Altered balance control in thoracic adolescent idiopathic scoliosis during obstructed gait

Adolescent idiopathic scoliosis (AIS) is the most common spinal deformity during adolescence, leading to altered postural control with compromised stability. To identify the effects of AIS on whole-body balance control during obstacle-crossing, 14 adolescents with Lenke 1 thoracic AIS and 14 healthy controls were compared in terms of the inclination angle (IA) of the body’s center of mass (COM) relative to the center of pressure (COP), the rate of change of IA (RCIA) and the jerk index of IA. Between-side comparisons were also performed for the AIS group. The patients showed less smooth COM-COP motion in the sagittal plane with significantly increased anterior RCIA and IA jerk index during crossing with either the concave side (p = 0.001) or the convex side (p = 0.001) leading when compared to healthy controls. In the frontal plane, the patients showed close-to-zero RCIA (p = 0.002) while crossing with the leading limb, with an increased IA magnitude (p = 0.039) only while crossing with the concave-side limb leading. The patients with Lenke 1 thoracic AIS were found to cross obstacles with altered, compromised COM-COP control in both sagittal and frontal planes when compared to healthy controls. The results suggest that the thoracic spinal deformity in Lenke 1 AIS affects the whole-body balance control during obstacle-crossing, which should be monitored for signs of increased risk of loss of balance in the management of such patient groups.


Introduction
Adolescent idiopathic scoliosis (AIS) is the most common spinal deformity during adolescence, characterized by a lateral spinal curvature with an axial rotation and loss of sagittal kyphosis of the thoracic spine [1,2]. This three-dimensional spinal deformity often alters the orientations of the head, trunk and pelvis in all anatomical planes [3][4][5], as well as resulting in vestibular and trunk muscle imbalance [6][7][8], and proprioceptive disorders [9], leading to postural control problems [10] with compromised postural stability [10,11].
Poor body segmental alignment as a result of AIS has been associated with increased lateral displacement of the body's center of mass (COM) [12], affecting the body's dynamic balance a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 (XcoM) is an imaginary point combining the position and velocity of the COM based on an inverted pendulum model, which enables the definition of the region of dynamic stability [31,32]. While these methods are useful for studying factors influencing dynamic stability during level walking, such as walking speed and foot placement, they may not be suitable for activities or studies in which the assumptions and conditions involved do not hold. In contrast, the IA and RCIA are defined by the measured COM and COP motions without involving mathematical models and their associated assumptions that are not met in the current study. Therefore, the IA and RCIA variables are considered more appropriate for the current study.
Patients with AIS have been shown to change the motions of the trunk and the pelvis-leg apparatus during obstacle-crossing as compared to level walking [33]. These changes are expected to affect the balance control of the body and increase the neuromechanical demand on the locomotor system. However, to the best knowledge of the authors, no study has investigated the effects of AIS spinal deformity on balance control in terms of COM-COP motions during obstacle-crossing. Therefore, the purpose of this study was to identify the influence of spinal deformity on the whole-body balance control during obstacle-crossing in terms of COM-COP motions in patients with Lenke 1 thoracic AIS. It was hypothesized that patients with AIS would cross obstacles with altered COM-COP control when compared to healthy controls, and that the changes would be greater when crossing with the limb on the concave side leading than when crossing with the convex side leading.

Methods
This study was approved by the Research Ethics Committee of National Taiwan University Hospital (Permit Number: 201306013RINB). All methods were carried out in accordance with relevant guidelines and regulations. Fourteen adolescents with right thoracic AIS (AIS group; age: 14.0±1.8 years, height: 154.8±4.7 cm, mass: 42.0±7.5 kg) were recruited from the university hospital between July 2013 and May 2016. All participants gave written informed consents signed by both them and their legal guardians. All the patients were determined radiographically to have a Lenke 1 thoracic curve [34] with Cobb angles of 59.9±18.9˚and kyphosis angles of 28.2±8.8˚. The patients had normal lower limb muscle strength, with corrected vision and right leg dominance without limitations in performing daily or sports activities. The only treatment was wearing a brace during the daytime. All indicated studies were performed without the brace. Participants were excluded if they had leg length discrepancies greater than 1 cm as determined by scanogram, or other musculoskeletal diseases that would affect their gait performance. Fourteen healthy adolescents (Control group; age: 14.4±2.0 years, height: 158.4±6.2 cm, body mass: 48.6±8.9 kg) were selected to match with the AIS group for sex, age and BMI. An a priori power analysis based on pilot results of sagittal and frontal IA and RCIA from four patients with AIS and four healthy subjects using GPOWER 3 [35] determined that a projected sample size of ten subjects for each group would be enough for a power of 0.8 and an effect size of 0.9 at a significance level of 0.05. Thus, 14 subjects for each, namely the AIS and the control group, were adequate for achieving the main objectives of this observational, cross-sectional study.
In a university hospital gait laboratory, each subject walked at a self-selected pace on a 10-meter walkway and crossed an obstacle composed of a 1.5 m long tube with a diameter of 1.5 cm placed horizontally across a height-adjustable metal frame [36]. Three obstacle heights (i.e., 10%, 20% and 30% of leg length) were tested with each leg leading while the motions of the body segments were monitored by an 8-camera motion analysis system (Vicon MX T-40) at 200 Hz using 39 infrared-retroreflective skin markers [37]. The ground reaction forces were also measured simultaneously at 2000 Hz using two forceplates (OR6-7, AMTI) placed on either side of the obstacle in the middle of the walkway. With the current forceplate setup, the crossing cycle was defined from the toe-off of the leading limb before crossing to the toe-off of the leading limb after crossing. The ground reaction force data were used to determine the toe-off of the leading limb, heel-strike of the leading limb, and toe-off of the contralateral limb. The marker data were used to determine the instants when the leading toe was above the obstacle, and when the trailing toe was above the obstacle as well as the heel-strike of the contralateral limb. Data for three complete crossing cycles for each lower limb leading were obtained for each obstacle condition. For the AIS group, the crossing cycle with the limb on the convex side leading was denoted AIS-V and the cycle with the limb on the concave side leading was denoted AIS-A. For the crossing limb conditions (AIS-V vs. AIS-A) a counterbalanced measures design was used, while the sequence of the obstacle conditions was decided by a random number table.
The body's COM position was calculated as the weighted sum of the positions of the COMs of all the body segments using the marker data and segmental inertial properties. Subject-specific body segmental inertial properties were obtained using an optimization-based method, which has been shown to reduce errors in the calculated center of mass motions when compared to commonly used prediction methods [38]. The COP position was calculated using forces and moments measured by the forceplates. The COM-COP inclination angles (IA) in the sagittal and frontal planes were then calculated as follows: Frontal IA ¼ À sin À 1 ðt X Þ; for the right limb sin À 1 ðt X Þ; for the left limb ð3Þ ( where P ⃑ COMÀ COP was the vector pointing from the COP to the COM, Z ⃑ was the unit vector of the vertical and X ⃑ was the unit vector pointing in the direction of progression. With the current forceplate setup, the IAs were calculated from the beginning of swing phase of the leading limb to the subsequent contralateral heel-strike. The RCIA were calculated by smoothing and differentiating the IA trajectories using the GCVSPL method [39]. For the leading limb, positive sagittal and frontal IA indicate that the COM is anterior to and away from the COP toward the contralateral limb, respectively. The greater the IA, the greater the COM-COP vector deviates from the vertical and the greater the effort needed to bring the COM back to be above the COP unless accompanied by an appropriate RCIA to reduce IA, corresponding to the position and velocity control of the COM described by Pai et al. [23]. To quantify the smoothness of the balance control, the jerk index of the COM-COP motion was calculated for contralateral single-limb support (SLS) and double-limb support (DLS) during crossing using the third derivatives of the IA trajectories as follows [40][41][42][43].
where t i and t f are, respectively, the beginning and end of the sub-phase considered.
Crossing speed was calculated as the ratio of the distance traveled by mid-ASISs in the walking direction and the time spent from leading toe-off immediately before crossing to trailing heel-strike immediately after crossing. Toe-clearance for both the leading and trailing limb was calculated as the vertical distance between the toe marker of the swing limb and the obstacle when the swing toe was directly above the obstacle. The trailing toe-obstacle horizontal distance was defined as the horizontal distance between the obstacle and the toe marker of the trailing limb during stance immediately before stepping over the obstacle. The leading heelobstacle horizontal distance was defined as the horizontal distance between the obstacle and the heel marker of the leading limb during stance immediately after stepping over the obstacle. The leading toe-obstacle horizontal distance was defined as the horizontal distance between the obstacle and the toe marker of the leading limb during stance immediately before stepping over the obstacle [24].
For statistical analysis, the values of the IA and RCIA when the leading and trailing toes were directly above the obstacle were obtained. The range of motion and mean of IA, and the peak RCIA during DLS as well as the mean of IA during SLS of the trailing limb were also obtained. For each calculated variable, data from three trials were averaged for AIS-V and for AIS-A, while those from both sides (i.e., six trials) were averaged for Control. All calculated variables were determined to be normally distributed by a Shapiro-Wilk test and the homogeneity of variance across groups was confirmed by the Levene's test. For between-group comparisons, a two-way mixed-design analysis of variance (ANOVA) was used to compare the differences in all the calculated variables with one between-subject factor (i.e., AIS-V vs. Control, and AIS-A vs. Control) and one within-subject factor (obstacle height). For between-side comparisons in AIS, a two-way repeated-measures ANOVA was used to compare the differences with one within-subject factor (AIS-V vs. AIS-A) and another separate within-subject factor (obstacle height). Whenever an obstacle height effect was found, a post hoc trend analysis was performed to determine the trend of the variable. A significance level of α = 0.05 was set for all test conditions. All statistical analyses were performed using SPSS version 20 (SPSS Inc., Chicago, IL, USA).

Results
Compared to the healthy controls, no significant differences were found in the crossing speed ( Table 1). The AIS group showed significantly increased leading heel-obstacle horizontal distance for all obstacle heights, both when crossing with the limb on the concave side leading (AIS-A) and when crossing with the convex side leading (AIS-V), while the trailing toe-obstacle horizontal distances were not significantly different between groups. Double-limb support (DLS) time for both AIS-V and AIS-A was significantly decreased but only AIS-V showed significantly increased leading swing phase time when compared to the control group.
In the sagittal plane, averaged IA values over the leading swing phase and the ranges of IA during DLS were increased when the limb on the convex side was leading (Fig 1 and Table 2). Compared to Control, no significant differences were found in any of the IA-related variables during AIS-A, except for the increased averaged values over the DLS. Both AIS-V and AIS-A showed significantly increased anterior RCIA when the leading toe was above the obstacle ( Fig  2) but only AIS-V showed increased anterior RCIA when the trailing toe was above the obstacle. Both AIS-V and AIS-A showed significantly decreased mean and peak values of RCIA over DLS (Table 3). When compared to Control, both AIS-V and AIS-A showed an increased jerk index of IA over DLS and CSLS (Table 4).
In the frontal plane the AIS-A showed a significantly increased IA magnitude compared to Control while the AIS-V showed unaltered IA when the leading toe was above the obstacle (Figs 2 and 3), with RCIA magnitudes very close to zero for both AIS-A and AIS-V (Table 2). Over the leading swing phase, the AIS-A showed significantly increased IA magnitudes compared to AIS-V. During DLS, while the AIS-V maintained an unaltered mean IA value, the AIS-A significantly altered the IA control to show a negative mean IA when compared to both Control and AIS-V (Table 2). Both AIS-V and AIS-A also showed significantly increased peak values of RCIA but only AIS-V showed significantly increased mean values of RCIA over DLS when compared to Control (Table 3). Over the DLS and CSLS, the AIS-V showed an increased jerk index when compared to Control (Table 4).
With increasing obstacle height, crossing speeds were decreased linearly, while the trailing toe-obstacle distance and leading heel-obstacle distance were found to increase linearly for both groups (Table 1). Linearly decreased DLS time but linearly increased leading swing phase time were found in both groups. In the sagittal plane, both groups linearly decreased anterior IA when the leading toe was above the obstacle (Table 2). When the trailing toe was above the obstacle, increased anterior IA but decreased anterior RCIA were found in both groups (Table 3). Averaged values of the anterior RCIA over the leading swing phase decreased but the peak RCIA was increased with increasing obstacle height (Table 3). No significant obstacle height effects were found in the frontal plane (Tables 2 and 3).

Discussion
The current study aimed to identify the effects of thoracic AIS on the whole-body balance control during obstacle-crossing in terms of COM-COP IA and RCIA. The results support the hypothesis that, when compared to healthy controls, the patients with AIS crossed obstacles with altered, compromised balance control.
In the sagittal plane, the patients crossed the obstacle with unaltered IA but increased anterior RCIA when crossing, whether with the concave-side or the convex-side leading. Since the COM was ahead of the stance limb when the leading limb was crossing, increased anterior RCIA indicated that the COM was moving away from the stance limb at a faster speed, increasing the difficulty in maintaining a smooth COM motion relative to the stance limb, as indicated by the increased jerk index of IA. Increased anterior RCIA also appeared to contribute to the increased horizontal heel-obstacle distance, requiring greater braking force to slow down the anterior motion of the COM, which appeared to lead to an increased posterior RCIA during DLS for both AIS-A and AIS-V. Rapid changes in RCIA suggest challenges for a smooth transition between swing and stance, as well as for weight transfer during DLS (Table 4). In the frontal plane, compared to healthy controls the patients appeared to have a compromised balance control with increased IA magnitude but reduced RCIA when the swing toe of the concave-side leading-limb was above the obstacle during crossing, suggesting an increased risk of loss of balance during crossing. Previous studies have shown that the M/L COM stability is an important parameter for distinguishing older people with imbalance, because they Table 2  showed greater M/L inclination angles [22,44]. Increased frontal IA, i.e., increased medio-lateral COM-COP separation, increases the difficulty in controlling a smooth and stable COM motion relative to the COP, which was found to lead to loss of balance in elderly fallers, especially under unexpected postural disturbances such as tripping or slipping [22]. For leading with the limb on the convex side, the increased bending of the trunk towards the limb on the concave side [33] may help minimize the separation of the COM and COP, leading to a COM-COP inclination angle similar to that of the healthy controls. These results suggest that the whole-body balance control during obstacle-crossing in patients with Lenke 1 AIS should be monitored for signs of increased risk of loss of balance in the management of such a patient group, especially when crossing obstacles with the limb on the concave side leading. During the DLS phase after the leading limb had crossed the obstacle, the COM was controlled within a relatively small range while the COP traveled from the trailing to the leading limb during the body weight transfer. Therefore, the smooth motion of the COM relative to the COP in terms of a smooth change in RCIA, and normal or reduced jerk index of IA is an indication of a well-controlled transfer of the body weight while maintaining dynamic balance. In the sagittal plane, both AIS-A and AIS-V increased the peak and mean posterior RCIA, with an increased jerk index of IA during DLS, suggesting a less well-controlled COM-COP motion and an increased effort in transferring the body weight to the leading limb. In the frontal plane during the DLS of AIS-A, the patients appeared to alter the IA control significantly, with prolonged duration of the weight release phase, i.e., the body weight was kept on the trailing limb for longer before the COP traveled to be right below the COM (i.e., zero IA), leading to a negative mean IA during DLS as compared to Control. These results suggest that patients Table 3. Means (standard deviations, SD) of the sagittal and frontal rates of change of COM-COP inclination angles (˚/s) in the patients with adolescent idiopathic scoliosis and healthy controls when crossing obstacles of heights of 10%, 20% and 30% of subjects' leg length. with AIS had a reduced ability in maintaining a smooth and stable transfer of the body weight and COM-COP control from the trailing to the leading limb during obstacle-crossing when leading with the concave-side limb. The vast changes of RCIA around the instances of transfer from swing to DLS, and from DLS to swing, particularly shortly before the beginning and after the end of the DLS when the largest IA magnitudes occurred, are also an indication of an increased difficulty in COM-COP control and thus an increased risk of loss of balance for AIS.

Study limitations
The current study was the first attempt to identify the effects of AIS on the control of the body's COM motion relative to the COP in terms of IA and RCIA during obstacle-crossing. The patient group was limited to Lenke 1 thoracic scoliosis without compensatory thoracolumbar curves for better homogeneity of the patient group. Nonetheless, this type of spinal deformity, and thus the trunk shape in AIS, has been shown to affect the motion of the trunk during activities [8,45,46], and presumably the COM-COP control. Therefore, further studies are needed to identify how the spinal deformity type and the severity of AIS would affect the COM-COP control during walking and obstacle-crossing. Another factor that may limit the patient's ability in controlling the trunk and COM-COP motion is muscle strength imbalance [4,45,47], which also deserves further investigation.

Conclusions
The patients with Lenke 1 thoracic AIS were found to cross obstacles with altered, compromised COM-COP control in both sagittal and frontal planes when compared to healthy controls. The patients increased anterior RCIA with an increased jerk index of IA during crossing with either the limb on the concave side or convex side leading, indicating their difficulty in maintaining a smooth COM-COP motion in the sagittal plane. In the frontal plane, the  patients adopted a conservative COM-COP control strategy with close-to-zero RCIA during the leading limb crossing. However, their balance control appeared to be compromised with an increased IA magnitude when crossing with the concave-side limb leading. The current results suggest that the thoracic spinal deformity in Lenke 1 AIS affected the whole-body balance control during obstacle-crossing, especially when the concave-side was leading. Thus, balance control should be monitored for signs of increased risk of loss of balance in the management of such patient groups.