One-Minute Walk and modified Timed Up and Go tests in children with cerebral palsy: performance and minimum clinically important differences

Authors


Abstract

Aim

This prospective multicenter study assessed performance and changes over time, with and without surgical intervention, in the modified Timed Up and Go (mTUG) and One-Minute Walk tests (1MWT) in children with bilateral cerebral palsy (CP). Minimum clinically important differences (MCIDs) were established for these tools.

Method

Two hundred and nineteen participants with bilateral spastic CP (Gross Motor Function Classification System [GMFCS] levels I–III) were evaluated at baseline and 12 months follow-up. The non-surgical group (n=168; 54 females, 114 males; mean age 12y 11mo, [SD 2y 7mo], range 8y 1mo–19y) had no surgical interventions during the study. The surgical group (n=51; 19 females, 32 males; mean age 12y 10mo [SD 2y 8mo] range 8y 2mo–17y 5mo) underwent soft-tissue and/or bony procedures within 12 months from baseline. The mTUG and 1MWT were collected and MCIDs were established from the change scores of the non-surgical group.

Results

Dependent walkers (GMFCS level III) required more time to complete the mTUG (p≤0.01) than independent walkers (GMFCS levels I and II). For the 1MWT, distance walked decreased with increasing impairment (p≤0.01). 1MWT and mTUG change scores were not significantly different at any GMFCS level for either the surgical or non-surgical groups (p≤0.01).

Interpretation

Children with varying levels of function (GMFCS level) perform differently on the 1MWT and mTUG. The data and MCID values can assist clinicians in interpreting changes over time and in assessing interventions.

Abbreviations
1MWT

One-Minute Walk Test

MCID

Minimum clinically important difference

mTUG

modified Timed Up and Go

Outcome tools are used to assess performance at a single time, as well as over time or over a course of treatment. The usefulness of these tools is improved with the ability to compare scores to typical performance by peers and to determine if changes observed are clinically meaningful. Previous investigators have reported minimal clinically important differences (MCIDs) to help clinicians interpret changes seen on outcome tools.[1-3] For ambulatory children with cerebral palsy (CP), MCID values have been established for several outcome tools including the Gross Motor Function Measure, Pediatric Outcomes Data Collection Instrument, Pediatric Quality of Life, Pediatric Functional Independence Measure, oxygen cost, Pediatric Evaluation of Disability Inventory, gait profile score, pediatric motor activity log, and gait temporal–spatial parameters, to assist clinicians in evaluating interventions and interpreting results.[1-5] For these outcome measures to be useful and practical they should not be time consuming, costly, nor require specialized training to administer. In this regard, two quick, accessible, and affordable outcome tools that help clinicians assess balance and walking ability are the modified Timed Up and Go (mTUG) test and One-Minute Walk Test (1MWT; Table 1).[6, 7] The interpretation of the data obtained from these measures is limited because of (1) inadequate information is available about typical performance values, changes that occur with time or following surgical intervention, and (2) there is a lack of MCID values for comparison of children with spastic diplegic CP stratified by functional level using the Gross Motor Function Classification System (GMFCS).

Table 1. Review of One-Minute Walk Test (1MWT) and modified Timed Up and Go (mTUG)
Outcome measurePurpose of testPopulationClinical useLevel of measurementStandardization (manual)ReliabilityValidity
  1. aDistinguish among individuals at a single point in time.[27] bAssess change over time within a participant.[27] +Instrument meets criteria.

1MWT[6, 26]Discriminativea Evaluativeb

Disorder: children with cerebral palsy

Age: 4–16y

Cognitive ability: must be able to follow instructions

Clarity of instructions: easily administered performance test with single tester

Cost: readily available; inexpensive

Time to administer: 5min or less

Format: measure distance in meters to walk 1min as fast as possible

Training needed: none

Ratio; distance in meters

Test is administered by a standardized script.

No manual, refer to referenced manuscript

++
mTUG[13]Discriminativea Evaluativeb

Disorder: children with cerebral palsy or myelomeningocele

Age: 3–19y

Cognitive ability: must be able to follow instructions

Clarity of instructions: easily administered performance test with single tester

Cost: readily available; inexpensive

Time to administer: 15min or less

Format: measure time in seconds to rise from sitting from a standard chair, walk 3m, turn, walk back to the chair, and sit down. Can use a walking aid.

Training needed: none

Ratio; time in seconds

Test is administered by a standardized script.

No manual, refer to referenced manuscript

++

The TUG is a fast, reliable, and practical screening tool that measures basic functional mobility and balance in adults and children. It was adapted by Podsiadlo and Richardson[8] from the Get Up and Go Test.[9-12] Its original purpose was to measure basic mobility skills of frail elderly persons. Williams et al.[13] later modified the TUG (mTUG) to evaluate children with and without disabilities with the introduction of a task. The group investigated the reliability and validity of the mTUG in 176 children without physical disabilities and 41 children with CP or myelomeningocele, classified in GMFCS levels I to III.[13] The mTUG was able to detect significant differences among the three GMFCS levels. The mTUG times showed moderate negative correlation with the standing (dimension D) and walking (dimension E) of the Gross Motor Function Measure.

The 1MWT is a validated and user-friendly tool to evaluate walking ability and endurance in individuals with CP.[6] McDowell et al. reported typical walking distances for the 1MWT and found the instrument to be significantly correlated to the Gross Motor Function Measure.[6] However, the study used a small heterogeneous sample of children with CP.

The usefulness of both the mTUG and 1MWT for individuals with CP would be improved by establishing typical performance scores, defining changes with time or surgical intervention, and establishing MCIDs. These data should be specific to functional ability, or GMFCS level, and the distribution of limb involvement (diplegic or hemiplegic), using a large population sample.

One difficulty in establishing MCIDs in this population of patients is identifying natural progression because many ambulatory children with diplegic CP undergo orthopedic surgeries. Orthopedic surgery is used as standard care in individuals with CP to improve function, decrease discomfort, and prevent disabling structural changes.[14] Gorton et al.,[15] in a 6-year prospective study involving seven institutions, identified that 16% of 562 children underwent orthopedic surgery over a 1-year period. Thus, one must consider an effect of orthopedic surgery when evaluating the natural progression of children with CP.

The purpose of this study was to evaluate the mTUG and 1MWT at baseline and 1 year later in ambulatory children with bilateral spastic CP with and without surgical intervention. Differences among GMFCS levels, between non-surgical and surgical groups, and over time were assessed. MCID values were established.

Method

Participants

This study was part of a 4-year prospective multicenter study with cross-sectional and longitudinal components. Participants were a convenience sample of eligible patients from seven pediatric orthopedic hospitals. Institutional Review Board approval was obtained at each site, and consent and Health Insurance Portability and Accountability Act forms were completed.

Inclusion criteria were participants with a diagnosis of bilateral spastic CP, in GMFCS levels I to III, aged 8 to 18 years, who were able to complete an instrumented gait analysis with or without assistive devices. Participants were excluded if they had lower extremity orthopedic surgery within 1 year from the baseline evaluation, botulinum toxin A injections in the previous 4 months, or a currently implanted and operating baclofen pump. They were also excluded if they were unable to follow instructions to perform the tests and/or were visually impaired.

Two hundred and eighty individuals with bilateral spastic CP were assessed at baseline. Two hundred and nineteen (78%) of the participants completed both baseline and follow-up assessments, but not every participant completed both the 1MWT and TUG (Tables 2 and 3).

Table 2. Demographic data for the non-surgical and surgical groups at baseline assessment
 Non-surgical GroupSurgical Group
OverallGMFCS IGMFCS IIGMFCS IIIOverallGMFCS IGMFCS IIGMFCS III
  1. GMFCS, Gross Motor Function Classification System.

Sex (n)
Female5416191919496
Male114305331326197
Age at study (y)
Mean (SD)12y 9mo (2y 6mo)13y 4mo (2y 5mo)12y 9mo (2y 8mo)12y 7mo (2y 5mo)12y 8mo (2y 7mo)11y 1mo (2y 4mo)12y 9mo (2y 7mo)13y 8mo (2y 3mo)
Range8y 1mo–19y8y 5mo–19y8y 4mo–18y 7mo8y 1mo–18y 4mo8y 2mo–17y 4mo8y 6mo–14y 8mo8y 2mo–17y 4mo8y 3mo–16y 8mo
Standing height (cm)
Mean (SD)147.7 (14.7)151.9 (13.6)148.6 (15.2)142.4 (13.7)144.5 (13.2)140.5 (14.0)145.8 (14.1)144.8 (11.0)
Range109.2–176.0125.0–174.0118.0–176.0109.2–169.0121.0–174.6121.6–162.5121.0–174.6121.9–163.1
Weight (kg)
Mean (SD)45.7 (16.9)47.6 (18.5)45.9 (16.9)43.6 (15.5)40.7 (15.3)35.2 (14.4)43.3 (16.2)39.5 (13.4)
Range18.0–120.921.4–120.920.5–97.618.0–87.020.5–78.620.9–72.320.5–78.622.5–73.4
Table 3. Demographic data for the non-surgical and surgical groups at 12 months assessment
 Non-Surgical GroupSurgical Group
OverallGMFCS IGMFCS IIGMFCS IIIOverallGMFCS IGMFCS IIGMFCS III
  1. GMFCS, Gross Motor Function Classification System.

Sex (n)
Female5416191919496
Male114305331326197
Age at study (y)
Mean (SD)14y 1mo (2y 7mo)14y 5mo (2y 5mo)14y (2y 8mo)13y 8mo (2y 5mo)14y (2y 7mo)12y 4mo (2y 5mo)14y 1mo (2y 7mo)15y 2mo (2y 4mo)
Range9y–19y 9mo9y 6mo–19y 9mo9y 3mo–19y 8mo9y–19y 6mo9y 3mo–18y 7mo9y 7mo–16y 3mo9y 5mo–18y 7mo9y 3mo–18y 3mo
Standing height (cm)
Mean (SD)151.8 (13.5)156.4 (12.4)152.6 (13.9)146.4 (12.4)148.8 (13.2)145.6 (15.5)150.0 (13.1)148.6 (12.1)
Range110.5–179.0129.0–175.3129.5–179.0110.5–167.0124.0–175.2124.0–173.0125.0–175.2127.0–169.0
Weight (kg)
Mean (SD)49.6 (17.2)52.5 (19.6)49.8 (17.0)46.6 (15.0)44.1 (15.4)41.2 (16.9)46.8 (15.6)40.6 (13.6)
Range20.9–126.421.6–126.423.6–98.220.9–88.422.5–83.923.2–83.922.5–78.223.0–73.2
Time between baseline and follow-up assessments (mo)
Mean (SD)13.2 (1.9)13.0 (1.8)13.3 (2.2)13.2 (1.7)14.9 (3.1)15.2 (4.1)14.2 (2.5)16.2 (3.2)
Range9.0–20.59.0–18.210.2–20.59.9–19.79.6–23.99.6–23.910.8–20.611.0–21.3

The non-surgical group consisted of 168 participants who did not undergo a surgical procedure between their two assessments (baseline and 12mo follow-up). The non-surgical group was assessed at baseline and 12 months after their initial evaluation. Those with implant removal were included in the non-surgical group and were assessed 12 months from surgery. The surgical group consisted of 51 participants who underwent surgical intervention within 12 months from baseline. Follow-up assessment was performed 12 months after surgery. The surgical procedures ranged from single-level soft-tissue release to multilevel bony and/or soft-tissue procedures, totaling 171 procedures (Fig. 1 and Table 4). At baseline, demographic characteristics of individuals who did not complete the follow-up assessments were not different from those who did.

Table 4. Number of bony and soft-tissue procedures performed in the surgical group
Soft-tissue proceduresNumber of procedures
Heelcord lengthening23
Hamstring lengthening45
Psoas lengthening4
Rectus femoris transfer9
Adductor lengthening10
Other7
Total soft-tissue procedures98
Bony procedures
Foot/lateral column lengthening10
Femoral osteotomy28
Tibia/fibula osteotomy14
Pelvic osteotomy1
Total bony procedures53
Other
Patella (reconstruction dislocating patella with extensor realignment)4
Other bony and soft tissue16
Total other20
Total procedures171
Figure 1.

Summary of surgical procedures performed.

Data management

Before study initiation, all coordinators attended a training session to review and standardize administration of the outcome tools and data collection processes. All participant data were entered directly into a custom database. De-identified data from each site were compiled for analysis. Data quality and consistency was ensured by the project manager and site coordinators.

Assessment tools

Each participant completed two assessments (baseline and 1y follow-up) that included a history of the patient, GMFCS level, height, weight, and mTUG, and 1MWT scores.

The mTUG is a tool used to assess a person's mobility and requires both static and dynamic balance.[8] This is a measure of activity in the World Health Organization's International Classification of Functioning, Disability and Health (ICF) model.[16] Standard procedures for the mTUG were followed.[13] The length of time (in seconds) to rise from sitting in a chair, walk 3m at a comfortable velocity, touch the wall, turn around, walk back to the chair, and sit down using customary walking aids, orthoses and shoes was recorded. Two trials were performed and the average score was used for data analysis.

Walking capacity was assessed using the 1MWT, a measure of body structure and function based on the ICF.[6] For the 1MWT, participants were instructed to walk using their typical walking aids, orthoses, and shoes at their fastest pace, without running, for 1 minute. Location of administration of the 1MWT was specific to each institution but ensured that the participant could walk as far as possible at their fastest speed for a full minute. The total distance (in meters) walked was measured using a measuring wheel and recorded.

Evaluators

Trained evaluators at each of the seven participating sites collected both the 1MWT and mTUG data. Consistency among the evaluators was assessed during annual training sessions and locally at each site when there were multiple assessors at a single site. However, to ensure reliability between assessments, all efforts were made to have the same evaluator assess the same participants at each test session.

Data analysis

Using data from the baseline assessment for all participants, descriptive statistics were generated for the mTUG and 1MWT by GMFCS level. Data were then divided into the surgical and non-surgical groups and change scores were calculated as the difference between the follow-up and baseline assessments. Unpaired t-tests were used to assess differences between the surgical and non-surgical groups. Analyses of variance (ANOVAs) were used to analyze differences among GMFCS levels. Post-hoc tests were performed to examine significant main effects. Owing to the large number of parallel tests, a value of p≤0.01 was set a priori as the significance level for all statistical tests. pc-sas version 9.3 (SAS Institute, Cary, NC, USA) was used for all analyses.

The mTUG and 1MWT MCID threshold values were calculated for each GMFCS level using the non-surgical group change scores and the method previously reported by Oeffinger et al.[1] The equation used was

display math

where the desired effect size corresponds to both medium (0.5) and large (0.8) effect sizes needed to obtain a clinically meaningful change, r is the correlation between the repeated measures, and SD is the standard deviation of the change score.

Results

One hundred and sixty-eight individuals who did not undergo a surgical procedure (54 females, 114 males; mean age 12y 11mo, [SD 2y 7mo], range 8y 1mo–19y) and 51 individuals who underwent surgical intervention (19 females, 32 males; mean age 12y 10mo [SD 2y 8mo] range 8y 2mo–17y 5mo) participated in the study. Demographic data for the study population at baseline and follow-up assessments for both are reported in Tables 2 and 3. We analyzed data to assess site bias and found no significant differences among the seven sites (ANOVA, p≤0.05) at any GMFCS level. Demographic data were not significantly different between the surgical and non-surgical groups at baseline and/or follow-up.

Sample mean, SD, 95% confidence interval, and interquartile range for mTUG and 1MWT for the entire group are reported by GMFCS level in Table 5. The interquartile ranges illustrate the range of scores obtained by the middle 50% of the participants for each GMFCS level and illustrate variability within GMFCS level and the degree of overlap between GMFCS levels. Comparisons among GMFCS levels showed the time required to complete the mTUG at baseline differed between GMFCS levels I and III (p≤0.01) and levels II and III (p≤0.01). Time increased with increasing severity. Significant differences in the distance walked were found among all GMFCS levels for the 1MWT (p≤0.01), with decreasing distance seen with increasing impairment.

Table 5. Descriptive data for the modified Timed Up and Go (mTUG) and One-Minute Walk Test (1MWT) for the overall group
GMFCS level n mTUG baseline, s (SD)Lower 95% CI, sUpper 95% CI, sInterquartile (mid-50%) range, s
I, mean (SD)567.2 (1.5)b6.87.64.011.2
II, mean (SD)1009.0 (2.6)c8.59.54.323.1
III, mean (SD)6125.1 (15.8)21.129.27.382.5
GMFCS level n 1MWT baseline, m (SD)Lower 95% CI, mUpper 95% CI, mInterquartile (mid-50%) range, m
  1. ap<0.01 between levels I and II. bp<0.01 between levels I and III. cp<0.01 between levels II and III. GMFCS, Gross Motor Function Classification System.

I, mean (SD)5684.9 (13.5)a,b81.288.555.7113.9
II, mean (SD)9969.9 (17.7)c66.373.432.0115.7
III, mean (SD)6347.9 (17.0)43.652.220.289.8

For the non-surgical and surgical groups, 1MWT and mTUG means and SDs at baseline, and follow-up, and change scores are reported by GMFCS level in Table 6. Comparisons among GMFCS levels are consistent with the findings from the entire group for both the mTUG and 1MWT. Change scores were not significantly (p≤0.01) different among GMFCS levels for either the 1MWT or mTUG in either the surgical or non-surgical groups.

Table 6. Change scores for the non-surgical and surgical groups
GMFCS level n mTUG baseline, s (SD)mTUG at 12mo, s (SD)mTUG change scores n 1MWT baseline, m (SD)1MWT at 12mo, m (SD)1MWT change scores
  1. ap<0.01 between levels I and II. bp<0.01 between levels I and III. cp<0.01 between levels II and III. dp<0.01 between level II non-surgical and level II surgical groups. GMFCS, Gross Motor Function Classification System; mTUG, modified Timed Up and Go; 1MWT, One-Minute Walk Test (1MWT).

Non-surgical group
I, mean (SD)467.4 (1.5)b7.8 (1.9)b0.4 (1.9)4685.6 (13.8)a,b86.7 (14.2)a,b1.1 (12.6)
II, mean (SD)728.9 (2.3)c9.1 (2.5)c0.2 (1.9)7172.9 (17.1)c77.3 (16.1)c4.4 (13.6)
III, mean (SD)4823.2 (14.3)23.5 (14.3)−0.1 (5.9)5047.6 (16.5)51.6 (16.4)4.3 (11.3)
Surgical group
I, mean (SD)106.7 (1.5)b6.6 (1.3)b−0.1 (1.6)1081.5 (11.9)a,b92.2 (8.0)a,b10.7 (15.5)
II, mean (SD)289.3 (3.3)c10.1 (2.8)c0.6 (3.0)2862.3 (17.1)c,d68.0 (14.9)c,d5.6 (11.9)
III, mean (SD)1332.2 (19.4)35.3 (25.7)3.0 (16.3)1349.2 (19.6)44.3 (21.25)−4.9 (16.3)

At baseline and follow-up, the surgical group in GMFCS level II walked significantly shorter distances in the 1MWT than the non-surgical group in GMFCS level II (p≤0.01; Table 6). No other differences between the non-surgical and surgical groups were seen for the mTUG, 1MWT, or change scores in any GMFCS level (p≤0.01).

MCID threshold values were calculated and change scores needed to achieve MCIDs with medium and large effect sizes by GMFCS level are reported in Table 7. Because change scores for the study groups were not statistically significant, MCID values were not applied.

Table 7. Medium and large minimum clinically important difference (MCID) values for the modified Timed Up and Go (mTUG) and One-Minute Walk Test (1MWT) by GMFCS level
 GMFCS I (n=47)GMFCS II (n=72)GMFCS III (n=46)
MCID medium (0.5)MCID large (0.8)MCID medium (0.5)MCID large (0.8)MCID medium (0.5)MCID large (0.8)
  1. GMFCS, Gross Motor Function Classification System.

mTUG1.11.70.71.21.21.9
1MWT5.69.05.18.33.86.2

Discussion

The current study provided the 1MWT and mTUG scores for children with spastic diplegic CP stratified by impairment (GMFCS) with and without surgical intervention at baseline, 1-year follow-up, change scores, and MCID values, which have not been previously reported. Consistent with previous reports[13, 17] this study found that the time required to complete the TUG increased with increasing impairment (GMFCS level). The mTUG was able to discriminate between independent ambulators (levels I and II) and dependent ambulators (level III). However, it did not discriminate between level I (independent ambulators who walk without limitations) and level II (independent ambulators who walk with limitations). These findings are in agreement with Gan et al.[17] but differ slightly from Williams et al.[13] who found the mTUG was able to discriminate among all three GMFCS levels. The participants in the Williams et al. study included a more heterogeneous population of individuals with diagnoses of diplegic CP, hemiplegic CP, and myelomenigocele.[13] The present study focused on bilateral spastic CP, making it a more homogenous study group. A separate analysis by distribution of limb involvement is supported by Damiano et al., who reported that even within the same GMFCS level, children with hemiplegia outperformed children with diplegia in measurements for lower extremity function.[18]

The 1MWT was able to discriminate among all three GMFCS levels; distance walked decreased as GMFCS level, or impairment, increased. Our current study findings differ from those reported by McDowell et al.,[6] who found greater mean distances walked at each GMFCS level and no difference between GMFCS levels I and II. A smaller study sample and inclusion of individuals with both diplegic and hemiplegic involvement in that study might have contributed to these differences.

How much change is clinically meaningful can be challenging to quantify for a number of reasons, and several different approaches have been reported.[19-22] Some have suggested that MCIDs be subjectively defined based on patient or family rating, or by clinicians from training and personal experience.[19] Others have proposed objective techniques to define MCIDs.[20, 21] The MCIDs defined in this study used an objective technique based on the assumption that the mean change score needed to obtain a medium or large effect size was observable[23] and, therefore, clinically meaningful.

MCIDs can provide an objective measure to assist clinicians with determining if the changes are meaningful and larger than the variability expected for the group. Change scores can be statistically significant but not clinically important. The use of MCIDs can assist researchers in determining if significant changes seen over time have clinical implications. The use of MCIDs has been applied in the absence of statistical significance.[4] This application is debatable and should be applied cautiously.

In the current study, change scores over a 1-year period were not significant for either the non-surgical or surgical groups, nor were the differences between the surgical and non-surgical groups significant. The natural history of children with CP is characterized by a gradual deterioration of gait function over time.[24, 25] Therefore, minimal changes were expected over a year for the non-surgical group. This was supported by small change scores among GMFCS levels in both the mTUG and 1MWT. The lack of significance in the surgical group could be due to relatively small sample sizes or the variability in response to treatment. The surgical procedures included in this study ranged from isolated musculo-tendenous releases to single-event multilevel procedures. This variability of procedures may explain, in part, the large variations in outcomes resulting in a small mean difference over the study period.

Limitations

The 1-year follow-up of the study participants is a relative short-term follow-up and probably limited the magnitude of changes seen in both the surgical and non-surgical groups. Despite the initial large study sample when the group was stratified into surgical and non-surgical groups by GMFCS levels, the sample sizes in some of the subgroups became relatively small.

The MCIDs presented in this study are calculated using assumptions related to effect sizes. Ideally, these MCIDs would include validation using an external standard. Unfortunately, no external standard exists in the clinical settings for most outcome tools, including the 1MWT and mTUG. The 10% change for the 1MWT and 10% change for the mTUG required to meet the MCID values for a medium effect size are comparable to values clinicians expect for improved clinical performance on these measures.

Owing to the multicenter study design, there were multiple assessors for each of the studied tools. To reduce any potential errors associated with the number of evaluators, each evaluator was trained in the standardized protocol. Intrarater consistency was tested at the training sessions by having all evaluators collect data on the same participant concurrently. Data were collected until all evaluators' data were within 1 second. When multiple assessors collected data at the same site, consistency among evaluators was checked in the same manner. Site bias assessments performed during the study and at its conclusion did not demonstrate significant site bias among the seven sites. Evaluators were not blinded to the treatments received during the study period. This does not appear to have biased the results as change scores were not significant for either the surgical or non-surgical treatment groups.

Conclusion

This prospective, longitudinal, multi-center study of ambulatory children with diplegic CP has reported typical performance ranges, changes over a 1-year period with and without surgical intervention, and established MCID values for the mTUG and 1MWT. These results should improve the use of these measures within the clinical setting for this population. Children with varying levels of impairment function differently on the 1MWT and mTUG. The mTUG differentiated between independent (GMFCS levels I and II) and dependent (GMFCS level III) walkers; the 1MWT demonstrated differences among all GMFCS levels, with the distance walked decreasing with increasing impairment (GMFCS level). The data reported, stratified by GMFCS level, will provide background knowledge to aid clinicians in assessing patients and determining realistic goals for them. The MCID values improve the interpretation of change scores and can be used to assess the effectiveness of interventions for the study population.

Acknowledgements

The Functional Assessment Research Group acknowledges the funding of this work by Shriners Hospitals for Children, grant number 9158, the research coordinators and motion analysis laboratory staff at all participating facilities for their roles in data collection, and the participants and their families.

Ancillary