Abstract
Background A force-driven harmonic oscillator (FDHO) model reveals the elastic property of general muscular activity during walking.
Objective This study aimed to investigate whether children with Down syndrome (DS) have a lower K/G ratio, a primary variable derived from the FDHO model, compared with children with typical development during overground and treadmill walking and whether children with DS can adapt the K/G ratio to walking speeds, external ankle load, and a treadmill setting.
Design A cross-sectional study design was used that included 26 children with and without DS, aged 7 to 10 years, for overground walking and 20 of them for treadmill walking in a laboratory setting.
Methods During overground walking, participants walked at 2 speeds: normal and fastest speed. During treadmill walking, participants walked at 75% and 100% of their preferred overground speed. Two load conditions were manipulated for both overground and treadmill walking: no load and an ankle load that was equal to 2% of body mass on each side.
Results Children with DS showed a K/G ratio similar to that of their healthy peers and increased this ratio with walking speed regardless of ankle load during overground walking. Children with DS produced a lower K/G ratio at the fast speed of treadmill walking without ankle load, but ankle load helped them produce a K/G ratio similar to that of their healthy peers.
Limitations The FDHO model cannot specify what muscles are used or how muscles are coordinated for a given motor task.
Conclusions Children with DS show elastic property of general muscular activity similar to their healthy peers during overground walking. External ankle load helps children with DS increase general muscular activity and match their healthy peers while walking fast on a treadmill.
People with Down syndrome (DS) are known for their weak muscle strength, low muscle tone, and poor motor coordination.1–3 Children and adults with DS usually show a slower self-selected speed than their healthy peers while walking overground.4–8 Preadolescence is generally considered an optimal period for displaying the best motor ability in people with DS.9 However, even during preadolescence, children with DS display a less advanced gait pattern, such as a shorter stride length and a wider step width,9–12 a lower ankle plantar-flexion moment and power,7,11,13 and greater variability in the center-of-mass and joint angle excursions10,14 compared with typically developing children. Despite their motor deficits, children with DS are able to adapt to motor tasks and the environment. For instance, treadmill walking appears to be novel for preadolescents with DS. However, with limited practice, most children with DS aged 7 to 10 years are able to walk on a treadmill at different speeds, with and without external ankle load.9,12,15,16
Both time- and frequency-domain analyses demonstrate that a fast treadmill speed with the inclusion of ankle load helps children with DS produce a pattern of vertical ground reaction force similar to that of children with typical development.15,16 This walking paradigm appears to be promising in the development of an intervention program for children with DS to improve their motor adaptability. However, little research has been carried out to understand how children with DS modulate the mechanical property of the neuromuscular system during walking cycles.
A force-driven harmonic oscillator (FDHO) model has been used to understand the relationship between the elastic and gravitational properties of the neuromusculoskeletal system during the locomotion of quadrupeds and humans.17,18 The FDHO model can precisely predict cycle period of locomotion in quadrupeds and humans.17–20 Moreover, this model can estimate the elastic torque (active stiffness) of the neuromuscular system with the input of stride time, revealing the overall neuromuscular activity to overcome the passive stiffness of the leg due to gravity.20,21 This model was explored in the current study to help answer the aforementioned question on the mechanical property of the neuromuscular system during walking in children with DS.
The FDHO model is not the only mechanical model elucidating the kinetic and muscular characteristics of human walking. For instance, a simple inverted pendulum model can help estimate muscular torques at the joints by using inverse dynamics and the input of ground reaction force (GRF).22 However, this model does not quantify the general neuromuscular activity from the whole body. Another escapement-driven, damped inverted pendulum model was developed to estimate global muscular stiffness during the double-stance phase of walking.23–25 This model predicts that children with DS have global stiffness similar to their healthy peers during overground walking, but higher stiffness during treadmill walking.9 This model was not chosen in the current study because we aimed to estimate general neuromuscular stiffness during the whole walking cycle and to compare our results with quadrupeds and other human populations.
The FDHO model consists of a simple pendulum (the leg) and a spring (muscles and soft tissues) that is attached to the pendulum and provides a restoring force to facilitate the pendulum swing at its natural frequency. The ratio between the elastic (spring) and gravitational (pendulum) restoring torques (the K/G ratio as defined in the Method section) is the primary variable derived from the FDHO model and helps quantify the level of muscle contraction and co-contraction with respect to the gravitational torque. Although not identifying specific muscles, the K/G ratio presents the elastic property of general muscular activity of the neuromuscular system for a motor task. A ratio of 1 means a scaling of 1:1 between the elastic restoring torque from muscles and soft tissues and the gravitational torque of the pendulum. This ratio was found to be equal to 1 at the preferred locomotion speed in quadrupedal animals18 as well as in children with typical development aged 9 years and young adults.17,19 However, a ratio between 1.4 and 2.8 was observed in infants with typical development within 6 months after walking onset.20 Also, a ratio of 1.43 was reported in children with cerebral palsy during overground and treadmill walking.21 These observations indicate that the ratio between the elastic and gravitational restoring torques in the FDHO model can deviate from 1 in these populations due to their neuromuscular immaturity and deficits.
The purpose of this study was to apply the FDHO model to estimate the elastic property of general muscular activity during both overground and treadmill walking in children with and without DS. Given that children with DS are known to have low muscle tone, our first question was: Would children with DS show a lower ratio between the elastic and gravitational torques than their healthy peers? Furthermore, this ratio is speed-dependent, as it increases to 6 during trotting and 9 during cantering in quadrupedal locomotion.18 Our second question was: Would children with DS change this ratio at different walking speeds? As external ankle load has been shown to help children with DS walk with a kinetic pattern similar to that of children with typical development,15,16 our third question was: Would children with DS take advantage of ankle load and produce a ratio similar to that of their healthy peers? In addition, walking overground is an activity that children with DS practice every day; however, walking on a treadmill can be a challenging task because most children with DS usually do not have such experience. A previous study demonstrated that children with DS show similar global stiffness at their preferred comfort level between overground and treadmill walking but higher global stiffness at the same speed during treadmill walking.9 Therefore, our last question was: Would the K/G ratio from children with DS show the same pattern as global stiffness between overground and treadmill walking? We hypothesized that children with DS will: (1) produce a lower K/G ratio between the elastic and gravitational torques than children with typical development, (2) increase the K/G ratio at a faster speed like their healthy peers, (3) increase the K/G ratio with external ankle load like their healthy peers, and (4) produce a similar K/G ratio at the preferred comfort level between overground and treadmill walking (the speeds may be different) but a higher ratio at the same speed during treadmill walking.
Method
Participants
Twenty-six children, aged 7 to 10 years, with and without DS, completed the data collection of overground walking at the first laboratory visit; 20 of them completed the data collection of treadmill walking at the second laboratory visit (Fig. 1). The inclusion criterion was that participants were able to walk 10 m overground without assistance. The exclusion criterion was that participants had a history of medical conditions or known neuromuscular problems that prevented them from independent locomotion. The DS group was recruited through the Down Syndrome Association of Atlanta and the local parent support groups. The age- and sex-matched typically developing (TD) group was recruited from the local community through advertisements and personal contact. The DS group was shorter in height and had a higher body mass index than the TD group, but the 2 groups had similar body masses at each laboratory visit (Tab. 1). Written informed consent was obtained from all the participants and their parents or guardians before data collection.
Flow diagram of participant progress in this study. DS=Down syndrome, TD=typical development.
Mean (SD) of the Physical Characteristics of the Participants at Laboratory Visits 1 (Overground Walking) and 2 (Treadmill Walking)a
Data Collection
At the first laboratory visit, participants were tested for overground walking. Two overground walking speeds were manipulated: normal speed and fastest speed. Participants were instructed to walk at a comfortable speed in the normal speed condition and to walk as fast as possible in the fastest speed condition across a 10-m walkway. Also, 2 external ankle load conditions were manipulated on each side: no load (NL condition) and ankle load that was equal to 2% of the participant's body mass (AL condition). The AL condition presented a 39% increase of the moment of inertia of each leg about the hip joint.15,22 Another ankle load condition that was equal to 4% of body mass was proposed, but the majority of the DS group had difficulty in adapting to this load condition. Therefore, only the load condition equal to 2% of body mass (AL condition) was included in this study. There was no significant difference in ankle load between the DS and TD groups (Tab. 1).
A total of 4 conditions were tested combining the factors of walking speed and ankle load and were presented in a randomized order to the participants. Data for at least 4 trials were collected for each condition. Participants wore compression shorts or bathing suits during data collection. Reflective markers were placed bilaterally on the temporomandibular joint, acromion, lateral epicondyle of the humerus, greater trochanter, lateral epicondyle of the femur, lateral malleolus, and second metatarsal head. A 7-camera Vicon motion capture system (Vicon Denver, Centennial, Colorado) was used to record these markers during each trial. Kinematic data were collected at a sampling rate of 100 Hz. Adequate rest was provided between the trials and between the conditions to minimize fatigue, particularly in the DS group.
At the second laboratory visit, participants were tested for treadmill walking. The experimental setup was previously described.15,16 Each participant walked at a comfortable speed in the hallway outside of the laboratory 3 times, and the average speed was calculated as the preferred overground walking speed. As 75% of the preferred overground speed was reported to represent the preferred comfort level of treadmill walking in children with DS,9 we used 2 treadmill speeds—75% and 100% of the preferred overground speed—as normal speed (NS) and fastest speed (FS) conditions, respectively, in this study. The same 2 ankle load conditions were presented on each side as at the first laboratory visit: no load (NL condition) and ankle load that was equal to 2% of the participant's body mass (AL condition). There was no significant difference in ankle load between the DS and TD groups (Tab. 1). A total of 4 conditions were tested combining the factors of walking speed and ankle load and were presented in a randomized order to the participants.
A Zebris FDMT-S instrumented treadmill (Zebris Medical GmbH, Isny, Germany) was used to measure the timing and magnitude of vertical GRF during treadmill walking. Data for two 60-second trials were collected at a sampling rate of 100 Hz under each condition. A few participants in the DS group had difficulty starting the FS condition first, so we presented the NS condition first to allow acclimation in these participants.9 Two participants in the DS group did not complete the AL condition at the fastest speed, and one participant in this group did not complete the FS condition. A few participants in the DS group occasionally placed their hands on the handrails for a few steps. We verbally encouraged these participants to remove their hands from the handrails and complete treadmill walking without holding the handrails.
Data Analysis
For the kinematic data collected at the first laboratory visit, we manually identified the reflective markers using Vicon Nexus software (Vicon Denver) and determined the gait events of heel contact and toe-off from the heel and toe markers, respectively. The processed data were exported as text files and analyzed with custom-written Matlab programs (The Mathworks Inc, Natick, Massachusetts) to calculate spatiotemporal parameters, including speed, cadence, and stride length and time.
For the kinetic data collected at the second laboratory visit, the GRF data were processed first with Zebris FDM-TS software and exported as text files. Custom-written Matlab programs were used to determine gait events of heel contact and toe-off15 and calculate spatiotemporal parameters, including cadence and stride length and time. All of the spatiotemporal parameters are provided in Table 2, but the focus of this study was on the FDHO model. During overground walking, the DS group walked at a slower speed than the TD group in all the conditions except for the fastest speed without ankle load. During treadmill walking, the DS group walked at a slower speed than the TD group in both speed conditions regardless of ankle load (Tab. 2).
Mean (SD) of Spatiotemporal Gait Parameters During Overground and Treadmill Walkinga
FDHO Model
A complete description of an FDHO model is presented elsewhere.17,18 Briefly, the FDHO model represents a hybrid pendulum-spring system that consists of a single pendulum and a spring attached to the pendulum. The simple pendulum represents the gravitational contribution of the thigh-shank-foot system facilitating the passive dynamics of the system, and the spring represents the contribution of muscles and soft tissues facilitating the active dynamics of the system during locomotion. The spring component conserves mechanical energy by being passively stretched and produces mechanical energy by actively eliciting muscle contraction, ensuring a periodic restoring force to maintain the oscillations at the natural frequency.17,22 At resonance and for a small amplitude of oscillation during walking, the periodic duration (ie, stride time) can be predicted from this hybrid pendulum-spring model with the following equation17,18:
where τ is stride time, m is the mass, L is the equivalent leg length of the thigh-shank-foot system, g is gravitational acceleration of 9.81 m/s2, mL2 is the moment of inertia of the leg about the hip joint, mLg is the gravitational restoring torque associated with the size and mass of the system, and kb2 is the elastic restoring torque from muscles and soft tissues. This equation can be further reduced as shown below17,18:
where K represents kb2 and G represents mLg. The K/G ratio is the scaling ratio between the elastic and gravitational restoring torques (stiffness) in this hybrid pendulum-spring model. A case of kb2=mLg (ie, K/G ratio=1) represents a 1:1 scaling between the elastic and gravitational restoring torques, as reported in healthy children and adults while walking at the preferred speed.17,19 The above equation can be rearranged to calculate the K/G ratio as shown below17,18:
The equivalent leg length (L) of the thigh-shank-foot system can be calculated as shown below17,18:
where Isystem is the moment of inertia of the system about the hip joint, Msystem is the total mass of the system, and hsystem is the distance from the hip joint to the center of mass of the system. The mass and the location of the center of mass of the thigh, shank, and foot were estimated based on the regression equations previously reported in children aged 4 to 15 years.26 When an external ankle load was added, the thigh-shank-foot system became the thigh-shank-foot-ankle load system. The added ankle load increased the Msystem, Isystem, and hsystem of the thigh-shank-foot-ankle load system. The equivalent leg length L increased accordingly due to a greater increase in Isystem than in both Msystem and hsystem.
Data Analysis
Two 3-way analyses of variance (ANOVAs) (2 group × 2 speed × 2 load) with repeated measures on the last 2 factors were conducted on the K/G ratio for overground and treadmill walking separately to test the first 3 hypotheses. Two 3-way ANOVAs (2 group × 2 task × 2 load) with repeated measures on the last 2 factors were conducted on the K/G ratio to test the fourth hypothesis comparing overground and treadmill walking: (1) at a similar preferred comfort level of walking (overground versus treadmill walking in the NS condition) and (2) at similar walking speeds (overground in the NS condition versus treadmill in the FS condition). Post hoc pair-wise comparisons with Bonferroni adjustments were conducted when appropriate to examine the difference between the DS and TD groups at each condition and between the 2 load conditions at each speed in each group. We used SAS 9.2 statistical software (SAS Institute Inc, Cary, North Carolina) for all statistical analyses. The significance level was set at P<.05, unless otherwise specified.
Role of the Funding Source
This study was supported, in part, by a research grant from the Jerome Lejeune Foundation and a research initiation grant from Georgia State University. The study sponsors had no role in the study design, data collection and analysis, interpretation of data, writing of the manuscript, or the decision to submit the manuscript for publication.
Results
K/G Ratio During Overground Walking
The K/G ratio across the 2 groups was 1.07 in the NS condition and 3.02 in the FS condition. That is, the ratio between the elastic and gravitational torques was about 1:1 in the NS condition and 3:1 in the FS condition for the 2 groups regardless of external ankle load (Fig. 2A). Statistical analysis indicated there was no group effect, load effect, or group × load interaction on the K/G ratio between the 2 groups. However, there was a speed effect on the K/G ratio (F1,24=236.06, P<.001) such that both groups increased the K/G ratio from the NS condition to the FS condition.
Mean and standard deviation of the K/G ratio for: (A) overground walking and (B) treadmill walking. DS=Down syndrome, TD=typical development, NS=normal speed, FS=fast speed, NL=no load, AL=ankle load. * Significant difference between the DS and TD groups at P<.05. § Significant difference between the NL and AL conditions at a given speed condition within a group at P<.05.
K/G Ratio During Treadmill Walking
The K/G ratio across the 3 groups was 0.85 in the NS condition regardless of external ankle load. In the FS condition, the DS group produced a lower K/G ratio of 1.14 compared with 1.47 in the TD group without ankle load but had a K/G ratio of 1.44, similar to the K/G ratio of 1.43 in the TD group, with ankle load (Fig. 2B). Statistical analysis indicated there was a speed effect on the K/G ratio (F1,17=140.03, P<.001) such that both groups increased this ratio from the NS condition to the FS condition. Also, there was a group × load interaction on the K/G ratio (F1,18=9.11, P=.007). No difference in the K/G ratio was observed in the NS condition between the 2 groups. However, in the FS condition, the DS group increased the K/G ratio from the NL condition to the AL condition, and the TD group maintained the K/G ratio regardless of external ankle load.
Comparison Between Overground and Treadmill Walking
At a similar preferred comfort level of walking (ie, the NS condition), the DS group maintained a similar K/G ratio between overground and treadmill walking (Fig. 3A). Both the DS and TD groups increased the K/G ratio from the NL condition to the AL condition across overground and treadmill walking. Statistical analysis indicated that there was a group × task interaction (F1,18=4.51, P=.048) and a load effect (F1,23=5.38, P=.030) on the K/G ratio. Post hoc analysis revealed no difference in the K/G ratio between overground and treadmill walking in the DS group but a higher ratio during overground walking in the TD group.
Comparisons of mean and standard deviation of the K/G ratio between overground walking (OG) and treadmill walking (TM): (A) OG at normal speed (NS) vs TM at NS and (B) OG at NS vs TM at fast speed (FS). NL=no load, AL=ankle load. * Difference between the OG and TM conditions within a group for a given load condition at P<.05.
At a similar walking speed (ie, NS for overground walking and FS for treadmill walking), the DS group produced a higher K/G ratio during treadmill walking compared with overground walking regardless of ankle load (Fig. 3B). Statistical analysis indicated that there was a group × task × load interaction (F1,15=6.44, P=.023) on the K/G ratio. Post hoc analysis revealed that the DS group increased the K/G ratio from overground to treadmill walking and from the NL condition to the AL condition, and the TD group had a similar K/G ratio between overground and treadmill walking in the AL condition.
Discussion
Although the DS group walked at slower speeds than the TD group during both overground and treadmill walking conditions, the DS group had a K/G ratio similar to that of the TD group while walking overground at normal and fastest speeds regardless of ankle load. This trend also was observed for the normal speed during treadmill walking regardless of ankle load. However, the DS group produced a lower K/G ratio than the TD group at the fast speed of treadmill walking without ankle load, but the inclusion of ankle load helped the DS group produce a K/G ratio similar to that of the TD group. In addition, the DS group increased the K/G ratio with walking speed and ankle load. Between overground and treadmill walking, the DS group had a similar K/G ratio while walking at a similar preferred comfort level but produced a higher K/G ratio at a similar speed during treadmill walking.
Contrary to our first hypothesis, both the DS and TD groups showed a scaling of 1:1 between the elastic and gravitational restoring torques while walking overground at the preferred speed and a scaling of 3:1 at the fastest speed. Our results suggest that children with DS aged 7 to 10 years can match their healthy peers by producing similar elastic restoring torque from muscles and soft tissues to overcome the gravitational load of the thigh-shank-foot system during overground walking at both the preferred and fastest speeds. The scaling of 1:1 is consistent with the K/G ratio of 1 reported in quadrupedal animals, 9-year-old healthy children, and young adults.17–19 Our finding is also in agreement with the global stiffness predicted with an escapement-driven, damped inverted pendulum model such that children with and without DS display similar global stiffness at the stance phase of overground walking at the preferred speed.9
The increase in the K/G ratio at the fastest speed confirms our second hypothesis that children with DS are able to increase the elastic property of general muscular activity necessary to facilitate the dynamics of the leg swing when accommodating a faster walking speed. Even though children with DS are known to have low muscle tone and weak muscle strength,1 our results suggest that with many years of experience in walking overground, children with DS are able to match the general muscular function of children with typical development for this important daily task. Furthermore, a similar K/G ratio of 0.85 was found between the DS and TD groups while walking on a treadmill at the normal speed. As treadmill walking is a novel task for most children with DS, our results suggest that children with DS are able to translate their ability of controlling the dynamics of leg swing from well-practiced overground walking to novel and challenging treadmill walking at a similar preferred comfort level.
Walking on a treadmill at a faster speed presented a challenge to children with DS, as they showed a lower K/G ratio than their healthy peers. Compared with overground walking, novel and challenging treadmill walking at a fast speed may expose the neuromuscular limitations, such as poor balance control and motor coordination, in children with DS.14 However, the inclusion of external ankle load helped children with DS increase the K/G ratio to the level of children with typical development while walking faster on a treadmill. This finding partially confirms our third hypothesis, although no effect of ankle load was found on the K/G ratio during overground walking in the DS group. It also suggests that the amount of external ankle load used in this study may not be adequate to change neuromuscular activity for well-practiced overground walking.
However, when walking on a treadmill at a fast speed, external ankle load may have helped elicit an enhanced general muscular activity and increase elastic restoring torque from muscles and soft tissues in the DS group. External ankle load has been shown to be an important training component in treadmill intervention in infants with DS.27,28 Infants with DS who received a treadmill intervention with ankle load attained walking onset earlier and, after walking onset, walked faster overground, displayed better joint coordination, and stepped over an obstacle at a higher probability.28–31 Our previous study also showed that external ankle load improved the kinetic pattern of vertical GRF in children with DS in both time and frequency domains.15,16 Therefore, we propose that external ankle load may be a promising training component when designing a locomotor intervention for children with DS. Such an intervention protocol may help improve muscle strength and motor adaptation in children with DS for executing novel and challenging motor tasks and meanwhile may enhance bone mineral accrual in the critical preadolescent and adolescent periods.32
Consistent with our fourth hypothesis, the DS group produced a similar K/G ratio at the preferred comfort level between overground and treadmill walking but had a higher K/G ratio at a similar speed during treadmill walking. This finding suggests that although treadmill speed was slower than overground speed at a similar comfort level in the DS group, a similar level of elastic property of the neuromuscular system may be needed to overcome the challenge of treadmill walking. This need may be due to poor balance control and joint coordination in children with DS.14 From a motor control perspective, the related question becomes whether joint coordination is a key factor influencing the K/G ratio in different walking tasks? No study has used the FDHO model to examine the relationship between elastic and gravitational torques in children with coordination problems such as developmental coordination disorder (DCD). However, children with DCD display reduced ankle plantar flexion and an increased trunk inclination in the anteroposterior direction,33 as well an increased center-of-mass velocity and range of motion in the mediolateral direction, during walking.34 Also, our previous results showed that the DS group produced a less vertical propulsive force than the TD group.15
Based on these findings, we propose that further examination of the kinematic and neuromuscular patterns of walking in children with DS and DCD may provide evidence on the effect of motor coordination on the K/G ratio. In addition, our results confirm that walking on a treadmill at 75% of the preferred overground speed presents the preferred comfort level of treadmill walking in children with DS.9 When comparing similar walking speeds between overground and treadmill walking, the DS group increased the K/G ratio to elicit more neuromuscular activity for more challenging treadmill walking. This finding suggests that the K/G ratio depends not only on walking speed but also on the difficulty of a locomotion task.
One limitation of this study is that we used the kinematic data from overground walking and the kinetic data from treadmill walking to investigate the FDHO model of walking between children with and without DS. A previous study showed that both kinematic and kinetic patterns of walking on an AMTI instrumented treadmill (Advanced Mechanical Technology Inc, Watertown, Massachusetts) are qualitatively and quantitatively similar to those of overground gait.35 Furthermore, reliability of spatiotemporal and kinetic variables obtained from an FDM treadmill (Zebris Medical GmbH) was found to be good between testing days.36 We deem that it was appropriate to compare the results of the K/G ratio and spatiotemporal gait parameters between overground and treadmill walking in this study.
Although our results of the K/G ratio shed light on the relationship between the elastic and gravitational properties of the neuromuscular system and the dynamics of leg movement, we acknowledge that the FDHO model cannot specify what individual muscles are activated and how these muscles are coordinated to produce appropriate elastic restoring torque. The K/G ratio calculated from this model predicts the general contribution of muscle contraction and co-contraction with respect to the gravitational torque of the thigh-shank-foot system. The same K/G ratio may not mean that the same group of muscles is used; rather, it can be achieved by different combinations or levels of muscular activity. Therefore, future studies using electromyography or other interventions shall provide the detailed knowledge of specific muscular activity and coordination and help improve the interpretation of the FDHO model in explaining the dynamics of walking in children with DS.
In summary, despite physical constraints, children with DS show elastic property of general muscular activity similar to their healthy peers and increase the K/G ratio with walking speed during overground walking. Furthermore, external ankle load helps children with DS increase the elastic property of general muscular activity and match their healthy peers while walking fast on a treadmill. We concluded that the FDHO model shows its theoretical and clinical relevance in the assessment of motor adaptability in children with DS and should be included in future studies of neuromuscular activity in this population.
Footnotes
Dr Wu provided concept/idea/research design, project management, fund procurement, participants, facilities/equipment, and institutional liaisons. Dr Wu and Dr Ajisafe provided writing. All authors provided data collection, data analysis, and consultation (including review of manuscript before submission). The authors are grateful to all of the participants and their families for their participation in this study.
This study was approved by Georgia State University's Institutional Review Board.
This study was supported, in part, by a research grant from the Jerome Lejeune Foundation and a research initiation grant from Georgia State University.
- Received May 6, 2014.
- Accepted November 26, 2014.
- © 2015 American Physical Therapy Association