Obstacle Crossing During Gait in Children With Cerebral Palsy: Cross-Sectional Study With Kinematic Analysis of Dynamic Balance and Trunk Control

Discussion

Children with CP successfully cleared a 10-cm obstacle in stride during walking at their self-selected speed; however, their strategy for achieving clearance differed from that of children with TD. The differences between the groups reflected the challenges faced by children with CP in devising an efficient motor strategy to clear the obstacle. They were capable of achieving peak hip and knee flexion similar to that of their peers with TD, but their strategy was compromised by the addition of cross-planar compensatory movements at the trunk, pelvis, and hip and excessively high foot clearance of the lead limb. The findings indicated both an impairment of trunk and pelvic control during the challenging single-limb support phase and a lack of selectivity and efficiency of movement of the swinging limb itself. The swinging limb was forced to use a movement amplitude larger than that used in normal gait to avoid hitting the obstacle; this larger amplitude imposed additional stability challenges on the stance limb.

Children with CP showed an inefficient lead limb strategy, characterized by significantly higher foot clearance over the obstacle. This excessive height might have resulted from a lack of selectivity that prevented children with CP from fine-tuning the most efficient multijoint pattern to ensure clearance. Difficulties with visual perception might also have contributed to this feature. A recent systematic review found a 40% to 50% prevalence of visual-perceptual impairment in children with CP across all GMFCS levels.²² Visual-perceptual ability might influence motor planning around activities of daily living in people with CP,²³ and obstacle crossing could be impeded in the presence of a significant deficit. None of the children in our cohort had undergone tests of visual-perceptual function, so we cannot confirm this association; however, this topic is important for future research because it might have been a significant contributor to the deficits that we found.

Step length and single-support duration for both lead and trail limbs did not differ between the groups, unlike previous reports for these variables over level ground,²⁴ but this finding was not surprising. It suggested that the obstacle task forced minimum step length and single-support duration in order to clear the obstacle successfully and avoid falling. It has been noted that for adults with CP, the primary concern in obstacle clearance is safety; in contrast, unobstructed walking, which is associated with a lower risk of falls, is influenced more by minimizing energy cost.²⁵ Therefore, high clearance was prioritized over speed and efficiency.

These movement patterns reflected impaired control of the trunk in the unstable position of single-limb support while the contralateral limb cleared the obstacle. The coupling of movement between the pelvis and the trunk is an important part of the motor strategy for obstacle clearance. Children with CP often show a proximal-to-distal activation strategy, and their ankle muscles—critical for fine-tuning balance responses—are slower to activate than those of children with TD.³ This strategy has a significant impact on the trunk, yet deficits in trunk control are often underestimated. Recent research has shown that segmental trunk control has a strong association with gross motor function and mobility, with up to 40% of the variation in Gross Motor Function Measure scores being predicted by the results of a test of trunk control on linear regression.¹³ Furthermore, children with CP show excessive trunk movement during normal gait without an additional balance challenge.²⁶ These features present a dual source of difficulty for children with CP when dealing with dynamically challenging tasks, such as obstacle crossing: they not only lack responsiveness and control distally at the contact point with the foot but also show impaired proximal control of the trunk, which is forced to compensate further to maintain a stable CoM.

Our results support the consideration that the rehabilitation of dynamic balance in children with CP should focus on both distal and proximal control and stability. Exercises that could be considered include those targeting control by the hip and ankle during single-limb support and those focusing on the deep abdominal muscles through active alignment and trunk control in various postures, such as standing, 4-point kneeling, and stability ball sitting.

The results of the present study showed that children who have CP and GMFCS level I or II and are independently mobile can change their kinematics when crossing an obstacle to avoid a fall but that doing so challenges their trunk control and stability. This task may be a more sensitive indicator of dynamic balance; therefore, incorporating it into a gait analysis protocol for children who have CP but are active and ambulant may be of value. Three-dimensional measurement of the kinematics and control of the CoM during obstacle crossing also may be useful as an outcome measure for evaluating changes in dynamic balance and postural stability in such children.

In measuring the effectiveness of rehabilitation, detecting changes in gait kinematics through an analysis of level-ground walking can be difficult, as found by Taylor et al in a well-designed and adequately powered randomized controlled trial of progressive resistance exercise²⁷; however, in that case, the intervention was not specifically designed to target gait itself.²⁸ A study of adults who had a stroke showed improvements in some aspects of obstacle crossing after 1 month of rehabilitation.⁶ The responsiveness to change of this task has not been assessed in people with CP, but our results suggest that this task shows promise in this regard, particularly if the rehabilitation intervention specifically targets dynamic trunk and postural control.

The methodological requirements of the present study imposed some limitations on its design and interpretation. The 10-cm fixed height of the obstacle required greater adaptations for clearance by children of shorter stature. To counteract this possible confounding variable, we matched the groups in age and height to ensure that there were no differences; any effect of height would have been evenly spread across the 2 groups. The 10-cm obstacle was an average of 7.3% of the participants' average height, with a standard deviation of 0.9% and an absolute range of ±1.8%. We considered such a low percentage variation unlikely to cause significant kinematic deviations. Furthermore, obstacles in the everyday environment are of a fixed size, so our approach provided a more realistic scenario than scaling the obstacle to individual participant height. With regard to anthropometrics, the nonsignificant, 4.2-kg mean difference in weight between the groups was unlikely to be important clinically because we did not report segmental kinetic data. Mean body mass indexes were similar between the groups (16.9 kg/m² for children with CP and 17.7 kg/m² for children with TD).

Both hemiplegia and diplegia were included in the present study because we considered it to be important to include all children who had spastic CP but were active and ambulant, regardless of motor type. Obstacle crossing is a task that requires whole-body coordination, and previous studies did not distinguish between hemiplegia and diplegia in the interpretation of trunk control because deficits in trunk control are common to both conditions¹³; however, it is true that children with diplegia generally exhibit a greater range of trunk movement during walking.²⁶ The relatively equal distributions of diplegia and hemiplegia would have had a balancing effect. Furthermore, the present study was not powered at the outset to determine differences between the 2 motor types.

The inclusion of 3 children with GMFCS level II merits further discussion. When we set the objective of the present study a priori, our goal was to capture the most complete cohort of children who had CP but were active and independently ambulant in the community and likely to be participating in activities with their peers with TD in the everyday environment. All children who were recruited met this broad description, and all were able to cross a 10-cm obstacle unassisted during gait. Because this level of ability crosses into the milder presentations of GMFCS level II, we did not want to exclude children with GMFCS level II because the results also would have applied to them and they also were representative of the population in question. Therefore, we consider that our cohort of participants reflected a spectrum of ability related to community ambulation rather than a categorical distinction between GMFCS levels. This notion is reflected in the normal statistical distribution of the data.

The age range of the participants in the present study was large—5 to 18 years, with a mean of 10 years. We acknowledge that this range increased variability in the data; however, we included a large age range because we aimed to capture a cohort of children representative of those who are independently ambulant and active alongside their peers in the community but who experience balance difficulties. Had we excluded a range of ages to reduce variability, the results then could not be generalized to the excluded age groups, thereby affecting the relevance of the findings to the wider population of interest. The lower limit of 5 years was pragmatically based on the age at which children could adhere to the laboratory protocol. We acknowledge that balance would have been better developed in the older participants; however, the effect would have been minimized by the inclusion of a cohort of children who had TD and were carefully age matched for comparison.

In conclusion, children with CP required greater adjustments at the trunk and pelvis to achieve successful obstacle crossing. This finding reflected both impaired trunk control as a primary problem and compensatory strategies for reduced stability distally, particularly because obstacle crossing enforces longer single-support duration. Visual-perceptual difficulties and lack of selective motor control might have contributed to the differences in strategy.

The results of the present study provide a baseline for planning rehabilitation strategies to target deficits in dynamic balance during walking. Analysis of obstacle crossing might be more sensitive to changes in postural control than level-ground walking, which does not challenge stability as effectively. It could be used as a test in parallel with other clinical measures of balance to plan interventions and assess changes, particularly in trunk and CoM control.