the need for assistance when walking; and more generally promoting lower extremity gross motor skill development or task performance. Studies that aimed solely to improve aerobic fitness or to decrease body weight through greater caloric expenditure were excluded.

The population of interest was infants, children, adolescents, and young adults younger than 21 years of age who had or were at risk of developing a motor disability affecting gait coordination or function. We did not include studies of typically developing children, those that used treadmills for sports-related training, or those that addressed diminished exercise capacity because of asthma, cystic fibrosis, obesity, or acute medical illness. In studies that also included individuals 21 years or older, we only included those studies that provided individual data or separate analyses for those younger than 21 years of age.

The review was limited to studies published in peer-reviewed journals with full text available in English. All research reports were accepted, regardless of study design. We excluded studies published only in abstract or dissertation form and those investigating only the within-session effects of different walking conditions. Review articles on related topics were also excluded.

Search Strategy

We conducted a comprehensive systematic literature search to identify all relevant articles. Both authors first received training in electronic search methods and strategies from a medical librarian. The following databases were searched covering the time span from January 1, 1980 to May 31, 2008: MEDLINE, EMBASE, CINAHL Plus (cumulative index to nursing and allied health literature), PEDro (Physiotherapy Evidence Database), all databases within the Cochrane Library, and ERIC (Education Resources Information Center). We chose 1980 as the lower cutoff because BWSTT in humans did not emerge until the early to mid 1980s, following related discoveries made through animal research.

The EMBASE.com web-based search platform was used to simultaneously search the EMBASE and MEDLINE bibliographic databases. For each search term, selected options included mapping to preferred terminology (with spell check) and searching also for synonyms, with explosion on preferred terminology. Records were limited to Humans, In English, and Records added between January 1, 1980 and May 31, 2008. We searched PEDro for records that included either “treadmill” or “weight support” in the abstract/title field, within the subdiscipline “pediatrics.” All other searches were limited to English language publications between January 1980 and May 2008, and the strategy listed below was used:

1. Treadmill
2. Locomotor
3. “Over ground”
4. Overground
5. “Weight support”
6. Harness NOT Pavlik
7. Robotic NOT surgery
8. #1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7
9. Infant OR child OR adolescent
10. Training
11. #8 AND #9 AND #10
12. #11 NOT obese

The EMBASE/MEDLINE search returned 163 unique citations. As we proceeded sequentially through the list of databases and eliminated any previously identified citations, 32 additional unique citations were identified by CINAHL, 69 by the Cochrane Library, eight by PEDro, and four by ERIC. One additional relevant citation was found through examination of reference lists. These 277 unique citations were then examined further.

Each author independently screened each title and abstract to determine whether criteria for inclusion were potentially met. Selection results from the two authors were compared, and discrepancies were resolved by discussion. None required a third reviewer to reach a decision. With this process, we eliminated 218 articles that did not report outcomes of treadmill and/or partial body weight–supported gait training for individuals younger than 21 years of age with motor disabilities. Thirteen abstracts, two dissertations, three review articles, and two reports published in non–peer-reviewed journals were also excluded.

We then retrieved the remaining 39 full text articles that potentially met the search criteria, and each author reviewed each article separately. Three studies were excluded because no subjects younger than 21 years of age participated. Six included both pediatric and adult subjects, but were excluded from this review because data from the pediatric subjects were not reported separately. One study was eliminated because it compared different treadmill walking conditions within a single session. The remaining 29 articles met the inclusion criteria and were included in the review. Both authors independently read each article, rated each using the PEDro scale^7,8 (http://www.pedro.fhs.usyd.edu.au/scale_itemhtml#scale_1), and extracted data using the form provided on the website of the American Academy of Cerebral Palsy and Developmental Medicine as a guide (http://www.aacpdm.org). A consensus process was used to finalize the level and quality of evidence rating presented here.

Effect sizes were calculated for the three most common outcome measures in the group with central nervous system impairments: self-selected gait velocity, the Gross Motor Function Measure (GMFM) dimensions D and E, and only those studies that had the necessary data within the article to compute effect sizes could be included. This was not done for the SCI group because of the predominance of case reports rather than group data or for the Down syndrome group because the primary outcome there differed across studies. Effect sizes were calculated by (1) the difference in mean change scores across treatment groups divided by the standard deviation of the mean change score in the control group or (2) the mean change as a result of the intervention of interest divided by its standard deviation.

RESULTS

A general description of all 29 included studies^9–37 is provided in Table 1, including the basic study design, number and characteristics of the participants, outcome measures used, and summarized results. The information in Table 2 focuses more specifically on the intervention details including the parameters of the treadmill training, such as speed and duration, type and amount of body weight support provided, if any, and a description of other concurrent interventions, that may have been provided or that the child was permitted to continue during the study period. These are separated by the three subpopulation groupings that emerged in this review: (1) those with cerebral palsy (CP) or other central motor impairments, (2) those with SCIs, and (3) infants with Down syndrome. None of the studies used body weight support without a treadmill, although not all treadmill studies included the use of body weight support for all participants.^10,19,24 In most cases, the two intervention strategies were used in combination. The level and quality of evidence of the identified studies are summarized in Tables 3 to 5. Sackett's Levels of Evidence³⁸ (Table 3) was used to determine the strength of the evidence and the PEDro rating instrument⁸ (Table 4) was used to rate the quality of the studies. The level of evidence and quality rating for each of the included studies are listed in Table 5.

TABLE 1: Description of Study Design, Participants, Outcomes Measures and Results by Study

TABLE 1: Description of Study Design, Participants, Outcomes Measures and Results by Study. (continued)

TABLE 1: Description of Study Design, Participants, Outcomes Measures and Results by Study. (continued)

TABLE 1: Description of Study Design, Participants, Outcomes Measures and Results by Study. (continued)

TABLE 2: Description of Locomotor Training Methods by Study

TABLE 2: Description of Locomotor Training Methods by Study. (continued)

TABLE 3: Sackett's Levels of Evidence³⁸

TABLE 4: Physiotherapy Evidence Database Scale⁸

TABLE 5: Level and Quality of Evidence Supporting Treadmill and/or Body Weight Supported Gait Training

Outcome summaries are provided in Tables 6 to 8 based on those recommended in the American Academy of Cerebral Palsy and Developmental Medicine methodology document contained on their website, and key outcomes across studies are listed under the subcategories of body structures and functions or activity and participation as delineated in the International Classification of Functioning (ICF) recently published by the World Health Organization.³⁹ These are divided into three separate tables based on the subject population grouping. Although the primary goal of two of the studies was to demonstrate whether the training protocol could be feasibly implemented in the target population, motor-related outcome data were reported for all participants, and were included here. Table 6 includes all 17 studies on children who have, or are at risk for, a motor disability as a result of a various central motor impairments. The majority of the 114 total participants (>76%) had CP, but there was also a small cohort of children who underwent hemispherectomies for intractable seizures,¹⁷ an infant who had sustained an intraventricular hemorrhage,¹¹ several young girls with Rett syndrome,¹⁹ a young woman with cerebellar ataxia,¹³ and a young man with a traumatic brain injury.²⁵Table 7 includes the results of all currently published outcomes data collected by a single laboratory, reporting on results from two separate comprehensive home-based treadmill training protocols involving 60 infants with Down syndrome, 45 of whom participated in a treadmill intervention. Both protocols were randomized, controlled trials, the first of which examined the effects of treadmill training in a group of infants with Down syndrome compared with a similar control group not receiving the training; and the second of which compared the previous training paradigm with a more intense and progressive one in terms of speed, time, and additional limb loading while on the treadmill. The six different published reports each address a different set of primary or secondary outcomes that were collected before, during, and immediately after training, and/or up to 15 months after the intervention was discontinued (as determined by the onset of independent walking by each infant participant). Table 8 includes all data on seven total participants with SCI who were younger than 21 years of age, each identified as a participant from among six published case series or reports in this population, some of which also contained data on adults that was not included here.^32,34–36

TABLE 6: Cerebral Palsy and Other Central Motor Impairments

TABLE 7: Down Syndrome

TABLE 8: Spinal Cord Injury

Because the primary goal of this review was to evaluate the evidence supporting or failing to support the use of treadmill training and body weight support in pediatric therapy, this topic is addressed first for each of the three subject groupings.

Cerebral Palsy and Other Central Nervous System Disorders

Although this grouping contains the largest number of studies by far, 17 in all, no randomized clinical trial has been reported among these to evaluate the efficacy of BWSTT. The one level 2 study included here compares two types of treadmill training paradigms, with both groups showing significant increases in GMFM scores on the standing and walking, running, and jumping dimensions over time.¹⁴ Because the study did not include a no treatment condition, however, the ability to draw conclusions is limited. The strongest research available to address intervention effectiveness is a single level 3 study by Dodd et al,¹⁸ which is a nonrandomized, controlled trial comparing two matched cohorts, one receiving the intervention and one serving as a control group. That study did show a significant effect for increased gait speed in the training group during a 10-meter walk at the subjects' self-selected comfortable speed. The distance walked in 10 minutes was substantially higher (by nearly 20 meters, on average) in the treatment group, but this result did not reach significance, most likely because of the small group size (n = 7) and the variability across subjects in the amount of change. All of the 15 other studies were Evidence Level 4 (10 studies) and 5 (5 studies) studies, with a limited number of statistically favorable effects. Some positive effects emerged from across the multiple studies, but each outcome measure that showed positive results in one or more studies also had inconclusive or equivocal results in one or more other studies. For example, changes in self-selected gait velocity and in the GMFM dimensions D and E were the most frequently noted positive results, and they were also among the most frequently reported results that either did not show a difference or were inconclusive. Effect sizes for these three outcomes are shown in Figure 1. We used an effect size (d) of 0.20 as the lower cutoff for a small effect size, 0.50 for a medium effect size, and 0.80 for a large effect size.⁴⁰ As noted, all mean effects were positive, with the majority falling in the medium to large range.

FIGURE 1.: Effect sizes for the three most common outcomes reported in group studies within the central nervous system Impairment subgroup.

PEDro scores ranged from 2 to 6 across studies, with a median value of 2. Lower quality ratings were largely a function of the lower level study designs. A major weakness in this group of studies was the presence of co-interventions, which may have had large distorting effects. In some cases, the results may have been more closely related to the other interventions than to the intervention of interest. For example, the study by Blundell et al¹⁰ was primarily intended to increase strength, with treadmill training as one of many methods employed. Consequently, many of the outcome measures were assessments of functional strength and showed positive results that corroborated the isometric strength results. It is likely that the other strength training interventions had a larger effect on that outcome than the treadmill training component did. Other studies included botulinum toxin or recent surgery immediately before the intervention,^12,20 both of which could have potentially large positive or negative effects on outcomes depending on timing with respect to the treadmill training, muscles or joints addressed, and the aggressiveness or invasiveness of treatment. An interesting observation is that all columns in the summary table appear to show similar distributions across the ICF categories of body structures and functions and activity and participation. This may be related to the nature of the intervention, which involves training to improve a specific functional task, rather than to alleviate an impairment.

In summary, the strongest evidence, a single Level 3 study,¹⁸ suggests that BWSTT is effective in increasing self-selected gait speed. Although other positive, statistically supported outcomes have been identified, any positive effects found, while generally large, may not all be of clinical significance. The weakly positive or inconclusive outcomes from these pediatric studies are similar to those reported in other adult neurological conditions. This intervention has also not been compared sufficiently to other intervention approaches so that its relative benefits, as well as costs, cannot yet be adequately assessed. Larger studies including control and treatment comparison groups are necessary to determine efficacy foremost, and, if found, whether the effort and expense associated with BWSTT, in terms of equipment as well as therapist and patient time, are justifiable.

Down syndrome

The study published by Ulrich et al in 2001²⁸ has had a major impact on the field of infant development as well as on pediatric physical therapy because it was the first study to demonstrate that locomotor development, as measured by milestone achievement, in children with a known motor disability could be accelerated by as much as several months by practicing stepping on a treadmill while being supported by a parent for 8 minutes per day. The strength of the evidence from that initial study is rated as Level 2 and is strong because it was an adequately powered, randomized, controlled trial. Most importantly, it is the only study included in this review that demonstrates the efficacy of treadmill training in children with a motor disability compared with a control intervention. The PEDro score for that study was a 6/10, which equals the highest score assigned in this review, with points lost only because no one was blinded and allocation may not have been concealed. The second randomized trial that was done by this same group evaluated whether development could be further accelerated by increasing intensity in terms of longer treatment time, speed, and resistance, showing significant group differences in achievement of two milestones. Difference in treatment effects across groups were more modest than when a treatment group was compared with a control group because both groups received treadmill training.²⁹ Interestingly, compliance with the more intense but complex protocol was not as good as with the simpler, less intense protocol. Multiple secondary outcomes were measured in this second cohort and demonstrated that the more intense group showed several other beneficial short or longer term effects (as long as 15 months after training ceased) including a more mature gait pattern, as revealed through principal component analysis of multiple gait parameters and through significant differences on several, but not all, measured temporospatial and kinematic gait parameters.²⁷ The group that received more intense training also showed more advanced obstacle avoidance strategies³¹ and more time spent at a moderate-high activity level at 15 months post-training, suggesting a possible longer term effect on levels of physical activity.⁹ The lower intensity group showed more time spent at a low activity level and a shorter duration of moderate-high activity, a result that appears to favor the more intense training protocol. Wu et al³⁰ compared the second cohort with the controls from the first cohort and demonstrated that only the higher intensity group showed greater stride length, earlier age at walking onset, and less time from start of intervention until walking onset compared with controls.

In summary, five of six studies reported in infants with Down syndrome were classified as Level 2, with the one study by Wu et al³⁶ classified as Level 4 because it used a control group from a previously reported study. PEDro scores ranged from 3 to 6. Primary outcomes from each trial tended to be at the level of activity and participation with respect to the ICF, whereas secondary outcomes were mainly at the level of body structures and functions. A weakness noted in the reports resulting from the second training cohort was the fact that not all who were enrolled in that study participated in or completed the training or complied with all the many types of assessments, and mean ages and standard deviations provided in the substudies reflected the cohort who participated in the original training protocol, rather than the subsample of those who had data on the secondary outcome being reported in subsequent reports.

Finally, these studies support the efficacy of treadmill training for promoting the development of independent walking and for advancing other quantitative and qualitative aspects of gait performance. Some evidence further suggests that a higher intensity of training may be more effective than a less intense protocol. Although group differences for several outcome measures, as shown in Tables 1 and 7 did not reach statistical significance, many demonstrated similar trends of the significant findings and may have reached significance given a larger sample size.

SCI in Children and Young Adults

In contrast to the mainly high level of evidence in infants with Down syndrome, the studies identified for those with SCI are either individual case reports or individual subject data from multiple case series and are therefore classified as either Level 4 or 5, which can at best merely hint at causality. Each earned a PEDro score equal to 2. All the studies included other types of intervention including stretching, overground training, or other nonspecific physical and/or occupational therapy rehabilitation exercises. Outcomes were almost equally distributed across the ICF categories of body structures and functions and activity and participation. Most outcome results were positive, with some showing large and clearly clinically significant changes such as progression from no ability to step to walking independently with an assistive device by the end of training.^33,36,37 Although many of those included showed large effects, one participant made virtually no functionally relevant changes beyond improved stepping on the treadmill.³² A particularly interesting finding was the lack of change in the lower extremity motor score in a child who became a functional ambulator with a walker, showing amazingly that he could walk around his kindergarten classroom all day, but could not perform isolated knee extension.³³ This illustrates the task specificity of step training in this population. Data on children and adolescents with SCI are very limited compared with data on adults with SCI, even though it is possible that children may have greater potential for improvement, as well as a longer projected life span, which makes aggressive rehabilitation efforts even more critically important for them. Clearly, larger, more rigorous studies are needed and, given the promising preliminary evidence, are strongly warranted in this population.

Protocols

A secondary goal of this review was to evaluate whether an effective protocol emerged for specific groups to inform clinical practice. Table 2 lists the different protocols that were used across studies and are summarized primarily within groups. In the studies with SCI that included adults, the protocol was similar for those younger than 21 years of age. All the SCI case studies or series used body weight support typically starting at a very high percentage, given the level of involvement of those who underwent this type of training. The frequency cited ranged from three to five days per week. The shortest program duration was 12 weeks, and the longest was five months. Thirty minutes of training was a consistent upper goal across studies, with two reports failing to include this parameter.^34,35 Treadmill belt speeds were high with the target as normal gait speed, which is thought to be important to adequately stimulate both reciprocal stepping and arm swing.³² The speed often needed to be adjusted downward to optimize the stepping pattern for each subject, which often coincided with a decrease in the amount of manual assistance needed. Some other nuances were common in this population, such as the use of specific sensory inputs provided manually by therapists, particularly in the early stages of training, and the decision not to allow the use of orthotic devices or handrails.

For the infants with Down syndrome, the protocol was well defined for both treadmill training protocols that were conducted, and from which six articles have been published to date. In the first protocol used, the designated belt speed was 20 cm/sec and the frequency was five sessions per week for approximately eight minutes per session.²⁸ The other novel part of this program was the fact that it was done in the home by a trained parent who supported the infant from the front during the treadmill stepping. The second training protocol that was used had one group training for at least six minutes for five days per week at a speed of 18 cm/sec, similar to the first protocol. The more intense group progressed over time to a slightly higher speed of 22 cm/sec and also attempted to increase the length of the sessions by several minutes.²⁹ The final modification in that group was the addition of progressively increased ankle weights, which was found on later evaluation to actually decrease stepping behavior for a period in some of the infants. The added weight increased the difficulty and caused performance to deteriorate, presumably until strength increased sufficiently.²⁷ Speed and duration for the infants were understandably much lower compared with these in studies with older participants. Given the impressive compliance in the first landmark study,²⁸ the initial protocol in particular clearly demonstrates strong feasibility as well as efficacy.

For those with central motor impairments, the protocols used varied tremendously. Age of participants, levels of involvement, diagnoses, and intended goals of treatment were also quite variable, so lack of consistency across protocols is not surprising. Speeds ranged from 13 to 380 cm/sec and the duration of sessions ranged from four to 43 minutes, with the majority in the range of 20 to 30 minutes. Nearly all sessions were conducted in therapy, although several were conducted in a school setting,^10,18,19 an infant case report used a home-based program similar to that used for children with Down syndrome,¹¹ and one case report transitioned a child to a home treadmill based on the desires and resources of the subject's family.¹³ Frequency ranged from two to five times per week, and program duration was as short as two weeks and as long as five months. The amount of weight support provided ranged from none to 60% and in many studies or cases was not reported. Some of these parameters were determined primarily on an individual basis, often trying to provide only as much weight support as needed to optimize the walking pattern, and increasing speed depending on each child's tolerance. Standardized strategies for progression were rarely provided, and modifications to treatment parameters over time varied across studies and across participants within a study.

The determination of when to cease the training varied across studies as well and was related to either the predetermined program duration for the study, transition to overground gait training when that became possible, or when the participant either met the original treatment goal or reached a plateau. For the infant studies, a natural stopping point was identified, which was when the child could take steps independently and therefore practice stepping on his or her own. For those in whom the ambulation potential is uncertain, the stopping point is far less clear. In some of the individual case reports, treatment was continued for several weeks before the person being trained even began to take consistent steps. Some studies stopped the training when a plateau was reached^13,36; however, in a commentary to the more recent case report by Behrman et al,³³ Edgerton⁴¹ recognizes that plateaus can be misleading and that progress typically occurs in increments rather than a smooth linear trajectory, so a lull in progress may be just that and not a firm endpoint in recovery. Although one case report of a child with CP made the statement that the outcomes of treadmill training improve with the length of therapy,¹² this logical presumption has not yet been empirically substantiated.

In summary, demonstration of efficacy of various protocols in producing clinically important changes in the level of participation for children with specific disabilities or groups of disabilities is the critical first step, as has been shown nicely in the studies of infants with Down syndrome. Systematic refinement of the various parameters for optimizing outcomes can then be the next focus, assuming that treatment superiority has been demonstrated.

Adverse Effects

Only a few studies addressed adverse events during the course of the training. Schindl et al²⁴ noted that two of the 10 participants in their study found the program exhausting. Meyer-Heim et al²⁷ noted that no child suffered a hip dislocation as a result, which is an important factor to consider in a population that is at risk of hip subluxation or dislocation. Dodd and Foley¹⁸ reported that there were no falls, injuries, or soreness in the training group. In a study using robotic assistance in a child with CP,¹² it was stated explicitly that no complaints or adverse effects were reported. Richards et al²³ further reported that there were no adverse events and no increase in scissoring behaviors during their feasibility study. In the first study of infants with Down syndrome, Ulrich et al²⁸ noted that none of the seven infants in their study who had surgically corrected congenital heart defects demonstrated any observable problems during treadmill training.²⁸ In a case report of a child with SCI, Prosser³⁷ reported that there were no episodes of autonomic dysreflexia throughout the study duration. In summary, reports on adverse events consistently found that none of the possible risks that were anticipated and monitored were found to have occurred, pointing to the safety of these programs for children.

DISCUSSION

It has been recognized in recent years that rehabilitation strategies have not been intense or aggressive enough as seen from the positive functional results of implementing strength training programs in persons with chronic motor disorders, such as SCI, stroke, CP, and traumatic brain injury, who had long since reached a functional plateau and/or had been discharged from therapy.⁴² Increasing the amount and intensity of physical activity is critically important for general health and for participation of those with motor disabilities.⁴³ Those who have the greatest limitations also face the greatest challenges in accomplishing this, as well as the greatest need. Therefore, body weight support systems or devices that expand the use of treadmills across many rehabilitation populations and increase the feasibility of gait training even for those who are nonambulatory have been enthusiastically endorsed by the physical therapy field. Although statistical and anecdotal positive results have been reported, the level of evidence to support BWSTT in pediatric practice is generally inconclusive, even for answering the most basic questions of effectiveness. Task-specific gait training has been shown to be effective in adults with stroke and SCI for improving gait speed; however, the superiority of BWSTT over other gait training methods has also not been well established in the adult rehabilitation literature^4,5 and must be established to justify its use. Even for infants with Down syndrome in whom treadmill training has been shown to be efficacious, this intervention has not yet been compared with other possible methods of stepping practice or other intense training regimens, which may offer similar benefits. Larger controlled trials to address these important unanswered questions are critically needed.

A common limitation in many studies, particularly in the group with CP and other central motor disorders, is that the goal of the training often was not stated explicitly or precisely. From the standpoint of motor control, the role of the motorized treadmill is to provide repetitive, task-specific practice of walking. However, treadmill training protocols can be adapted to accomplish several different goals from the level of impairment to participation, as well summarized in the introduction to the case report by Cernak et al.¹³ For example, progressively decreasing body weight support in a nonambulatory child can be an effective way to increase lower extremity strength, with the additional benefit of this being accomplished in a task-specific manner, which should translate more readily to functional gains. In cases in which the goal may be simply to increase gait speed, the protocol can be optimized to meet that goal through progressive increases in belt speed, with or without adjustments in weight support. Session duration can be progressively increased to increase endurance and fitness levels, although increases in speed and weight-bearing can also help to accomplish that goal. Improving symmetry of gait patterns is important for those with unilateral or markedly asymmetrical involvement and again may require different, specific parameter adjustments to optimally achieve that goal.

The primary goal for using BWSTT in people with SCI has been to improve stepping performance through accessing and training the spinal locomotor circuits,² but additional benefits in persons with incomplete injuries such as improvements in strength may also be secondarily achieved and contribute positively to the ultimate outcome. The extent to which the spinal pathways are secondarily affected in CP and whether spinal circuits can be similarly accessed to improve motor coordination in conditions such as CP are fascinating and as yet unanswered questions.

In contrast to physical therapy for adults in which the goal is to restore walking, the goal for infants and young children with developmental delays or disorders is often to promote the development of walking. Based on dynamical systems theory, Thelen and Fisher⁴⁴ proposed that one of the major reasons why infants who had been able to take supported steps stopped stepping for a brief period in their development was the biomechanical fact that their rapid increases in growth outpaced the development of their lower extremity extensor strength.^44,45 Motor experience that includes repetitive limb loading or trunk and lower extremity strengthening can and does provide the stimulus for developing sufficient strength to walk,^46,47,48 suggesting that development of walking skills could potentially be augmented through training both in normal infants and those at risk for developmental delays in achieving motor milestones. Demonstration of the efficacy of treadmill training to alter developmental trajectories, and the suggestion of longer term benefits on activity levels, has major implications for early intervention therapy practices for children with multiple disabilities, and further research in this area that tracks both short- and long-term effects is strongly warranted.

Finally, other issues must be more thoroughly investigated. The safety of using a treadmill over longer periods for those who are at risk for joint deformity in the short term such as fracture and hip dislocation, or at risk for osteoarthritis in young adulthood must be more systematically and carefully evaluated. The use of other lower extremity reciprocal exercise devices, such as cycles, water-based treadmills or elliptical trainers, may offer similar benefits with less repetitive joint stress and should also be explored. The cost of treadmill training programs are high and this may dramatically limit their availability or duration, unless adequately justified to, and accepted by, third party payors. Perhaps finding ways to transition locomotor training to a home or community based setting as early as possible in the rehabilitation process may decrease the expense, increase access and compliance, and promote lifelong attention to physical activity, rather than providing only a short-term intervention.

CONCLUSION

The state of the evidence for BWSTT in pediatric rehabilitation varies across populations. Efficacy of this training compared with controls has been demonstrated in infants with Down syndrome. Although some individual results can be compelling, evidence in pediatric SCI is very limited in the number of studies and the strength and quality of the evidence, so no general conclusion can yet be made regarding efficacy or effectiveness in this population. Despite the increased number of studies in CP and other central motor disorders, the strength of the evidence is generally weak with no randomized clinical trial performed to date to address the efficacy of this intervention. Optimal protocol development is still in its infancy for all three populations.