Abstract
BACKGROUND:
Postural instability is an important pathomarker in children with cerebral palsy (CP), and is often implicated in gait disturbance.
OBJECTIVE:
The purpose of this study was to investigate the therapeutic effects of long-term robotic hippotherapy (HPOT) on postural muscles size and static and dynamic postural stability in a child with CP.
METHODS:
Ultrasonography was used to measure postural muscles size. We also evaluated the magnitude of the separation between the center of pressure (COP) and center of mass (COM) during quiet stance and gait initiation (GI) using an eight-camera motion capture system and two force plates. Robotic HPOT was provided as a 45-minute session once per week for 12 weeks.
RESULTS:
As transverse abdominal (12%) and lumbar multifidus (60%) muscles size improved, normalized sway area (16%) during the quiet stance decreased. Similarly, the maximal resultant COP-COM distance (12.84%) during the initial phase of GI increased.
CONCLUSIONS:
In a child with CP, robotic HPOT may be an important treatment for improving postural muscles size and postural stability in static and dynamic states.
Introduction
Postural instability is an important pathomarker in children with cerebral palsy (CP) that is often implicated in gait disturbance (Woollacott & Shumway-Cook, 2005). Neurodevelopmentally, postural stability is established with continued central nervous system (CNS) maturation at about 4.5 months (Kolar & Kobesova, 2010). Postural stabilization involves a subconscious feedforward mechanism by which the diaphragm, transversus abdominis (TrA), multifidus, and pelvic floor muscles are synergistically co-activated in coordination with superficial muscles for upright spinal postural stability (Kolar & Kobesova, 2010; Frank, Kobesova, & Kolar, 2013). Upright spinal postural stability is preceded by dynamic locomotor movement (Kolar & Kobesova, 2010). However, in children with CP, whose normal CNS development is structurally compromised, abnormal postural stabilization often manifests as excessive anterior pelvic tilt, forward head posture, chest fixation in the inspiratory position, and abnormal spinal curvature during both standing and walking (Woollacott & Shumway-Cook, 2005). As such, underlying postural instability is believed to be associated with altered core muscles activation patterns due to cortical disinhibition and associated sensorimotor coordination dysfunction (Dietz & Sinkjaer, 2007). Clinical evidence demonstrated excessive muscle imbalance between underactive deep core stabilizers (TrA and multifidus) and hyperactive superficial spinal extensors due to insufficient coordinated co-activation of the first and second oblique anterior and posterior chains in children with CP (Prosser, Lee, VanSant, Barbe, Lauer, 2010). To improve postural stability during gait performance, hippotherapy (HPOT) has been used to provide more engaging therapeutic stabilization and has yielded promising results (Encheff, Armstrong, Masterson, Fox, Gribble, 2012; Shurtleff & Engsberg, 2010).
However, the use of live horse in HPOT leads to inherent issues with accessibility, labor-intensive, reimbursement, weather dependence, and risk of injury. To mitigate these inherent issues, we recently studied a robotic HPOT system that is designed to simulate a live horse to achieve comparable therapeutic effects in children with CP. Specifically, the robotic HPOT system (FORTIS-102; Daewon Fortis, Hanam, South Korea) used in the present study demonstrated consistent anteroposterior and vertical movement patterns and similar time-to-peak acceleration movements compared to live horses (Park, Shurtleff, Engsberg, Rafferty, You JY, You IY, You SH, 2014). Moreover, robotic HPOT can be programmed for 100 different movement patterns (i.e., walking, trotting, and cantering) and with variable speed adjustments, providing variable movement challenges. These diverse vertical and sagittal movement patterns should be beneficial for the therapeutic sensorimotor facilitation required to improve postural development for uprighting as well as sagittal stabilization muscles (TrA and multifidus) development essential for locomotor control in children with CP (Kolar & Kobesova, 2010). Repetitive exteroceptive, proprioceptive, and vestibular stimuli can provide an enriched sensorimotor learning experience and contextual interaction to further normalize postural development because sensorimotor integration is compromised in children with CP (Ko, Sim, Kim, & Jeon, 2016; Bordoloi & Sharma, 2012). Recent motor learning evidence suggested that at least 400–600 repetitions per session are required to generate neuroplastic changes and ensure associated motor skill acquisition but are not affordable in the typical “manual” neurorehabilitation (Lang, MacDonald, & Gnip, 2007). Despite the advantages of robotic HPOT, its therapeutic benefits on postural stabilization during balance and locomotor control in CP remain unknown.
Therefore, the purpose of this study was to investigate therapeutic effects of long-term robotic HPOT on postural muscles size and static and dynamic postural stability in CP. We hypothesized that improvements in postural muscles size after robotic HPOT might improve static and dynamic postural stability in the quiet stance and initial phase of gait initiation (GI) of a child with CP.
Methods
Case description
An 11-year-old child with spastic CP was recruited (Table 1). The inclusion criteria were as follows: (1) spastic diplegia CP; (2) 5–12 years of age; (3) able to sit independently; (4) able to follow instructions; and (5) able to sit astride the robotic horse. The exclusion criteria were as follows: (1) previous experience riding robotic horses; (2) previous surgical, botulinum toxin, or baclofen intervention within 6 months; and (3) severe neuromuscular, cognitive, or visual problems that would affect the experiment. The study was approved by the Human Studies Committee of Washington University School of Medicine and the Institutional Review Board of St. Louis University. Informed consent was obtained from the child’s parents prior to participation.
Demographic and clinical characteristics of the participant
Demographic and clinical characteristics of the participant
BMI, body mass index; GMFCS, gross motor function classification system.
The intervention was performed by licensed occupational and physical therapists experienced with the robotic HPOT system (FORTIS-102; Daewon Fortis, Hanam, Kyungi, South Korea). The system was engineered to simulate live horse movements, including a walk (6 km/h), trot (13 km/h), canter (25 km/h), and gallop (40 km/h), as two-dimensional (anteroposterior, vertical) movement patterns with 100 different exercise modes (Park et al., 2014). Assigned therapists performed robotic HPOT treatment following the adopted industry standard methods and specific protocols of HPOT (Shurtleff & Engsberg, 2010). When mounted on a robotic horse during variable movement patterns and speeds in various positions (e.g., forward astride, reverse astride, side sit, tall kneel, and quadruped), often with transitions between positions and sometimes while the robotic horse is moving, the participant was instructed to practice catching, throwing, and placing balls or rings starting from the middle to lateral and crossing the midline (Shurtleff & Engsberg, 2010). The participant was allowed to interact with the therapists or parents nearby while on the robotic horse, and progressed along the hierarchy of postural control tasks in all directions. This intervention was provided in 45-minute sessions once per week for 12 weeks. A pre-test was administered before the 12-week robotic HPOT intervention, which was followed by a post-test.
Data acqusition, processing and analysis
The ultrasound imaging (LOGIQ 200 PRO, GE Medical Systems, WI, USA) with a 7.5-MHz linear transducer was used to measure TrA, internal oblique, external oblique, and lumbar multifidus (LM) during resting states. The child was comfortably placed in a relaxed hook lying position with approximately 40–80°of hip and knee flexion to eliminate lumbar lordosis. To measure abdominal muscle thickness, the inferior borders of the rib cage and iliac crest were palpated as a reference point. The transducer head was transversely positioned 25 mm anteromedial to the midway point between the 12th rib and the iliac crest (Ferreira, Ferreira, & Hodges, 2004). To determine the cross-sectional area of the LM muscle, the child was positioned in the prone position with a pillow under the abdomen to minimize the lumbar lordotic curve. The LM muscle was measured longitudinally from the tip of the L4–5 zygapophyseal joint to the superior musculofascial border of the multifidus muscle. The images were acquired at the end of the exhalation phase (Wallwork, Stanton, Freke, & Hides, 2009).
Images of postural muscle thickness were recorded on the screen from which muscle thickness (mm) and cross-sectional area (CSA; mm2) were measured using an electronic caliper. One investigator performed all of the muscle measurements under the supervision of the other investigator to ensure standardization; three consecutive measurements were obtained and averaged for the right and left sides. The reliability and validity of ultrasound imaging for assessing the lateral abdominal muscles and LM are well established (Koppenhaver, Hebert, Fritz, Parent, Teyhen, Magel, 2009; Teyhen, Gill, Whittaker, Henry, Hides, Hodges, 2007; Kiesel, Uhl, Underwood, Rodd, Nitz, 2007).
To measure changes in postural control, we used an eight-camera optical infrared motion capture system (MAC Eagle Digital Cameras, Motion Analysis Corporation, Santa Rosa, CA) and two 50 cm×50 cm force plates (Kistler, Winterthur, Switzerland). Three-dimensional kinematic data were acquired at 60 Hz and the kinetic data were simultaneously recorded at a sampling rate of 300 Hz and 26 reflective makers (9 mm) were placed on anatomical landmarks of the child’s head, trunk, and lower extremities. GI trials were performed along a 1-m walkway containing force plates surrounded by motion capture system. The force plates were embedded in the floor, camouflaged with carpet, so the child was not aware of them. The child stood barefoot on the force plate for 20 seconds in the quiet stance. With a verbal cue from the therapist, the child began walking a 1-m walkway for GI trials at a self-selected pace. The force plates captured center of pressure (COP) simultaneously with center of mass (COM), captured by a motion capture system. For the child, one to two practice trials were allowed to ensure that the child understood the procedure. Thus, 20 trials were repeated until five samples of accurate data were obtained.
Kinetic and kinematic data were digitally low-pass filtered using a fourth-order Butterworth filter with zero phase-lag at the cutoff frequency of 20 Hz and 7 Hz, respectively. The kinetic data were down-sampled to 60 Hz for comparison with the kinematic data. The Cortex software (Version 1.0.0.198, Motion Analysis Corporation, Santa Rosa, CA) integrated the kinematic and kinetic data. The ground reaction forces and moments collected from the force plates were processed and the participant’s static postural control was analyzed using COM and COP anteroposterior (AP) sway, mediolateral (ML) sway, and subsequent sway area (rectangle created by minimal and maximal AP and ML sway). During GI, the marker trajectory on the sacrum was based on previous studies that found this location to approximate whole-body COM as measured by segmental analysis (Jensen, 1986). Postural sway was determined by the displacement of COM from COP, which was calculated using the horizontal displacement between the centroids (two-dimensional averages) of COP and COM during the collection period. The distance (COP-COMMAX) between the COP and COM in the transverse plane in the AP and ML directions, and resultant moment arms during quiet stance and the initial phase of GI were calculated and normalized to the participant’s foot length (AP direction), stance width (ML direction), respectively. The actual robotic HPOT induced improvement (%) in postural muscles size and variables for static and dynamic postural control were determined using the following equation: (corresponding variable post-robotic HPOT – corresponding variable pre-robotic HPOT)/(corresponding variable post-robotic HPOT)×100.
Results
Postural muscles ultrasound imaging
As shown in Table 2, actual improvement in TrA thickness and LM CSA for the right (dominant) side were 11.55% and 60.31% from 31.87 (mm) and 105.97 (mm2) before robotic HPOT to 36.03 (mm) and 267 (mm2) after robotic HPOT, respectively.
Ultrasound imaging of the lateral abdominal and lumbar multifidus muscles
Ultrasound imaging of the lateral abdominal and lumbar multifidus muscles
Data are presented as mean (standard deviation). EO, external oblique; IO, internal oblique; TrA, transversus abdominis; LM, lumbar multifidus.
In quiet standing (Table 3; rows 3–8), the normalized range of AP sway (COPAP) decreased from 0.203 to 0.180, a 12.78% improvement. Similarly, the normalized range of ML sway (COPML) decreased from 0.141 to 0.127, a 11.02% improvement. Consequently, normalized sway area (COPSway) decreased from 0.029 to 0.025, a 16% improvement. The maximal COP-COM distance in AP and ML directions were improved by 41.67% and 25%, from 0.119 and 0.100 before robotic HPOT to 0.084 and 0.080 after robotic HPOT, respectively. The maximal resultant COP-COM distance (Fig. 1; top row) thus decreased from 0.013 to 0.007, an 85.71% improvement.
Comparison of the COP-COM distance (cm) during static quiet stance (top row) and dynamic gait initiation (bottom row) before (left column) and after (right column) robotic HPOT. COP, center of pressure; COM, center of mass; HPOT, hippotherapy; AP, anteroposterior; ML, mediolateral.
COP measures and COP-COM moment arms during quiet stance and gait initiation
Data are presented as normalized mean (standard deviation). COP, center of pressure; COM, center of mass; HPOT, hippotherapy; AP, anteroposterior; ML, mediolateral.
In the initial phase of GI (Table 3; rows 10–12), the maximal COP-COM distance in the AP and ML directions was improved by 4.5% and 7%, from 1.910 and 0.748 before robotic HPOT to 2.000 and 0.804 after robotic HPOT, respectively. Consequently, the maximal resultant COP-COM distance increased from 1.398 to 1.604, an 12.84% improvement (Fig. 1; bottom row), indicating that robotic HPOT was effective in restoring static and dynamic postural control.
This study provides evidence of the long-term effects of robotic HPOT on postural control for a child with spastic CP. Consistent with the proposed hypothesis, robotic HPOT indeed improved postural muscles size and associated static and dynamic stability in a child with CP.
Ultrasound imaging data demonstrated that improvements in TrA thickness and LM CSA were 11.55% and 60.31% in the right (dominant) side, respectively. The intensive, repetitive (3000–5000 repetitions), and variable movement (100 different cycles) of the robotic HPOT might challenge to activate postural muscles constantly so that the child can upright the spine and maintain upright stability (Park et al., 2014). Moreover, postural muscles size change is often associated with postural muscle strength and anticipatory postural adjustments, which are mediated by a subconscious feedforward mechanism resulting from variable perturbing movement. This finding is consistent with previous investigations linking postural muscles and stability. Using magnetic resonance imaging, Lee et al. reported that improved spinal alignment and associated muscles size resulted from robotic HPOT in child with postural deficits (Lee, Lee, Cha, You, Oh, Bang, 2011). Other studies have demonstrated that trunk-targeted interventions are effective for increasing abdominal muscles thickness and postural control in children with CP (Unger, Jelsma, & Stark, 2013; dos Santos, Serikawa, & Rocha, 2016).
Static postural sway analysis revealed decreases in COPAP (12.78%), COPML (11.02%) sways and resultant sway area (16%) during the quiet stance after the robotic HPOT. Concurrently, the maximal resultant COP-COM distance decreased about 85.71% after intervention, confirming the therapeutic effects of robotic HPOT. These improvements were a result of increased postural muscles size maintaining equilibrium, while at the same time child with CP may have controlled using the subconscious feedforward mechanism by which the TrA and multifidus are synergistically co-activated for upright spinal postural stability. That deep core stabilizers such as the TrA and multifidus contribute to static postural control is supported by Baek et al., who reported a significant improvement in abdominal muscular thickness and postural stability after horseback riding simulator training in patients with stroke (Baek & Kim, 2014). Similarly, Elshafey demonstrated a significant improvement in trunk alignment morphology and balance after hippotherapy stimulator treatment in children with hemiplegic CP (Elshafey, 2014).
In the initial phase of GI, the maximal COP-COM distance in the AP and ML directions was improved by 4.5% and 7% after robotic HPOT, respectively. Consequently, the maximal resultant COP-COM distance increased by 12.84%. Neurodevelopmentally, this is an important finding because it supports that improvement in deep core stabilizers size after robotic HPOT contributes to dynamic postural stability, which is preceded by dynamic locomotor movement (Kolar & Kobesova, 2010). With changes in postural position, the distance between COP and COM increases, making the body mass intrinsically unstable and requiring active postural control to return the COM to a stable position within the base of support. Unlike the quiet stance, with greater uncoupling of the COP and COM, greater active postural control is needed for dynamic balance maintenance (Martin, Shinberg, Kuchibhatla, Ray, Carollo, Schenkman, 2002). During the meaningful initial phase of GI, the COP initially shifted posterolaterally towards the stepping foot, whereas COM moved anterolaterally toward the stance foot to generate forward momentum. Although the present study was not a controlled clinical trial, our findings regarding the distance between COP and COM during GI provided clinical confirmation of previous studies suggesting that COP-COM distance is smaller in people with postural deficits compared to normal controls (Hass, Waddell, Fleming, Juncos, Gregor 2005; Hsue, Miller, & Su, 2009). Based on this evidence, it is possible that therapeutic robotic HPOT also increases separation of the COM and COP, possibly by enhancing active postural control of deep core stabilizers. Hubble et al. reported that progressive trunk exercises improved dynamic postural control and trunk muscle function in people with Parkinson’s disease (Hubble, Naughton, Silburn, & Cole, 2014). Hesary et al. also reported that core stabilization training focusing on deep trunk muscles was effective in improving balance in elderly women vulnerable to falls (Hesari, Mahdavi, & Abadi, 2012).
In light of our findings, several limitations need to be considered in future studies. First, this case report suggests that the therapeutic potential of robotic HPOT is promising, but randomized controlled trials with a larger sample will be needed to generalize our findings. A second limitation of our study is that postural core stability was not directly measured, but rather estimated from the COP value and COP-COM distance measurement. Further investigation is needed to validate the relationship between postural core stability and COP-COM distance. Lastly, we only investigated kinetic, kinematic and ultrasonography variables with respect to the observed differences before and after intervention. Future studies should consider clinical motor function and activity tests in observing to improvement of postural core stability.
Conflict of interest
The study was supported by the National Research Foundation of Korea through Brain Korea 21 Plus Project (grant no. 2016-51-0009) and Basic Science Research Program (grant no. 2017-51-0187). No conflicts of interest have been reported by the authors or by any individual in control of the content of this manuscript.
Author disclosures
This research received financial and administrative supports from the Brain Korea 21 PLUS Project (grant no. 2016-51-0009) and the Basic Science Research Program (grant no. 2017-51-0187) sponsored by the Korean Research Foundation for the Department of Physical Therapy in Graduate School, Yonsei University. This manuscript has been submitted solely to this journal and has not been published, or submitted elsewhere.
Footnotes
Acknowledgments
We would like to thank all of the professionals who assisted with data collection.