Evaluation of Knee Position Sense in Children with Motor Disabilities and Children with Typical Development: A Cross-Sectional Study

Background: In children with motor disabilities, knee position during walking is often of concern in rehabilitation. This study aimed to investigate knee joint position sense. Thirty-seven children with Cerebral Palsy (CP), 21 with Myelomeningocele (MMC), 19 with Arthrogryposis (AMC), and 42 TD children participated in the study. Knee joint position sense, i.e., the difference between the criterion angle and the reproduced angle (JPS-error), was assessed in sitting while 3D motion capture was recorded at flexed knee 70 (Knee70), 45 (Knee45), and 20 (Knee20) degrees, and after three seconds at maintained criterion angle (CAM) and maintained reproduced angle (RAM). No differences were found between the groups in JPS-error, CAM, and RAM. At Knee70, CAM differed between the right and left legs in the TD group (p = 0.014) and RAM in the MMC group (p = 0.021). In the CP group, CAM was greater than RAM at Knee70 in the left leg (p = 0.002), at Knee45 in both legs (p = 0.004, p = 0.025), and at Knee20 in the right leg (p = 0.038). Difficulties in maintaining the knee position at CAM in the CP group sheds light on the need for complementary judgments of limb proprioception in space to explore the potential influence on knee position during walking.


Introduction
In children with motor disabilities, altered knee position during walking is often of concern in rehabilitation. The cause of the observed difficulties varies with respect to the nature of the motor disability. In some children, knee flexion during walking is encountered, whereas, in others, knee hyperextension is observed.
Children with bilateral spastic Cerebral Palsy (CP), Myelomeningocele (MMC), and Arthrogryposis Multiplex Congenita (AMC) are often referred to gait laboratories for evaluation of walking function.
CP is a descriptive diagnosis including a group of movement and posture disorders attributed to an injury in the developing brain. Apart from motor disorders, disturbances with sensation, perception, cognition, communication, behavior, and epilepsy are common [1]. The current overall CP birth prevalence for high-income countries is 1.6 per 1000 live births [2]. The motor impairments, with reduced motor repertoire, hypertonia, and progressive muscle changes related to neuronal-, nutritional-and/or mechanical factors, may lead to musculoskeletal deformities [3]. In ambulating children with spastic cerebral palsy, crouch gait, characterized by increased knee flexion throughout stance, is a common gait pattern that is very tiring to sustain and, in some cases, accompanied by rushing forward [3]. Dropping into increasing ankle dorsiflexion and knee and hip flexion soon after rising to stand has been interpreted as an exhaustibility of the support reaction of the legs in children with spastic CP [4]. development and function, likely due to the loss of specific proteins in afferent neurons, leading to disturbed proprioception [22] with an unstable wide-based gait and coordination defects [23]. Since in MMC, both motor and sensory deficits are present [6], it seems plausible that gait pattern [24] is influenced by sensory dysfunction as well. In infants with MMC at a lumbosacral level, investigations have demonstrated that at neurological levels that affect major locomotor muscles, the infants find ways to respond and adapt their motor output to changes in sensory input after treadmill stepping [25].
In children with CP, musculoskeletal deformities in the lower limbs are commonly addressed by both skeletal and soft-tissue surgery [26]. In the orthopedic field of MMC, common procedures are the treatment of knee flexion contractures in ambulatory patients [27]. In the management of lower limb deformities in children with AMC, orthopedic surgery is performed to improve functional outcomes with early management in intensive physiotherapy and bracing [28]. In typically developing children and adolescents, the threshold to detect passive motion was increased after a knee ligament injury contributing to proprioceptive deficits [29]. In view of this, it is assumed that children with motor disabilities with associated joint deformities [10], muscle paresis [5], or disturbed muscle morphology [30] may have reduced proprioception in the knee. Our hypothesis was that children with central nervous system involvement, such as CP, and children with predominantly joint deformities (AMC) or peripheral lesions (MMC) would present with reduced knee proprioception compared to children with Typical Development (TD).

Participants
Children with motor disabilities who are commonly referred to the assessment of gait and motor functioning in the neuropediatric department of Karolinska University Hospital, Stockholm, Sweden, were invited to participate in a cross-sectional study between January 2019 and December 2020. Seventy-seven participants between the ages of 5 and 18 years were grouped according to their diagnosis: 37 with bilateral spastic CP, 21 with MMC, and 19 with AMC. All participants had an ambulatory function; however, two participants with CP walked only for transfer in a closed environment, and one participant with MMC had ceased walking as a teenager. Ambulation according to the various diagnosis was documented with the Gross Motor Function Classification System (GMFCS E&R) [31] in children with CP, and according to Hoffer in the MMC [32] and AMC [33] groups. Performed orthopedic surgery was documented in the medical chart. Participants had to meet the following criteria: knee extension muscle strength "good" (Grade 4) with the ability to withstand considerable manual resistance when being asked to hold the leg not allowing the examiner to "break" the hold or "normal" (Grade 5) with ability to maintain end-point range against maximal resistance [34]; passive knee motion available to perform the test; and no orthopedic surgery in the last year. In addition, the ability to understand verbal instructions necessary to accomplish the test was required. Forty-two individuals with TD constituted a control group. The TD children were recruited among siblings of patients, through advertisements at the hospital, and among acquaintances of the researchers.
The study was approved by the Regional Ethical Review Board in Stockholm, code nr 2017/953-31/2. Before taking part in the study, written informed consent was given by the caregivers, and a verbal agreement was confirmed by the children.

Knee Joint Position Sense
Knee joint position sense was measured in a sitting position on a customized stool with a 60 × 60 cm area seat and a height of 70 cm. A wedge was placed under the distal part of the thigh to ensure a hip flexion angle of 90 degrees and to adjust for a knee flexion angle of 90 degrees. The legs hung freely to eliminate any external cues. The participants were examined barefoot and dressed in shorts. Examinations were performed in both legs at three knee flexion angles, namely 70 (Knee70), 45 (Knee45), and 20 (Knee20) degrees [35]. A height-adjustable tray table was arranged in front of the participant to prevent visual control of leg movements during testing ( Figure 1). By gripping a molded strap with a soft lining to minimize cutaneous sensation, placed distally at one-third of the participant's shank length, the examiner passively elevated the participant's leg to a Criterion Angle (CA) (Figure 2a). To set the CA, the examiner was guided by a red laser pointer which was placed on the stool attached to a goniometer ( Figure 2b). Next, the participant was asked to actively maintain and memorize the CA position for three seconds. After a resting period of approximately 5 s, the participant was asked to reproduce the leg position at the CA by actively lifting the leg to the perceived position, giving the Reproduced Angle (RA), and maintaining that position for 3 s before returning their leg to its resting position ( Figure 2c). The trials started with a knee flexion angle of 70 degrees, followed by 45 and then 20 degrees. The left and right legs were tested in randomized order. Prior to the test, thorough explanations were given to the participant, and familiarization trials were undertaken to ensure the participant's understanding of the test procedure. There were two trials for each angle, with a short break of approximately ten seconds between trials. Some participants had difficulty actively maintaining the knee position for three seconds; additional measurements were therefore made to assess the ability to maintain leg positions at the CA (Criterion Angle Maintained, CAM) and at the RA (Reproduced Angle Maintained, RAM). angle of 90 degrees. The legs hung freely to eliminate any external cues. The participants were examined barefoot and dressed in shorts. Examinations were performed in both legs at three knee flexion angles, namely 70 (Knee70), 45 (Knee45), and 20 (Knee20) degrees [35]. A height-adjustable tray table was arranged in front of the participant to prevent visual control of leg movements during testing ( Figure 1). By gripping a molded strap with a soft lining to minimize cutaneous sensation, placed distally at one-third of the participant's shank length, the examiner passively elevated the participant's leg to a Criterion Angle (CA) (Figure 2a). To set the CA, the examiner was guided by a red laser pointer which was placed on the stool attached to a goniometer ( Figure 2b). Next, the participant was asked to actively maintain and memorize the CA position for three seconds. After a resting period of approximately 5 s, the participant was asked to reproduce the leg position at the CA by actively lifting the leg to the perceived position, giving the Reproduced Angle (RA), and maintaining that position for 3 s before returning their leg to its resting position ( Figure 2c). The trials started with a knee flexion angle of 70 degrees, followed by 45 and then 20 degrees. The left and right legs were tested in randomized order. Prior to the test, thorough explanations were given to the participant, and familiarization trials were undertaken to ensure the participant's understanding of the test procedure. There were two trials for each angle, with a short break of approximately ten seconds between trials. Some participants had difficulty actively maintaining the knee position for three seconds; additional measurements were therefore made to assess the ability to maintain leg positions at the CA (Criterion Angle Maintained, CAM) and at the RA (Reproduced Angle Maintained, RAM).  The JPS test was performed at the Motion Analysis Laboratory at Karolinska University Hospital on one occasion. Four examiners, specialized in each of the three patient categories, were involved in data collection.

Motion Analysis
The knee angles were recorded with 3D motion analysis using a 12-camera system (Vicon MX40 ® Oxford, UK) with a sampling rate of 100 Hz. The stool was placed in the middle of the laboratory room so that all markers (attached in sitting position) were visible during data capture, with the participant's hands placed on the sides of the stool to avoid obscuring the thigh markers. A kinematic lower body model consisting of 16 reflective markers, of which eight were on each side, were attached on the anterior superior iliac spine, posterior superior iliac spine, thigh, femur condyle, shank, lateral malleolus, Children 2023, 10, 1056 5 of 15 heel, and forefoot, according to the Newington model [36]. Video recordings were made simultaneously. The JPS test was performed at the Motion Analysis Laboratory at Karolinska University Hospital on one occasion. Four examiners, specialized in each of the three patient categories, were involved in data collection.

Motion Analysis
The knee angles were recorded with 3D motion analysis using a 12-camera system (Vicon MX40 ® Oxford, UK) with a sampling rate of 100 Hz. The stool was placed in the middle of the laboratory room so that all markers (attached in sitting position) were visible

Data Analysis
Vicon Nexus was used to process the trials. Data were missing in the TD group: in one left trial at Knee70; in the CP group: in 11 left and right trials at Knee20; in the MMC group: in one left trial at Knee45 and three left and two right trials at Knee20; and in the Children 2023, 10, 1056 6 of 15 AMC group: two left and four right trials at Knee70, one right trial at Knee45, and three left and right trials at Knee20.
The ability to sense knee joint position was assessed using the absolute difference in degrees between the CA and RA (JPS-error) [37]. The knee position angle after three seconds relative to the CA (CA-3s) and the knee position angle after 3 s relative to the RA (RA-3s) provided an assessment of the ability to maintain the knee position by taking the difference in degrees between the CA and CA-3s (CAM) and between the RA and RA-3s (RAM) (Figure 3). and forefoot, according to the Newington model [36]. Video recordings were made simultaneously.

Data Analysis
Vicon Nexus was used to process the trials. Data were missing in the TD group: in one left trial at Knee70; in the CP group: in 11 left and right trials at Knee20; in the MMC group: in one left trial at Knee45 and three left and two right trials at Knee20; and in the AMC group: two left and four right trials at Knee70, one right trial at Knee45, and three left and right trials at Knee20.
The ability to sense knee joint position was assessed using the absolute difference in degrees between the CA and RA (JPS-error) [37]. The knee position angle after three seconds relative to the CA (CA-3s) and the knee position angle after 3 s relative to the RA (RA-3s) provided an assessment of the ability to maintain the knee position by taking the difference in degrees between the CA and CA-3s (CAM) and between the RA and RA-3s (RAM) (Figure 3).

Statistics
For each leg and each of the measured angles, the trial with the smallest difference between the CA and RA (that is, the best-performing trial) was chosen for analysis. Descriptive data are presented in terms of medians and min-max. Nonparametric methods were used due to the absence of normally distributed data. A Kruskal-Wallis test with a post-hoc comparison using the Mann-Whitney U test was used to analyze possible differences in age, weight, and height and performed orthopedic surgery between the TD, CP, MMC, and AMC groups. A Wilcoxon test was used to analyze possible differences between the left and right legs in JPS-error, CAM, and RAM and differences between the CAM and RAM according to each of the disability groups. To explore possible differences in JPS-error, CAM, and RAM between groups, a Kruskal-Wallis test was applied, and differences in JPS-error with respect to ambulatory function within each of the CP, MMC, and AMC groups. Statistics were computed using commercially available software (SPSS, version 26) at a significance level of 0.05.

Statistics
For each leg and each of the measured angles, the trial with the smallest difference between the CA and RA (that is, the best-performing trial) was chosen for analysis. Descriptive data are presented in terms of medians and min-max. Nonparametric methods were used due to the absence of normally distributed data. A Kruskal-Wallis test with a post-hoc comparison using the Mann-Whitney U test was used to analyze possible differences in age, weight, and height and performed orthopedic surgery between the TD, CP, MMC, and AMC groups. A Wilcoxon test was used to analyze possible differences between the left and right legs in JPS-error, CAM, and RAM and differences between the CAM and RAM according to each of the disability groups. To explore possible differences in JPS-error, CAM, and RAM between groups, a Kruskal-Wallis test was applied, and differences in JPS-error with respect to ambulatory function within each of the CP, MMC, and AMC groups. Statistics were computed using commercially available software (SPSS, version 26) at a significance level of 0.05.

Results
Participant characteristics, orthopedic surgery, and ambulatory function in participants with TD, CP, MMC, and AMC are shown in Table 1. The median ages for the CP and MMC groups were higher than the TD group (p = 0.008 and p = 0.016, respectively). There were no differences between the TD, CP, MMC, and AMC groups with respect to height, weight, or gender. Orthopedic surgery had been performed significantly more in participants in the MMC and AMC groups than in the CP group, in the ankle (p = 0.005 and <0.001 respectively) and in the knee (<0.001 and knee 0.006 respectively), and more in the hip in MMC than in the CP group (p = 0.037).

JPS-Error
The JPS error did not differ significantly in any leg between the TD, CP, MMC, and AMC groups at Knee70, Knee45, or Knee20 (Table 2). There were no differences in the JPS error between the left and right legs. No differences were found in JPS-error at Knee70, Knee45, or Knee20 with respect to the GMFCS levels in the CP group or to ambulation groups in the MMC and AMC groups. Table 2. Joint-position sense as the difference between the Criterion Angle (CA) and Reproduced Angle (RA) (JPS error), CA with respect to CA maintained for 3 s (CAM), and RA with respect to RA maintained for 3 s (RAM) presented as median absolute errors in participants With Typical Development (TD), Bilateral Spastic Cerebral Palsy (CP), Myelomeningocele (MMC), and Arthrogryposis Multiplex Congenita (AMC). "n" indicates performed trials within each group.

CAM
There were no significant differences in CAM between the TD, CP, MMC, and AMC groups at Knee70, Knee45, or Knee20 (

RAM
There were no statistically significant differences in maintaining any leg at RAM between the TD, CP, MMC, and AMC groups at Knee70, Knee45, or Knee20 (Table 2)

Discussion
To our knowledge, this is the first study examining knee joint position sense involving children with various motor disabilities who are commonly referred for assessments of gait and or motor functioning at a neuropediatric department. Despite the often observed altered knee position during walking in these children, our hypothesis that knee joint position sense was reduced in children with CP, AMC, and MMC compared to TD children could not be confirmed. When addressing gait in children with motor disabilities, it is common to determine a more or less dominant leg. Since the investigation was performed in the sitting position with the inclusion criteria being good to normal muscle strength [34] of the knee extensors, we did not consider a choice of a more involved side to be of importance in this study. Accordingly, when calculating the differences between the sides, no differences in JPS error were found, but they were present at Knee70 in CAM in the TD group and in RAM in the MMC group, values that were not considered clinically relevant. In the CP group, however, the highest JPS-error values with respect to the other disability groups were found but did not reach a significant level. Conversely, orthopedic surgery in the lower limbs, even in the knee, had been performed significantly less among participants in the CP group than in the MMC and AMC groups. This finding, therefore, does not further elucidate the possible influence of orthopedic surgery on the JPS results. Nor when discriminating for functional ambulatory level, differences in JPS error with respect to each of the diagnosis groups were found. Furthermore, in children with MMC with a paresis at a low neurological level, knee flexed position in an upright position is plausible due to insufficient calf muscle innervation. However, even though motor paresis is present in the biceps femoris muscle [8] at a low neurological level, this did not seem to have any impact on knee joint position sense in the unloaded sitting position. Nor did the oftenseen knee joint deformities in AMC [10] seem to play a determinant role in sensing knee position. Among the children with MMC, in one child with an asymmetric neurological lesion level, one leg was excluded due to insufficient quadriceps muscle strength, requiring a locked knee-ankle-foot orthosis when walking. All other children with MMC requiring orthoses were able to walk with non-restricted knee joints indicating at least good (grade 4) quadriceps strength [34]. There is important evidence that muscle spindles, being the principal kinaesthetic sensors, play a major role in kinaesthesia [13]. Since all children included in this study were able to activate their knee extensor muscles against manual resistance [34], one could speculate that muscle activation of the innervated knee extensors contributed sufficiently to the ability to sense the knee position in sitting, showing similar results in all groups. However, when grading muscle strength, it is challenging to minimize the subjective element by the examiner when performing the measurement [34].
The reason for assessing CAM and RAM derived from the experienced effort by some children to maintain the leg at the criterion angle set by the examiner versus when maintained at the self-initiated movement at the reproduced angle. The results confirm that there were discrepancies between the CAM and RAM in the CP group presenting with significantly larger CAM at four of the six tested angles, ranging between maximum values of approximately six degrees at Knee70 to eight to 17 degrees at Knee45 and 13 at Knee20, respectively. The difficulty in the CP group to maintain the knee joint extended for three seconds after a not self-initiated movement raises further questions about movement perception and motor planning in children with CP. However, a limitation of the study is that CAM and RAM have not been reported previously when assessing JPS and must be validated in larger groups. A further limitation of the study is the reduced sample size of children in the CP group that could perform the twenty-degree knee flexion trial (26 out of 37) due to contractures and/ or difficulties in selectively performing the task. Another weakness of the study is that in some (15 out of 19) of the children with AMC, the seventy-degree angle test could not be performed because of insufficient passive knee flexion. A limitation of the study is also the relatively small number of participants which precludes the generalization of our results.
According to Olsson et al. [16], it may be easiest for a participant to reproduce the position of the leg movement in sitting at a 70-degree knee angle since it is close to the start position of the 90-degree angle. In our study, with similar results at all tested angles, no variations with respect to the knee flexion angles of 70, 45, or 20 degrees could be confirmed. In the study by Olsson et al. [16], including healthy adults, participants were examined during five trials in each of the three test positions. The same cognitive attention span required could not be expected in children, particularly not within the CP group, since attentional disturbances may be present by definition [3]. In our study, the decision to perform two trials for each angle was based on the experiences from our pilot study, indicating short attention spans in some of the children. Experiences within the pilot study also led to the decision to minimize the time spent sitting on the equipment stool during preparation. The markers for kinematic data were therefore applied to the child's body in advance while the child was seated on a regular chair, allowing adjustments of marker placement to be made as close as possible to the start of the proprioception trials. The markers were applied by examiners who were trained in the gait laboratory, and the same technician performed all kinematic data collection. Since full attention to the interaction between examiner and child could not always be maintained throughout the trials, some measurements should be interpreted with caution.
Impaired joint position sense of lower extremities measured isolated in ankles and knees has been reported to correlate with balance function in children with developmental coordination disorder [38]. Some authors, however, recommend estimating body position during standing in children to provide information on joint positions of the ankle and from the loading of foot mechanoreceptors [39]. However, in many children with motor disabilities, muscle weakness, spasticity, or exhaustibility of the support reaction lead to insufficient stabilization of ankle joints, impeding taking valid measurements in the standing position. Providing an ankle-foot orthosis would ensure a stable ankle joint, but the impact on other body segments and their contribution to postural control would remain unclear. Although, since in the present study, knee JPS per se, as investigated in the sitting position, cannot be considered to primarily contribute to difficulties in controlling the knee joint during walking, studies on isolated ankle joint position sense may add further information. For such a study, however, factors of muscle strength, movement selection, and sensation should be taken into consideration.

Conclusions
By investigating knee joint position sense with 3D-motion analysis in a non-weightbearing position in children with sufficient knee extension muscle strength, we found no significant differences in JPS-error, CAM, or RAM between any of the groups with motor disabilities compared to children with TD, neither with respect to age nor ambulatory level. Lately, suggestions have been made to incorporate proprioceptive low-level tests detecting such as joint matching movements and proprioceptive high-level judgments indicating the location of limbs in space performed in a different frame of reference [40]. Most recently, proprioception was defined as the sense of where one's limbs are in space in a study identifying the role of somatosensory neurons for musculoskeletal development in children with distal AMC [41]. Accordingly, the discrepancy found between CAM and RAM in the CP group in the present study sheds light on the need for complementary assessments using joint position matching and detecting leg position in space [40]. This will benefit further exploration of the role of proprioception and its potential influence on knee position during walking with respect to various motor disabilities.  Data Availability Statement: The datasets generated during and analyzed during the current study are available from the corresponding author upon reasonable request.