Combined Ankle/Knee Stretching and Pivoting Stepping Training for Children with Cerebral Palsy

Abstract

Objective:

Although various treatment methods have been investigated to reduce spasticity and intoeing gait in children with cerebral palsy (CP), methods to concurrently reduce an intoeing gait and associated ankle/knee stiffness and spasticity according to a child’s specific needs are lacking. This study aimed to develop a training program to improve walking function and transverse-plane (pivoting) neuromuscular control and reduce spasticity and intoeing gait deviations.

Methods:

Eight children with diplegic CP and intoeing gait participated in this 6-week combined robotic ankle and/or knee intelligent stretching and pivoting neuromuscular control training program (Subject-specific Stretching and Pivoting Off-axis Neuromuscular control Training, [SS-POINT]). The effects of SS-POINT were evaluated using neuromechanical, functional, and clinical outcome measures and compared to those for eight children with CP and intoeing gait who participated in pivoting neuromuscular control training (POINT) alone in a previous study.

Results:

After the SS-POINT program, subjects with CP showed reduced knee stiffness and intoeing angle, and improvements in both joint and leg functions in terms of ankle and knee active range of motion, ankle dorsiflexor strength, proprioception, walking speed, balance, and minimum pivoting angle. Furthermore, improvements in proprioceptive acuity and minimum pivoting angle after the SS-POINT were greater than those after the POINT.

Conclusion:

The SS-POINT approach can be used as a subject-specific training program for improving leg and walking functions and reducing intoeing during gait.

Significance:

This approach can serve as an individualized intervention at the joint and walking levels to maximize intervention effects by adjusting training targets, sequences, and intensities to meet the individual needs of children with CP.

I. Introduction

CEREBRAL palsy (CP) is the leading cause of childhood orthopedic disabilities. The prevalence of CP is in the range of 1.5 and 3.6 cases per 1000 in the US, with an estimated lifetime economic costs of $921,000 per case [1-3]. Children with CP often have many neuromuscular and musculoskeletal problems such as spasticity, dystonia, muscle contractures, bony deformities, poor balance, loss of selective motor control, and muscle weakness [4]. In particular, spastic hypertonia in the affected limbs is the most common disability associated with CP [1, 5]. A lack of mobilization and prolonged spasticity may be accompanied by structural changes to the muscle fibers and connective tissues that may reduce joint range of motion (ROM) and lead to clinical contracture and severe pain [4, 6-8].

In addition to the spastic hypertonia found in children with CP, the prevalence of an intoeing gait is reportedly 64%; these children often demonstrate a reduced ambulatory ability [9-11]. An intoeing gait in children with CP is a common problem associated with frequent falling [12], patellofemoral and hip pain, and osteoarthritis that interferes with their functional mobility, independence, and ability to participate in social activities [13-16]. Common causes of an intoeing gait in children with CP include internal hip rotation, internal tibial torsion, and spasticity of the hip internal rotators and adductors and the medial hamstrings [11, 17-19].

Various methods including Botox injection, surgery, physical therapy, orthosis, shoe modifications, and casting are available to manage spasticity and reduce intoeing gait deviations [4, 18, 20, 21]. However, no objective assessments and intervention tools with subject-specific treatment methods are currently available for children with CP to concurrently improve leg and walking functions and reduce intoeing gait deviations.

There is a recent trend of utilizing robotic training to improve ankle ROM and leg function [22-24], and intoeing gait [25] in a separate manner. However, considering that children with CP often present with both stiff lower limb joint(s) and an intoeing gait, it is unknown whether the combined joint stretching and multi-joint exercise for improving the twisting direction of movement effectively reduces spasticity and intoeing angle during walking. Therefore, there is a strong need for the development of objective assessments and targeted intervention protocols to improve leg and walking functions and reduce intoeing gait deviations in children with CP that meet their specific needs.

This study aimed to develop a training program utilizing robotic devices to improve walking functions and transverse-plane (pivoting) neuromuscular control and reduce spasticity and intoeing gait deviations, namely SS-POINT. We hypothesized that, following the 6-week training program, the combined stretching and off-axis elliptical training program will: 1) improve walking functions as increase in walking speed and reduction in 20-m walk time and intoeing angle; 2) increase pediatric balance scale; 3) improve pivoting neuromuscular control as decrease in pivoting instability and increase in pivoting proprioceptive acuity; and 4) improve joint functions by increasing ankle and knee active ROM, improving muscle strength, and reducing spasticity measured by the Modified Ashworth Scale (MAS). Furthermore, the effects of SS-POINT were compared to those of the training to improve pivoting control (POINT) among children with CP in the previously reported study [25].

II. Methods:

A. Subjects

Eight children with diplegic CP participated in the 6-week combined robotic ankle and/or knee intelligent stretching and off-axis neuromuscular control training program at the Rehabilitation Institute of Chicago. Each subject was recruited by the pediatric physiatrist. Exclusion criteria were other unrelated neurological impairments or musculoskeletal injuries, severe cardiovascular disease, hypertension, atrial fibrillation, and congestive heart failure. Detailed subject characteristics are shown in Table 1. All subjects read and signed an informed consent form approved by the Institutional Review Board of Northwestern University prior to the experiment.

TABLE I.

Subjects’ characteristics

Subject	Age (yr)	Height (cm)	Mass (kg)	Assistive device*	Sex	GMFCS	Impairment type	No. of study sessions
S1	18	172.1	58.5	None	M	I	Diplegia	16
S2	13	161	53.5	None	F	II	Diplegia	19
S3	10	150	35.1	None	F	I	Diplegia	18
S4	11	156	60.1	Hand	F	III	Diplegia	19
S5	11	149	39.1	Walker	F	III	Diplegia	19
S6	13	155	52.8	None	F	II	Diplegia	19
S7	14	138	41.8	Walker	M	III	Diplegia	19
S8	14	168	54.1	Crutch	M	III	Diplegia	18
Mean ± SD	13 ± 2.5	156.1 ± 10.9	49.4 ± 9.4		5F, 3M			18.4

^*Assistive device during evaluation; GMFCS, Gross Motor Function Classification System

B. Study Design

Each subject participated in 18 sessions of SS-POINT (3 sessions/week for 6 weeks). The program consisted of 16 training sessions, a pre-evaluation session (E1), and a post-evaluation session (E2); an ankle and/or knee intelligent stretching robot (Figs. 1(A), 1(B)) and a novel off-axis elliptical trainer (ET) (Fig. 1(C)) were used. The duration of each training session was approximately 45–60 minutes including exercise and rest. Furthermore, a follow-up evaluation (E3) was also conducted 6 weeks after E2. Thereby, each subject could participate in a total of 19 study sessions. Two subjects did not participate in E3 due to patient request (S1) and loss of contact (S3). Detailed information for each robotic system is available elsewhere [23, 26]. Briefly, the ankle intelligent stretching robot was developed to reduce ankle passive stiffness, increase ankle ROM, and improve gait function [23]. Similarly, the knee intelligent stretching robot was developed to reduce knee passive stiffness and increase knee/gait function. The off-axis ET system is used to improve pivoting neuromuscular control and walking function [27].Furthermore, this study compared several common parameters obtained from the previously published historical control group [25] and that from the current study group. The subjects in the historical control group participated in the 18-session training study on the off-axis ET to improve pivoting neuromuscular control. Thereby, in this study, the historical group is named the off-axis ET (POINT) group. The mean age was 15.5 ± 4.1 yr, height was 159.0 ± 13.5 cm, and mass was 53.9 ± 12.3 kg; 4 females and 4 males with Gross Motor Function Classification System (GMFCS) levels of I–III participated in the study [25]. Briefly, following the training program on the off-axis ET (POINT), Tsai et al. [25] reported significant improvement in pivoting neuromuscular control as a reduction in pivoting instability and pivoting ROM during stepping on the off-axis ET and an increase in leg off-axis strength. Significant improvements were also found in reduced 10-m walk time and intoeing angle as well as an increase in Pediatric Balance Scale (PBS) score. No follow-up evaluation was reported.

Fig. 1.

Each intelligent stretching robot consists of a footplate, support/fixation, control unit, and safety stop. (a) Ankle joint equipment. (b) Knee joint equipment. Each control unit consists of a torque sensor, encoder, and digital signal processor to control joint movement, communicate with the display computer, and measure isometric joint torque and passive and active ranges of motion. (c) Off-axis ET system with a subject during a typical pivoting neuromuscular training session with visual feedback. Each subject was instructed to align their foot pivoting angle (yellow dotted line) with the green line target. Each subject received a virtual golden coin as a reward when their feet were successfully maintained in the desired position.

C. Training Protocol

Depending on the more affected side and joints based on each child’s MAS score of the ankle and knee joint at E1 conducted by the physiatrist, the ankle and/or knee joint was passively stretched [6] using the ankle or the knee intelligent stretching robot (Figs. 1(A), 1(B)) for 15 minutes. At each training session, each subject received the different stretching sequence (ankle or knee, left or right side) depending on his/her symptoms (Table 2). The subject’s foot was strapped to the footplate, and the robot adjusted to fit each subject so that the axis of rotation of the robot was aligned with the ankle joint. Similarly, if the knee robot was used, the subject’s foot was strapped to the footplate and the lower leg was strapped to the sidebar so that the axis of rotation of the robot was aligned with the knee joint. Before stretching, the subject’s ankle or knee joint was pushed manually by the operator to the extreme position. During stretching, the subject was instructed to relax, and the robot moved the ankle or knee following the intelligent stretching algorithm. The stretching velocity was inversely proportional to the resistance torque for strenuous and safe passive stretching and the end position can be a few degrees beyond the manually predetermined ROM within the resistance torque limits [6, 22-24].

TABLE II.

Stretching sequence of each subject

	S1				S2				S3				S4				S5				S6				S7				S8
	L A	R A	L K	R K	L A	R A	L K	R K	L A	R A	L K	R K	L A	R A	L K	R K	L A	R A	L K	R K	L A	R A	L K	R K	L A	R A	L K	R K	L A	R A	L K	R K
T1
T2
T3
T4
T5
T6
T7
T8
T9
T10
T11
T12
T13
T14
T15
T16

Stretching sequence of each subject for the left ankle (LA), right ankle (RA), left knee (LK), and right knee (RK) at each training session for the most affected joint depending on subjects’ initial Modified Ashworth Scale score and feedback with predetermined range of motion for the robot. There were some sessions when some subjects did not receive stretching protocol to focus on elliptical trainer training. S1 to S8 are the subjects; T1 to T16 are the training sessions.

After completing the passive stretching procedure, the subjects wore a safety harness and stood on the footplates of the off-axis ET (Fig. 1(C)) with each tibial long axis aligned with the center of the pivoting axis on each footplate. In this study, no subjects used body weight supports; rather, they used only a safety harness. They were instructed to maintain their feet in the forward direction position (green line) (Fig. 1(C)) while stepping on the off-axis ET. Each trial was 12–16 minutes long and consisted of three 3–4-minute challenging tasks during which to control pivoting movements when both footplates moved in pivoting direction and two 2-minute stepping tasks when the footplates were locked, similar to stepping on a conventional ET. Each shoe was strapped to the footplate with toe and heel straps so that the foot and the footplate were rotated together to help a child acquire motor skills to improve walking function while reducing intoeing gait deviations and preventing secondary injuries.

As described previously, a spring mode was used to teach the children to control their pivoting motions while stepping [25]. In this mode, the footplates were pushed from both sides with assistive spring torque. Difficulty level was adjusted by the normalized adjustable stiffness value range of 0–1, where stiffness value decreases as assistive spring torque increases the difficulty of controlling the footplates [25]. With less assistive spring torque, the footplates were free to pivot (free pivoting task [FPT]). During the FPT, subjects felt that they were walking on ice due to the minimally assistive spring torque of the footplates. During the assistive spring torque task (ASST), subjects received assistant spring torque to maintain their target position because the restoring torque from the virtual springs helped them stay in the target position [26]. To increase training efficacy, real-time audio and visual biofeedback of the pivoting angles was displayed on a computer monitor in front of the ET (Fig. 1(C)). This feedback was further integrated with a simple video game to encourage active participation for the children with CP (Fig. 1(C)). A video camera was also placed in front of ET to provide biofeedback of the children’s lower limb alignment during the training sessions.

D. Outcome Evaluations

Functional, clinical, joint-level biomechanical, and walking-level neuromechanical measures were assessed to evaluate the combined joint stretching and pivoting neuromuscular control training program at E1, E2, and E3.

Functional measures included walking speed and intoeing angle assessed during the 20-m walk test using inertial measurement units (Mobility Lab; APDM Inc., Portland, OR, USA) [25] positioned atop each foot. At the beginning of each walking test, each foot was instructed to assume a second toe forward position, so the consistent initial position was considered to compute intoeing angle. Clinical measures included PBS and MAS scores of ankle dorsiflexion and plantarflexion and knee extension and flexion assessed by the physiatrist. The MAS was averaged from the left and right leg for further statistical analysis.

Joint-level biomechanical measures included active ROM (AROM) of the ankle and knee, ankle dorsiflexor strength (DS)/plantar-flexor strength (PS), knee flexor/extensor strength, and passive ROM (PROM) and passive stiffness measured at the more affected side of the leg from the ankle/knee intelligent stretching robot. Ankle AROM was assessed when the patient was instructed to move the ankle as much as possible without any assistance from the ankle robot from the neutral position defined as the 10° dorsiflexion angle to the dorsiflexion direction and from the neutral position to the plantarflexion direction. Similarly, knee AROM was assessed when the knee was instructed to be moved by the subject without any assistance from the knee robot from the neutral position as the 90° of knee flexion in the knee extension direction as much as possible. When the ankle robot was locked at 10° of dorsiflexion, the subject was asked to push the ankle in the dorsiflexion direction as hard as possible for 3 seconds to measure DS. Similarly, the subject was asked to push the ankle in the plantarflexion direction to measure PS. When the knee robot was locked at the 90° knee flexion, the subject was asked to push the knee in the extension direction as hard as possible for 3 seconds to measure the knee extensor strength (KES). Similarly, the subject was asked to push the knee in the knee flexion direction to measure the knee flexor strength. Furthermore, using stretching data from the training sessions, PROM and passive stiffness of the ankle and knee joint of the most affected side were evaluated. Considering differences in body size and relative strength among subjects, different peak torques were used for them; however, the level of torque used in the two training sessions (first and last available sessions in the training program) within a subject was the same [23]. Joint stiffness is defined as the slope of the ascending limb of the torque-angle relationship at 3° of dorsiflexion for ankle passive stiffness and that at 40° of knee extension for knee passive stiffness.

Walking-level neuromechanical measures included proprioceptive acuity, a minimum and maximum pivoting angle measured from the off-axis ET system. Proprioception was tested to assess how quickly the subjects could detect the subtle passive pivoting movement (1°/sec) on the off-axis ET. At the lowest position of the right footplate on the off-axis ET, subjects put all their weight on the right leg with full knee extension and minimally bear their weight on the left leg due to the mechanical structure of the off-axis ET [27]. The initial pivoting position was at the second toe forward position. In that posture, the subjects were asked to press a handheld button immediately and report the direction when they felt a subtle movement in four possible pivoting directions, namely, left and right internal pivoting and left and right external pivoting. To be considered a successful trial, the subjects were needed to correctly identify the direction and leg moved. A total of 16 trials could be performed, but some subjects were not able to complete all tasks due to the difficulty maintaining the posture during the experiment. Thereby, the number of completed proprioception tasks was also considered. Next, the same protocol was performed with a switched leg as at the lowest position of the left footplate on the off-axis ET, and the subjects put all their weight on the left leg with a full knee extension and minimally bore their weight on the right leg due to the mechanical structure of the off-axis ET. The proprioception tasks were performed in a quiet environment and the subjects were asked to close their eyes during the tasks to exclude visual feedback. The pivoting angle as proprioceptive acuity, the number of completed proprioception tasks, and the percentage of correct proprioception tasks (indicating correct direction and leg) were considered. The mean results from the left and the right legs were considered in the statistical analysis. An averaged stepping speed was computed per task for each subject based on the time intervals between successive events when the same footplate reached the most anterior position [28]. Peak vertical ground reaction force (GRF) during the stepping movement was also computed. Pivoting instability was quantified as the root mean square of the pivoting angle of the right leg during the FPT [26, 27]. Lower pivoting instability indicates better task performance in the FPT. Leg minimum and maximum pivoting angles were assessed when the subjects stepped on the off-axis ET while they needed to control the pivoting motion at a comfortable speed for 2 minutes. The minimum and maximum pivoting angles were computed during the loading and unloading phases. The loading phase is defined as the period from when a footplate reaches the most anterior position to when the same footplate reaches the most posterior position [28]. The unloading phase is the period from when a footplate reaches the most posterior position to when the same footplate reaches the most anterior position. Here the minimum pivoting angle indicates the maximum inward angle from the initial position, which is similar to the intoeing angle during walking. Similarly, the maximum pivoting angle indicates the maximum outward angle from the initial position.

E. Statistical Analysis

Due to the small sample size, non-parametric tests were performed. Friedman tests (non-parametric version of repeated-measure analysis of variance) were conducted to investigate whether the SS-POINT affected the aforementioned neuromechanical, functional, and clinical outcome measures with a significance level of p < 0.05. If p < 0.05, Wilcoxon signed rank tests were performed to investigate whether the effect on the specific measures were between E1 and E2, E1 and E3, or E2 and E3. An independent t-test was applied to compare the effect of the SS-POINT investigated in the current study versus the POINT in the previously reported study among children with CP [25] on the differences between E1 and E2 in the PBS, walking speed, and intoeing angle during overground walking, proprioception angle, stepping difference, pivoting instability, and minimum pivoting angle on the off-axis ET with significance level of p < 0.05. Furthermore, a hierarchical model [29] was also applied to evaluate potential bias of the results that could be raised due to inhomogeneity of the subjects between the POINT and SS-POINT groups. The model can be realized using a linear mixed model with group as a fixed factor subject as a random factor with GMFCS level as a covariate and aforementioned variables as dependent variables.

III. Results

A. Overground Walking Tests

After the SS-POINT program, the subjects were able to walk significantly faster during the 20-m walking test in terms of walking time (χ² = 6.522, p = 0.038), and walking speed (χ² = 6.522, p = 0.038). There were significant reductions in walking time from E1 to E2 (z = −2.103, p = 0.035) and from E1 to E3 (z = −2.201, p = 0.028). There were significant increases in the walking speed from E1 to E2 (z = −2.100, p = 0.036) and from E1 to E3 (z = −2.201, p = 0.028; Table 3). The subjects also significantly reduced the maximum intoeing angle during the 20-m walking test following SS-POINT (χ² = 9.333, p = 0.009). A significant reduction in the maximum intoeing angle was seen from E1 to E2 (z = −2.521, p = 0.012) and from E1 to E3 (z = −2.201, p = 0.028; Table 3).

TABLE III.

Test results

Measure	E1	E2	E3
Overground walking test
20-m walk (sec)	24.8 (21.5, 27.6)	22.6 (19.5, 24.4)*	22.3 (19.8, 25.5)*
Walking speed (m/s)	0.81 (0.72, 0.93)	0.89 (0.82, 1.03)*	0.90 (0.78, 1.01)*
Intoeing angle (°)	18.9 (7.0, 26.0)	11.8 (3.2, 22.0)*	16.8 (5.5, 19.3)*
Clinical tests
PBS	41.5 (30, 46.5)	44 (31, 50.5)*	40 (31, 47)*
MAS in ankle dorsiflexion	2.0 (1.8, 2.8)	1.9 (1.8, 2.8)*	1.0 (1.0, 1.5)+
MAS in ankle plantarflexion	1.3 (0.8, 2.0)	1.1 (0.3, 1.4)	0.3 (0.0, 1.0)
MAS in knee flexion	1.1 (0.3, 1.9)	0.8 (0.0, 1.9)	0.0 (0.0, 0.0)
MAS in knee extension	1.0 (0.9, 2.1)	0.5 (0.0, 1.4)	0.5 (0.0, 1.0)
Joint-level measures
Ankle joint AROM (°)	32.5 (29.0, 44.7)	56.6 (43.3, 67.5)*	60.3 (56.1, 75.0)*
Knee Joint AROM (°)	65.7 (53.6, 74.7)	65.9 (62.2, 81.5)	71.5 (62.2, 78.3)*
DS (Nm)	2.8 (1.8, 9.4)	7.4 (4.4, 12.8)	4.5 (3.0, 12)+
PS (Nm)	16.9 (14.2, 19.5)	19.2 (16.5, 22.2)	18.5 (17.2, 24.4)
KES (Nm)	36.0 (32.4, 40.5)	38.1 (30.1, 46)	34.3 (28.6, 54.0)
KFS (Nm)	14.8 (4.5, 19.4)	9.7 (8.5, 21.4)	7.1 (5.2, 17.1)

Median (25^th, 75^th quantile) in measures during overground walking tasks and clinical tests at E1, E2, and E3.

^*p < 0.05 indicates a significant difference between E1 and other evaluation.

⁺p < 0.05 indicates a significant difference between E2 and E3. PBS, Pediatric Balance Score; MAS, Modified Ashworth Scale; AROM, active range of motion; DS, dorsiflexor strength; PS, plantar-flexor strength; KES, knee extensor strength; KFS, knee flexor strength

B. Clinical Tests

The subjects showed significantly increased PBS scores following SS-POINT program (χ² = 7.364, p = 0.025). PBS increased from E1 to E2 (z = −2.120, p = 0.034) but decreased from E1 to E3 (z = −2.207, p = 0.027; Table 3). As seen in Table 3, there was a significant reduction of MAS in ankle dorsiflexion after the SS-POINT program (χ² = 7.900, p = 0.019). MAS in ankle dorsiflexion decreased from E1 to E3 (z = −2.023, p = 0.043) and MAS in ankle dorsiflexion decreased significantly from E2 to E3 (z = −2.032, p = 0.042; Table 3). MAS decreased in plantarflexion and knee flexion and extension (Table 3), but the changes were not significant.

C. Joint-Level Biomechanical Tests

AROM in the ankle and knee joint were significantly increased after the SS-POINT program (χ² = 7.600, p = 0.022 and χ² = 7.600, p = 0.022, respectively; Table 3). Ankle AROM increased significantly from E1 to E2 (z = −2.380, p = 0.017) and from E1 to E3 (z = −2.023, p = 0.043). Knee AROM increased significantly from E1 to E3 (z = −2.023, p = 0.043, Table 3). DS increased significantly after the SS-POINT program (χ² = 8.400, p = 0.015). A marginally significant increase in DS was seen from E1 to E2 (z = −1.820, p = 0.069), whereas DS significantly decreased from E2 to E3 (z = −2.023, p = 0.043; Table 3). There were increasing trends in PS and KES following the training, but they were not statistically significant (Table 3). There was a significant decrease in knee joint passive stiffness (Nm/Deg) at 40° of knee extension from the first to the last available sessions in the training program as from median (25^th, 75^th quantile) as 0.25 (0.17, 0.26) to 0.18 (0.08, 0.22, z = −2.023, p = 0.043) and a reducing trend in ankle joint passive stiffness (Nm/Deg) at 3° of dorsiflexion from 0.31 (0.25, 0.40) to 0.30 (0.26, 0.33). There were increasing trends in PROM (°) of the ankle joint from the first to the last available sessions in the training program of the most affected side from 25.4 (17.3, 42.3) to 33.5 (22.9,40.8) and in PROM (°) of the knee joint from 67.0 (56.9, 83.4) to 71.3 (65.6, 80.0).

D. Walking-Level Neuromechanical Tests

Although there was an improving trend in pivoting proprioceptive acuity in terms of reducing pivoting angle after the SS-POINT program, it was not statistically significant (Fig. 2). The number of completed proprioception tasks was significantly increased after the SS-POINT program (χ² = 8.435, p = 0.015). The number of completed proprioception tasks increased significantly from E1 to E2 (z = −2.201, p = 0.028) and from E1 to E3 (z = −2.023, p = 0.043). The percentage of correct proprioception tasks (indicating correct direction and leg) was significantly increased after the SS-POINT program (χ² = 6.870, p = 0.032). The percentage of correct proprioception tasks increased significantly from E1 to E2 (z = −2.197, p = 0.028) and from E1 to E3 (z = −2.201, p = 0.028).

Fig. 2.

(a) Proprioceptive acuity in degrees, (b) Number of completed tasks, (c) Percentage of correct tasks that the subjects identified in the correct direction and leg during the weight-bearing proprioception tasks at E1, E2, and E3. In the boxplot, three lines indicate the 25^th (Q1), 50^th (Q2), and 75^th (Q3) percentiles of the data. The whisker indicates Q3+1.5*(Q3-Q1), and Q1-1.5*(Q3-Q1) and the red cruciate points indicate the value outside of this range.

While there was no significant improvement in stepping speed during the FPT after the 6-week SS-POINT program (Fig. 3(A)), a significant increase in peak vertical GRF was found (χ² = 12.00, p = 0.002). Significant increases in peak vertical GRF were seen from E1 to E2 (z = −2.240, p = 0.025), E1 to E3 (z = −2.201, p = 0.028), and E2 to E3 (z = −2.201, p = 0.028). No significant improvement in pivoting neuromuscular control was seen during the FPT after the SS-POINT program in any evaluations. However, a significant reduction in pivoting instability was seen from E1 to E2 (z = −2.100, p = 0.036). There was a significant reduction in minimum pivoting angle (χ² = 7.00, p = 0.03) and an increase in maximum pivoting angle (χ² = 7.00, p = 0.03) after the SS-POINT program. A significant reduction in minimum pivoting angle occurred from E1 to E2 (z = −2.240, p = 0.025) and from E1 to E3 (z = −2.201, p = 0.028). A significant increase in maximum pivoting from E1 to E3 was noted after the SS-POINT program (z = −2.201, p = 0.028).

Fig. 3.

(a) Stepping speed in revolutions per minute (RPM), (b) peak vertical ground reaction force (GRF), (c) pivoting instability (deg), (d) minimum pivoting angle (deg) during the loading and unloading phases of the free pivoting task (FPT), and (e) maximum pivoting angle (deg) during the loading and unloading phases of the FPT at E1, E2, and E3. The black boxplots represent the loading phase, while the gray boxplots represent the unloading phase. In the boxplot, the three lines indicate the 25^th (Q1), 50^th (Q2), and 75^th (Q3) percentiles of the data. The whisker plot indicates Q3+1.5*(Q3-Q1) and Q1-1.5*(Q3-Q1), and the red cruciate points indicate the value outside of this range.

E. Comparison of Training Effects between the SS-POINT and POINT Groups

Differences in the common measures between E1 and E2 were compared between the SS-POINT and POINT groups (Table 4). There were significant intergroup differences in proprioception angle between E1 and E2 (t = 2.564, p = 0.028) as well as in the minimum pivoting angle during the FPT (t = −2.20, p = 0.047). No significant intergroup differences were found in improvements in PBS, walking speed, intoeing gait during overground walking, stepping speed, or pivoting instability during the FPT.

TABLE IV.

Comparison of changes

Measure		Combined Stretching and Off-axis ET group (SS-POINT)	Off-axis ET group (POINT)
Pediatric Balance Scale score	ΔEval E2-E1	3.38 (3.74)	2.13 (2.47)
Walking speed (m/s)		0.11 (0.08)	0.07 (0.03)
Intoeing angle (°)		−3.40 (5.44)	−4.46 (7.83)
Proprioception (°)		−3.66 (4.40)*,+	0.74 (2.06)
Stepping speed (RPM)		0.82 (4.54)	6.83 (10.05)
Pivoting instability (°)		−3.29 (3.00)	−2.71 (4.19)
Minimum pivoting angle (°)		5.88 (6.61)*,+	−0.93 (5.16)

Comparison of changes (improvements) from E1 to E2 m the combmed stretching and off-axis ET group (SS-POINT) and the off-axis ET group (POINT).

^*p < 0.05 from the independent t-test,

⁺p < 0.05 from the hierarchical model. RPM, revolutions per minute

IV. Discussion

After completing the 6-week SS-POINT program, subjects with CP demonstrated improved walking speed and a reduced intoeing angle. The improved gait function and reduction in intoeing angle following the training were also accompanied by reduction in passive knee stiffness and improvements in muscle strength, joint ROM, functional balance, proprioception, and leg pivoting neuromuscular control. These findings support our hypotheses that the SS-POINT program improves walking and leg functions as well as pivoting neuromuscular control. To our best of knowledge, this is the first study incorporating stretching and pivoting direction exercise at subject-specific needs to reduce spasticity and intoeing angle gait deviation simultaneously. The current approach can serve as a foundation for individualized leg and walking interventions to maximize intervention effects by adjusting training targets, sequencing, and intensity for ankle/knee stretching and pivoting neuromuscular training at subject-specific needs. Furthermore, the current approach can be combined with other conventional treatments such as bracing or surgery to improve leg and gait functions of children with orthopedic disabilities.

While the previous approach of combined passive-active training focused on the ankle joints of children with CP [22-24], the unique feature of the current study is that the training protocol was designed to improve joint and leg functions of children with CP who had more complex symptoms of ankle and/or knee spasticity with intoeing angle. Although studies have shown that passive stretching can only increase joint ROM [30], it is recommended that a stretching protocol be included as an adjunct to other treatment techniques for reducing soft-tissue tightness and managing spasticity and contractures [30, 31]. The focuses of the SS-POINT program were first to loosen the stiff ankle or knee joint depending on the more affected side and joint for 15 minutes and then improve the pivoting leg neuromuscular control for 12–16 minutes. Previous studies using the robotic ankle intelligent stretcher revealed that the 6-week SS-POINT program reduced MAS in the ankle joint and improved the ankle and walking function of children with CP who had ankle impairments [22-24]. With this trend, the combination of the passive joint stretching and pivoting leg neuromuscular training in our study likely helped improve the ankle and knee AROM and strength and reduce the MAS of ankle dorsiflexion with a reducing trend in MAS for ankle plantarflexion, knee extension, and flexion accompanied by a reduction in knee extension passive stiffness and a reducing trend in ankle dorsiflexion passive stiffness. However, a significant reduction in DS from E2 to E3 may indicate that a more frequent or longer training period might be necessary. Further studies that aim to identify a more systematic optimal stretching sequence to facilitate the functional neuromuscular training might be necessary. Despite this limitation, this is the first study to investigate different combinations of stretching and pivoting motions, taking us one step further to a subject-specific intervention.

This is the first study to report on the effects of SS-POINT on proprioceptive acuity under weight-bearing conditions among children with CP, although few studies have reported upper-limb deficits in proprioceptive acuity in children with CP [32, 33]. With the trend toward improved proprioceptive acuity in terms of reducing the pivoting proprioception angle, the interesting finding from the study was that, after the SS-POINT program, the number of completed proprioception tasks increased significantly from a median of 5.8 to 11.3, while the percentage of correct proprioception tasks increased from a median 63.3% to 92.2%. These results indicate that, prior to the SS-POINT program, the subjects were only able to complete 36.3% of the proprioception tasks while detecting only 63.3% of the correct directions and legs. A few factors can contribute to the poor percentage of tasks completed. Since the task required subjects to put their own weight on one leg, when subjects did not have sufficient strength to maintain the required postures on the off-axis ET, those with higher GMFCS levels who needed walking aids were not able to complete the task due to a lack of endurance. Compared to the previous study investigating the effects of POINT in children with CP [25], subjects who participated in the SS-POINT significantly improved their proprioceptive acuity. Based on our study findings, adding a joint-stretching protocol to the POINT program might better improve proprioceptive acuity by increasing the sensitivity of the peripheral muscle spindles and other proprioceptive receptors around the leg joints coupled with central facilitation of the neural information and selective attention to tasks [27, 34, 35]. During the SS-POINT program, the subjects were instructed to learn to control their pivoting directions of movement. Our unique weight-bearing test condition is functionally relevant since the proprioception test investigated resultant integrated somatosensory inputs from the hip, knee, and ankle [27]. Children with CP following the SS-POINT program may better sense pivoting positions during foot-ground contact to properly walk and prevent secondary injuries such as falls caused by a severe intoeing angle.

Following the SS-POINT program, significantly improved PBS with reduced intoeing angle, pivoting instability, and minimum pivoting angle were found. It is notable that the significant increase in PBS and reduction in pivoting instability from E1 and E2 was not retained at E3. While overall improvement in PBS was noted, its overall significance in pivoting instability considering all evaluations was not determined. The result implies that it is possible that frequent pivoting neuromuscular training might be necessary to maintain the training effects, and more than a 6-week training period performed on a regular basis might be required to improve balance and pivoting neuromuscular control under slippery conditions. Based on the fact that both the POINT program from the previous study [25] and the SS-POINT program of the current study in children with CP improved PBS without intergroup differences, our functionally relevant weight-bearing pivoting neuromuscular training paradigm could help improve functional balance in terms of PBS.

Although the inhomogeneity of subjects between the current and previous study groups [25] were considered in the statistical analysis, it is possible that the results were influenced by the inhomogeneity. Further randomized control clinical trials with larger sample sizes or a longitudinal cross-over design to control inhomogeneity of a study group might be necessary to investigate the benefits of adding a stretching paradigm to the POINT program on PBS and other measures. In addition, a longer than 6-week training period might be beneficial to determine therapy schedules and dosages whether it might be necessary to continue SS-POINT on a regular basis or a certain intervention period could provide permanent improvements in leg and walking functions and reduced intoeing during gait. A strength of this study is that it measured both sensory and neuromuscular performance outcomes under a functionally relevant weight-bearing condition in addition to joint-level clinical outcomes. Combined joint stretching and pivoting neuromuscular control reduced not only minimum pivoting angle in the loading phase on ET, it translated into improved walking functions in terms of reducing intoeing angle and increasing walking speed accompanied by improved balance and proprioception.

V. Conclusion

There is a strong need for practical, safe, and convenient rehabilitation protocols to improve leg and walking functions of the CP population and children with orthopedic disabilities. It is unknown from this study whether the current approach that combines stretching and pivoting neuromuscular training is superior to another previously reported intervention method that is used to reduce spasticity and intoeing gait deviations. Despite the small sample size of the current study, our findings are significant to the development of a conservative exercise intervention approach for children with orthopedic disabilities to improve their leg and walking functions and reduce intoeing gait deviations.