Biofeedback to Improve Gait in Children with Cerebral Palsy

Hele tekst

(1)Adam Booth. Biofeedback to Improve Gait in Children with Cerebral Palsy.

(2) BIOFEEDBACK TO IMPROVE GAIT IN CHILDREN WITH CEREBRAL PALSY. Adam Thomas Crawford Booth.

(3) The printing of this thesis was financially supported by Motek Medical B.V. and Vicon. The work presented within this thesis was carried out as part of the PACE ITN. It was financially supported from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 642961. This PhD thesis was embedded within Amsterdam Movement Sciences research institute, at the Department of Rehabilitation Medicine, Amsterdam UMC, Vrije Universiteit Amsterdam, the Netherlands. Layout & Cover: Adam Booth Artwork, Chapter 10: Dorothy Crawford Printed by Proefschrift-AIO.nl ISBN: 978-94-92801-87-6 © 2019 Adam T.C. Booth.

(4) VRIJE UNIVERSITEIT. BIOFEEDBACK TO IMPROVE GAIT IN CHILDREN WITH CEREBRAL PALSY. ACADEMISCH PROEFSCHRIFT ter verkrijging van de graad Doctor of Philosophy aan de Vrije Universiteit Amsterdam, op gezag van de rector magnificus prof.dr. V. Subramaniam, in het openbaar te verdedigen ten overstaan van de promotiecommissie van de Faculteit der Geneeskunde op vrijdag 5 juli 2019 om 9.45 uur in de aula van de universiteit, De Boelelaan 1105. door Adam Thomas Crawford Booth geboren te Stirling, Verenigd Koninkrijk.

(5) promotor: . prof.dr.ir. J. Harlaar. copromotoren: . dr. M.M. van der Krogt. . dr. A.I. Buizer. . dr. F. Steenbrink.

(6) Contents Chapter 1. General introduction. 7. Chapter 2. How normal is normal? Consequences of not accounting for stride-to stride variability in paediatric reference gait data.. 23. Chapter 3. The efficacy of functional gait training in children and young adults with cerebral palsy: a systematic review and meta-analysis.. 43. Chapter 4. Real-time feedback to improve gait in children with cerebral palsy.. 75. Chapter 5. Immediate effects of immersive biofeedback on gait in children with cerebral palsy.. 89. Chapter 6. Muscle synergies in response to biofeedback-driven gait adaptations in children with cerebral palsy. 105. Chapter 7. The validity and feasibility of an eight marker model for avatarbased biofeedback gait training.. 123. Chapter 8. General discussion. 141. Summary. 162. Words of thanks. 166.

(7)

(8) Chapter 1. General introduction.

(9) Chapter 1. 1. U. pright, bipedal locomotion is a central trait that defines the hominid lineage and is distinct in humans 1. Millions of years of evolution have specialised the human form for stable and energy efficient walking. Whilst there is debate as to the precise evolutionary drive for this 2,3, it is clear that walking plays a significant role in everyday life, allowing for efficient movement from point to point and freeing the upper limbs for additional tasks. For many, walking is a little considered action. Nevertheless, it requires precise control of many muscles to generate the cyclic stepping pattern, while minimising energy cost and maintaining stability. The stepping motion is innate to new born infants, however, it takes many years of practice to learn, refine and optimise this movement 4. In individuals with cerebral palsy (CP), this process is impaired, resulting in limited walking ability and reduced quality of life. In this thesis, the potential of a novel use of biofeedback to improve walking ability in children with CP will be explored. CEREBRAL PALSY Children with CP account for the largest proportion of paediatric physical disability. Approximately 1 in every 500 live born children are diagnosed with CP 5. It is a static neurological condition, resulting from brain injury that occurs in the developing foetal or infant brain. While causes may involve trauma, intracranial haemorrhage and infection, a large proportion of prenatal cases result from unknown causes. Risk factors may include premature birth or low birth weight 6. Despite improvements in neonatal care, with greater survival of premature infants, the prevalence of CP has remained relatively consistent in recent decades 5. Genetic pathways may play a role in development of CP 7. This is an emerging area and while genetic contributions are not fully understood, it may provide insights into future treatment of the condition 8. With greater understanding of genetic pathway, children previously diagnosed as CP may have hereditary influence. Indeed, children with CP are often be grouped with patients with Hereditary Spastic Paraplegia. Clinical presentation is highly akin to CP, with the primary difference that hereditary conditions tend to be progressive, with gradual worsening over the years. CP refers to a spectrum of conditions with often variable manifestations and clinical presentations, due to the nature and location of the brain injury. A generally accepted definition of CP includes the following key elements; i) a group of disorders; ii) permanent but not unchanging; iii) involves a disorder of movement and/or posture and of motor function; iv) is due to a non-progressive interference/lesion/abnormality; v) this interference/lesion/abnormality arises in the developing/immature brain 9. The disorders in movement may be characterized further by neurological subtype; bilateral spastic, unilateral spastic, dyskinetic, or ataxic 10. Spastic CP is by far the most common, accounting for over 80% of the population 11. The understanding of spasticity in CP has shifted in recent years, however, it can be described as a ‘disorder of sensorimotor control, resulting from an upper motor neurone lesion, presenting as intermittent or sustained involuntary activation of muscles’ 12. Spasticity is a contributing factor to joint hyper-resistance. Clinical presentation of hyper-resistance at a joint includes both nonneural elements, such as muscle/tendon tissue stiffness and neural aspects that involve spasticity (hyperreflexia) and involuntary background activation, distinguished by speed of movement 13. Spasticity can be considered a primary impairment – i.e. impairment directly relating to the brain dysfunction. Primary impairments in CP also include impaired. 8.

(10) General introduction selective motor control and paresis. As a result of these primary impairments, secondary impairments may develop over time. Muscles may shorten, stiffen and atrophy, leading to deformation of the bones. While at present there is no cure for CP, the symptoms may be managed with correct treatment based on a structured, and patient specific treatment plan. Current therapeutic interventions in children with CP may take many forms 14. Interventions generally target the periphery, acting to reduce the presenting symptoms. The primary impairment of spasticity is commonly treated with the use of antispasticity medication (botulinum toxin injection or oral baclofen). For suitable children, surgical intervention (selective dorsal rhizotomy) may permanently reduce spasticity by sectioning of targeted rootlets in spinal cord 15. Further surgical intervention may include multi-level surgery, with orthopaedic correction of musculoskeletal problems 16 . These interventions are often supplemented with some form of physical therapy or strength training to target rehabilitation of motor performance. While significant improvement can be achieved with successful surgical interventions, outcomes can be mixed 17. Indeed it is often considered that children with CP have limited ability to adapt and improve motor function, reaching a plateau in ability 18.. 1. ASSESSMENT OF FUNCTIONING In treatment of disability, there is an impetus for change from impairment focused rehabilitation towards maximizing the functional ability of an individual. This is part of the International Classification of Functioning, disability and health (ICF) 19. The ICF framework is a widely implemented, standardised method for describing functioning ability in people with disabilities. In the ICF, functioning and disability is a multifaceted, dynamic model (Figure 1), relating to: • • •. Body functions and structures, i.e. impairments (functioning at the level of the body) Activities of individuals, i.e. limitations (functioning at the level of the individual) Participation or involvement of individuals in all areas of life, i.e. restriction (functioning of a person as a member of society). These factors are complex, interacting, and may be further influenced by environmental and personal factors (acting both as facilitators and barriers). In recent years, a version specific to children (ICF-CY), aged up to 18, was developed to include specific paediatric elements such as gait pattern functions (b770) and school life and related activities (d835) 20. The ICF classification system also provides a platform for standardised assessment of disability in children with CP. The use of classification in children with CP is important for patients and their families in setting goals and outlining treatment plans. CP is characterized by impaired motor function. As such, the majority of rehabilitation outcome measures are focused on this aspect. Individuals are often grouped according to the severity of impairment, known as the gross motor function classification system (GMFCS) 21 . The GMFCS outlines five levels of disability, with level increasing with impairment. In this thesis we are primarily focused on individuals who are able to walk independently without (GMFCS I, II), and to a lesser extent, with (GMFCS III) assistive devices. Children 9.

(11) Chapter 1. 1.

(12)

(13) .

(14).

(15)

(16) .

(17)

(18). .

(19) . . .

(20)

(21)

(22) .

(23) . !.

(24) . Figure 1. International Classification of Functioning, disability and health model. World Health Organization.. classified as GMFCS IV and V often rely on a wheelchair for mobility and have difficultly ambulating without body-weight support, which is considered in aspects of this thesis. While the GMFCS is a widely implemented and accepted functionality measure for children with CP, it is not sufficiently detailed to encompass all noteworthy gait-related deviations 22. Additional measures may provide more detailed insights. For example, broad measures of functioning such as the Gross Motor Function Measure (GMFM) 23, Paediatric Evaluation of Disability Inventory (PEDI) 24 and Child Health Questionnaire (CHQ) 25 are widely reported for evaluating outcomes in clinical trials 26. In contrast to these overall assessment schemes, assessment of CP routinely includes quantitative assessment of gait pattern by means of gait analysis to identify precisely how activity is limited. GAIT ANALYSIS Clinical gait analysis can be described as “the process of recording and interpreting biomechanical measurements of walking in order to support clinical decision-making in case of gait dysfunction” 27. Using motion capture, force plates, electromyography (EMG) and biomechanical models, detailed information regarding the action of muscles, position of joints and forces acting on the body can be calculated reliably and repeatably 28 . Gait analysis outputs a vast volume of data and requires extensive clinical experience and training to extract and interpret salient information. Data is often visualised as waveforms such as joint angle across the gait cycle, from foot contact to foot contact. In this method, a patient’s data may be compared to reference data sets to identify and classify deviations in joint patterns that are present 29–31. Identifying quantifiable impairments in gait is then linked to the underlying primary impairment 27. As such, 10.

(25) General introduction gait analysis is a valuable tool for objectively assessing the impairments that limit an individual’s gait, monitoring progression and guiding treatment decisions 32.. 1. Gait analysis has traditionally been performed overground in a dedicated gait laboratory. While this is considered common practice, and the de facto gold standard, this approach may have a number of limitations. Principally this relates to the collection of a limited number of steps used in analysis. Patients must walk up and down a short walkway, with only steps in which the foot hits a small force plate able to be used in full analysis. As such, the limited number of steps measured may not reflect typical walking of the individual in their daily life. An alternative approach may be that of instrumented treadmill, incorporating motion capture systems. Subtle differences in gait may be observed between treadmill and over ground walking. Treadmill walking tends to result in slightly higher cadence, shorter stance time and reduced preferred walking speed when compared with overground walking 33,34. Small differences in kinematics, kinetics, and EMG are also reported 33–36. Recent developments in technology may act to mitigate the potential differences in overground and treadmill walking. The Gait Real-time Analysis Interactive Lab (GRAIL, Motek Medical, Amsterdam, the Netherlands) integrates instrumented treadmill, motion capture and an immersive virtual reality (VR) environment to allow for congruent optical flow with walking speed (Figure 2). When investigated, children perceived walking on a treadmill with VR as more similar to overground walking than without VR 37. Another difference in treadmill walking arises from the natural variation in speed associated with the unrestricted nature of overground walking. In treadmill walking, this can be partially recreated by means of real-time positional analysis. This is termed self-paced walking, in which the subject controls his/her own speed by positioning on the treadmill belt. The system acts to maintain their position in the centre of the treadmill by accelerating and decelerating as the individual speeds up towards the front, or slows down towards the back of the belt 38,39 . To summarize, differences between treadmill and overground walking are generally small but should be taken into consideration in reporting of clinical gait analysis. As such, the subtle differences in using a treadmill may be outweighed by the potential added value of VR 37,38,40. A further advantage of using such a system is the potential for real-time analysis, allowing for interaction between the user and the VR environment 41 (D-Flow, Motek Medical, Amsterdam, the Netherlands). In the simplest form, this allows for congruent optic flow when walking, for example down a busy city street in a VR environment. In a more complex application, the biomechanical outcomes from a gait analysis can be calculated in real-time and integrated to the VR environment 42. This development in technology enables interaction between the patient’s movement and their sensory surroundings. The addition of VR and this user interaction may harness the potential to stimulate motor learning in rehabilitation and facilitate gait retraining. TREATMENT OF GAIT IMPAIRMENTS Targeting improved walking ability, with training, may lead to gains in increased independence and follow with increased participation in daily life. While there may be a number of competing factors relating to impaired walking ability, it is often. 11.

(26) Chapter 1. 1. Figure 2. The Gait Real-time Analysis Interactive Lab (GRAIL). Integrates motion capture with instrumented dual-belt treadmill in immersive virtual environment.. considered that children with CP develop, or learn, an impaired gait pattern and this remains relatively rigid throughout development. The brain is remarkably plastic and with greater understanding of neurology and motor learning we may hope to improve treatment of gait limitations in children with CP. We know that repetition of task specific activities is important in shaping the learning of a motor action and there is evidence to suggest that biofeedback may provide additional benefit for children with CP, where usual sensory feedback may be impaired. Therefore, in order to improve walking, one avenue may be to train children with task specific activities of walking. Gait training Gait training encompasses a range of diverse interventions with the same treatment goal. It can be defined as actively practising the task of walking, to improve walking ability. This can involve overground gait training or treadmill-based gait training. The addition of a treadmill allows a greater repetition of stepping in a safe, controlled environment, enabling increased intensity, compared with overground. Both methods may incorporate the use of partial body weight support systems. Body weight support acts to reduce load on lower limbs, allowing upright posture and gait facilitation. This could be important for individuals in GMFCS levels III and IV, where self-driven gait training would be challenging without intensive facilitation from therapists. Literature 12.

(27) General introduction investigating gait training reports a wide variety of interventions and outcome measures relating to an individual’s walking ability, making it difficult to establish the true effect of intervention. Further interventions of gait training may include the use of electromechanical gait trainers 43 and functional electrical stimulation of muscles during gait 44. These may be considered end-effector devices used, in varying extents, to simulate walking movements, which is inherently different in neural control to gait training in which gait is actively achieved by the patient. Gait training has shown promise in improving walking function in children with CP 45–47. However, currently there is no established optimal protocol for gait training as an intervention in children and young adults with CP. There is limited comparison between gait training methods in the literature and consequently no evidence about which method is most effective. For an optimal gait training intervention, motor learning principles should be implemented.. 1. Motor learning The sole performance of an action does not imply learning. Learning is the development of an action over time, observing characteristics such as improvement, consistency, stability, persistence and adaptability 48. Repeated practice of a task leads to improved performance. While this basic principle holds true, it does the complexity of the topic injustice. A three-stage model of learning was set out by Fitts and Posner 49 (Figure 3). During the first stage, the cognitive stage, the individual new to the motor skill attempts to work out the objective of the action. They attempt to establish to move what, where and when. This requires a high level of conscious control, i.e. cognitive thought to co-ordinate groups of muscles to carry out the action. The cognitive stage is discernible by its high error and variability in the outcome of the motor task. The second stage, the associative stage, occurs as the subject establishes the correct co-ordination of movements and links this with timing relating to environmental cues. There is a reduction in error and variability as the movement is refined. Eventually the action may progress to the autonomous stage. At this point the movement is habitual and can be carried out with minimal conscious thought, there is high accuracy and consistency as the movement is highly refined 49. Walking can be considered at this autonomous level among healthy subjects. There is not a definitive length of time, or number of practice sessions that defines these stages, they may be influenced by personal traits and can be influenced by instruction and augmented feedback 48. Training interventions should be focused and specific to the intended treatment goal . For example, if improved walking function is the treatment goal, then this should be directly trained. While task specificity is important, this can become monotonous. As such, variable practise should also be implemented. In this regard VR may again be implemented to provide a limitless range of environments or scenarios, introducing variation in the context of the activity. Slight variation in the motor task is beneficial. Instead of training the exact same movement over and over, small task variations will result in more robust motor learning 51. Higher variability increases the freedom of the individual for motor exploration, which allows the user to find multiple solutions for the task. This results in improved retention and transfer of the trained motor task to different scenarios 52. 50. 13.

(28) Chapter 1. 1. For effective training therapy, the goal is to instil autonomous, improved motor performance. As such, best practice in rehabilitation should be rooted in motor learning principles. We can attempt to implement a number of these key principles in rehabilitation programmes. A fundamental doctrine of rehabilitation, as touched on earlier, is practise. Repeated practise, leads to improved performance. Maintaining a high intensity during prolonged practice is important. High intensity training is likely to be a contributory factor to maximising treatment outcomes 53. The treadmill setting may act to increase the potential for more intensive gait training interventions. Treadmills allow for high intensity of practice in a controlled environment. By imposing a set speed on a patient you can increase the number of step repetitions tenfold in a single therapy session 54. Delivering high intensity can be challenging in practice and requires high motivation of patients, otherwise the risk of drop-out may be elevated. This challenge may be even greater in paediatric patient populations undergoing prolonged therapy throughout development to adolescent. One potential solution to maintaining motivation is the integration of gaming aspect in therapy. Purposeful gaming, or exergaming, involves elements of competition, a scoring or points system, a storyline or additional forms of reward for successful participation in the therapeutic game. This, the combination of challenge and reward has been shown to be effective in rehabilitation of children with CP, however the games themselves must be grounded in therapeutic principles 55. Incorporating feedback on the performance, or result, of the motor action within this gaming environment may build on an important pillar of motor learning. Biofeedback. Figure 3. Schematic representation of the three stage model of motor learning as laid out by Fitts and Posner (1967).. 14.

(29) General introduction Feedback is a universal tool for improving performance in everyday life, sport and rehabilitation. Biofeedback is a term that refers to the method of providing additional, or augmented, biological information that would otherwise not be apparent to the individual. This information may supplement the intrinsic information the individual receives via sensory input. In patient populations, intrinsic sensory information may be impaired and providing additional supportive information is even more valuable. Biofeedback in rehabilitation encompasses a diverse range of factors, from biomechanical (e.g. movement and posture) to physiological (e.g. neuromuscular and cardiovascular) signals 56.. 1. The literature surrounding the use of biofeedback in rehabilitation of walking in children with CP is to date limited, yet positive. It has been shown that children can improve aspects of walking and neuromuscular control when provided with biofeedback on muscle activation, visualized graphically on a simple screen 57–59. With rhythmic auditory stepping cues, children are able to adapt spatiotemporal aspects of walking such as step length, cadence and velocity 60,61. Auditory cues may also be used to encourage modification of specific pathologic gait patterns. Children with pathologic toe-walking can be discouraged from this walking pattern with auditory error feedback 62. How the information regarding the biological signals is provided to the user, defines how it is interpreted. Therefore, the method of delivery of biofeedback dictates the success of its use. MacIntosh et al. outline a set of characteristics that define biofeedback relating to CP 63. The method of presentation describes how the feedback is delivered to the individual, this may be haptic, visual, auditory or an integration of multiple facets. The variable upon which feedback is provided relates to how the biofeedback is perceived as an external or internal focused task. For example, when performing a squat, internal focused feedback may be “keep your knees over your toes”, while external focus would be “imagine you are sitting on a box”. Focusing attention on the effect of movement on the environment (external), rather than how to achieve the movement itself (internal) is thought to encourage faster automation in learning novel motor skills 64. As an extension of this, focus of attention may also describe how the movement is carried out. For instance, knowledge of performance describes how the movement is achieved, while knowledge of result focuses on the outcome of the movement. The last characteristic relates to the timing of the presentation of biofeedback to the individual. For example, information may be presented in real-time, or as summary measures after a session. If biofeedback is presented in real-time the frequency of this is influential. If presented continuously, the individual may begin to become overly reliant on the biofeedback and would not be able to carry out the action when the stimulus is not present. Allowing the patient autonomy to choose when the biofeedback is given may assist the learning process and empower the patient. There is currently no optimal methodology for provision of biofeedback and it is likely to vary with level of skill, cognitive and functional ability 65,66. Refining the delivery of biofeedback and taking into account the specialist needs of paediatric rehabilitation populations is essential. Understanding biofeedback on often complex, biomechanical parameters may be particularly difficult for children. Walking is a movement of the whole body and providing visual biofeedback relating to this global aspect may ultimately be beneficial. Coupled with the potential of VR, the concept of avatar based biofeedback in rehabilitation has been investigated as a promising avenue. 15.

(30) Chapter 1. 1. . Virtual avatars may provide additional value in reducing complexity of the image. For example, with a mirror or video-based feedback, patients may be distracted by selfimage issues and not able to objectively evaluate performance. A simplified avatar, or mannequin, representing a patient’s real-time actions, may facilitate understanding of the movement problem and the purpose of rehabilitation tasks. In qualitative assessment of avatar based stroke rehabilitation, communication between the therapist and patient was enhanced, with the anonymous nature of the avatar allowing for a more objective discussion on movement of the avatar and not the individual 67. The avatar also allowed for clear visual representation of progress through treatment, with improvements visualized in a comprehensive manner as opposed to represented numerically or graphically 67. 67. AIMS AND OUTLINE OF THESIS In developing novel treatment of gait limitations in children with CP there are many questions that require exploration. While therapeutic interventions in children with CP have been extensively studied, there are no guidelines for best practise in targeting gait improvement. Developments in technology have enabled the integration of real-time biomechanical analysis with VR, opening the door to a range of novel treatment areas. With greater understanding of the pathways for motor skill acquisition, the integration of biofeedback on gait may be constructive for teaching improved gait pattern and facilitate communication between the therapist and patient. The use of biofeedback and VR in CP gait rehabilitation is an emerging topic. Little is known about the extent of adaptability of gait in children with CP and if biofeedback and VR can positively influence this. The overarching aim of this thesis is to establish the extent to which children with CP can adapt and improve their gait through the novel use of biofeedback in an immersive VR environment. In order to achieve this aim there are a number of research areas that must be investigated. Firstly, we must generate a lab-specific database to further define gait in TD children to allow fair comparison of pathological gait. Then, we must ascertain to what extent gait training can lead to functional benefits in children and young adults with CP. Developing concepts and optimisation of biofeedback, we aim to explore if children can understand and adapt gait in response to real-time biofeedback on gait biomechanics. If so, then we aim to establish how individuals achieve this change at the neural level. Finally, to lay the foundation for biofeedback enhanced gait training we must show it is feasible for implementation as part of clinical practise in gait retraining. The first part of this thesis focuses on the investigation into the development of what is considered to be the typical walking pattern in healthy children. In chapter 2, a large database of gait analysis of typically developing children is reported, both over ground and on a treadmill placed in a virtual environment. This report acts as a basis upon which the gait-related outcomes in children with CP can be compared and contrasted. In chapter 3, the effects of gait training in children and young adults with CP is investigated in a systematic review and meta-analysis comparing its use to standard physical therapy. This acts to establish the potential of task specific training for functional benefits. In addition, it provides insights as to whether biofeedback and VR may act to 16.

(31) General introduction enhance outcomes. From chapter 4, we begin to explore the feasibility and potential of biofeedback to drive changes in gait in children with CP. Chapter 5 develops this concept further, with investigations into optimization of visualisation of biofeedback with use of an avatar to represent movement, targeting a range of gait parameters.. 1. Chapter 6 seeks to unravel the underlying impairments in selective motor control during gait that face children with CP. Using synergy analysis of measured muscle activations, we explore if adaptations in gait as a result of biofeedback are accompanied by changes in SMC. We also explore if this measure of SMC can indicate an individual’s ability to improve their gait. Chapter 7 brings the body of work full circle, involving the development, validation and feasibility of a clinically feasible biofeedback gait training tool. We outline the promise of this tool for future use to advance the concepts developed in this thesis. The main findings of these studies are summarised and discussed in chapter 8, with reflections on the potential of biofeedback within rehabilitation of children with CP. Gaps in the current research knowledge are identified and clinical implications are considered. We conclude with recommendations for further research avenues and outline suggestions for implementation of biofeedback in rehabilitation.. 17.

(32) Chapter 1. 1. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.. 18. Alexander, R. M. N. Bipedal animals and their differences from humans. J. Anat. 204, 321–330 (2004). Bramble, D. M. & Lieberman, D. E. Endurance running and the evolution of Homo. Nature 432, 345–352 (2004). Harcourt-Smith, W. & Aiello, L. Fossils, Feet and the Evolution of Bipedal Locomotion. J. Anat. 204, 403–416 (2004). Dominici, N. et al. Locomotor primitives in newborn babies and their development. Science (80-. ). 334, 997–999 (2011). Oskoui, M., Coutinho, F., Dykeman, J., Jetté, N. & Pringsheim, T. An update on the prevalence of cerebral palsy: A systematic review and meta-analysis. Dev. Med. Child Neurol. 55, 509–519 (2013). Odding, E., Roebroeck, M. E. & Stam, H. J. The epidemiology of cerebral palsy: Incidence, impairments and risk factors. Disabil. Rehabil. 28, 183–191 (2006). MacLennan, A. H., Thompson, S. C. & Gecz, J. Cerebral palsy: Causes, pathways, and the role of genetic variants. Am. J. Obstet. Gynecol. 213, 779–788 (2015). Fahey, M. C., Maclennan, A. H., Kretzschmar, D., Gecz, J. & Kruer, M. C. The genetic basis of cerebral palsy. Dev. Med. Child Neurol. 59, 462–469 (2017). Cans, C. Surveillance of cerebral palsy in Europe: a collaboration of cerebral palsy surveys and registers. Dev. Med. Child Neurol. 42, 816–824 (2007). Sellier, E. et al. Interrater reliability study of cerebral palsy diagnosis, neurological subtype, and gross motor function. Dev. Med. Child Neurol. 54, 815–21 (2012). Johnson, A., SCPE, S. of C. P. in E., SPCE, (Surveillance of Cerebral Palsy in Europe) & SCPE, S. of C. P. in E. Prevalence and characteristics of children with cerebral palsy in Europe. Dev. Med. Child Neurol. 44, 633–640 (2002). Pandyan, A. D. et al. Spasticity: clinical perceptions, neurological realities and meaningful measurement. Disabil. Rehabil. 27, 2–6 (2005). van den Noort, J. C. et al. European consensus on the concepts and measurement of the pathophysiological neuromuscular responses to passive muscle stretch. Eur. J. Neurol. 24, 981-e38 (2017). Novak, I. et al. A systematic review of interventions for children with cerebral palsy: state of the evidence. Dev. Med. Child Neurol. 55, 885–910 (2013). Trost, J. P., Schwartz, M. H., Krach, L. E., Dunn, M. E. & Novacheck, T. F. Comprehensive short-term outcome assessment of selective dorsal rhizotomy. Dev. Med. Child Neurol. 50, 765–771 (2008). McGinley, J. L. et al. Single-event multilevel surgery for children with cerebral palsy: A systematic review. Dev. Med. Child Neurol. 54, 117–128 (2012). Damiano, D. L. Rehabilitative therapies in cerebral palsy: The good, the not as good, and the possible. J. Child Neurol. 24, 1200–1204 (2009). Rosenbaum, P. L. et al. Prognosis for Gross Motor Function in cerebral palsy: creation of motor development curves. J. Am. Med. Assoc. 288, 1357–1363 (2002). World Health Organization. ICF : International classification of functioning, disability and health / World Health Organization. (World Health Organization, 2001). 20. World Health Organization. International Classification of Functioning, Disability and Health: Children & Youth version. (World Health Organization, 2007). Palisano, R. et al. Gross Motor Function Classification System. Dev Med Child Neurol 39, 214–223 (1997). Õunpuu, S. et al. Variation in kinematic and spatiotemporal gait parameters by Gross Motor Function Classification System level in children and adolescents with cerebral palsy. Dev. Med. Child Neurol. 57, 955–962 (2015). Russell, D., Rosembaum, P. L., Avery, L. & Lane, M. Gross Motor Function Measure (GMFM-66 and GMFM-88) User’s Manual. (Mac Keith Press, 2013). Feldman, A. B., Haley, S. M. & Coryell, J. Concurrent and Construct Validity of the Pediatric Evaluation of Disability Inventory. Phys. Ther. 70, 602–610 (1990). Landgraf, J. M. et al. Canadian-French, German and UK versions of the Child Health Questionnaire: methodology and preliminary item scaling results. Qual. Life Res. 7, 433–445 (1998). Schiariti, V. et al. Comparing contents of outcome measures in cerebral palsy using the international classification of functioning (ICF-CY): A systematic review. Eur. J. Paediatr. Neurol. 18, 1–12 (2014). Baker, R., Esquenazi, A., Benedetti, M. G. & Desloovere, K. Gait analysis: clinical facts. Eur. J. Phys. Rehabil. Med. 52, 560–74 (2016). McGinley, J. L., Baker, R., Wolfe, R. & Morris, M. E. The reliability of three-dimensional kinematic gait.

(33) General introduction measurements: A systematic review. Gait Posture 29, 360–369 (2009). 29. Nieuwenhuys, A. et al. Identification of joint patterns during gait in children with cerebral palsy: A Delphi consensus study. Dev. Med. Child Neurol. 58, 306–313 (2016). 30. Dobson, F., Morris, M. E., Baker, R., Wolfe, R. & Graham, H. Clinician agreement on gait pattern ratings in children with spastic hemiplegia. Dev. Med. Child Neurol. 48, 429–35 (2006). 31. Rodda, J. M., Graham, H. K., Carson, L., Galea, M. P. & Wolfe, R. Sagittal gait patterns in spastic diplegia. J. Bone Joint Surg. Br. 86, 251–258 (2004). 32. Cook, R. E., Schneider, I., Hazlewood, M. E., Hillman, S. J. & Robb, J. E. Gait analysis alters decisionmaking in cerebral palsy. J. Pediatr. Orthop. 23, 292–295 (2003). 33. Alton, F., Baldey, L., Caplan, S. & Morrissey, M. C. A kinematic comparision of overground and treadmill walking. Clin. Biomech. 13, 434–440 (1998). 34. Lee, S. J. & Hidler, J. Biomechanics of overground vs. treadmill walking in healthy individuals. J. Appl. Physiol. 104, 747–755 (2008). 35. Riley, P. O., Paolini, G., Della Croce, U., Paylo, K. W. & Kerrigan, D. C. A kinematic and kinetic comparison of overground and treadmill walking in healthy subjects. Gait Posture 26, 17–24 (2007). 36. Stolze, H. et al. Gait analysis during treadmill and overground locomotion in children and adults. Electroencephalogr. Clin. Neurophysiol. Mot. Control 105, 490–497 (1997). 37. Sloot, L. H., van der Krogt, M. M. & Harlaar, J. The biomechanical effect of a virtual reality depends on treadmill mode. Gait Posture 39, S49–S50 (2014). 38. Sloot, L. H., Harlaar, J., van der Krogt, M. M. & Krogt, M. M. Van Der. Self-paced versus fixed speed walking and the effect of virtual reality in children with cerebral palsy. Gait Posture 42, 498–504 (2015). 39. Plotnik, M. et al. Self-selected gait speed - Over ground versus self-paced treadmill walking, a solution for a paradox. J. Neuroeng. Rehabil. 12, (2015). 40. van der Krogt, M. M., Sloot, L. H. & Harlaar, J. Overground versus self-paced treadmill walking in a virtual environment in children with cerebral palsy. Gait Posture 40, 587–593 (2014). 41. Geijtenbeek, T. & Steenbrink, F. D-Flow : Immersive Virtual Reality and Real-Time Feedback for Rehabilitation. 10th Int. Conf. Virtual Real. Contin. Its Appl. Ind. 1, 201–208 (2011). 42. van den Bogert, A. J., Geijtenbeek, T., Even-Zohar, O., Steenbrink, F. & Hardin, E. C. A real-time system for biomechanical analysis of human movement and muscle function. Med. Biol. Eng. Comput. 51, 1069–77 (2013). 43. Lefmann, S., Russo, R. & Hillier, S. The effectiveness of robotic-assisted gait training for paediatric gait disorders: systematic review. J. Neuroeng. Rehabil. 14, 1 (2017). 44. Pool, D., Blackmore, A. M., Bear, N. & Valentine, J. Effects of functional electrical stimulation in children with spastic hemiplegia. Dev. Med. Child Neurol. 56, 9–10 (2014). 45. Mutlu, A., Krosschell, K. & Spira, D. G. Treadmill training with partial body-weight support in children with cerebral palsy: A systematic review. Dev. Med. Child Neurol. 51, 268–275 (2009). 46. Willoughby, K., Dodd, K. J. & Shields, N. The effectiveness and efficacy of treadmill training for children with cerebral palsy - A randomised controlled trial. Dev. Med. Child Neurol. 51, 10 (2009). 47. Damiano, D. L. D. et al. A systematic review of the effectiveness of treadmill training and body weight support in pediatric rehabilitation. J. Neurol. Phys. Ther. 33, 27–44 (2009). 48. Magill, R. a. Motor learning and control. (McGraw-Hill Education, 2007). 49. Fitts, P. & Postner, M. Human Preformance. (Brooks/Cole, 1967). 50. Barbeau, H. Locomotor training in neurorehabilitation: Emerging rehabilitation concepts. Neurorehabil. Neural Repair 17, 3–11 (2003). 51. Wu, H. G., Miyamoto, Y. R., Castro, L. N. G., Ölveczky, B. P. & Smith, M. A. Temporal structure of motor variability is dynamically regulated and predicts motor learning ability. Nat. Neurosci. 17, 312–321 (2014). 52. Shea, J. B. & Zimny, S. T. Context Effects in Memory and Learning Movement Information. in Memory and Control of Action (ed. Magill, R. A. B. T.-A. in P.) 12, 345–366 (North-Holland, 1983). 53. Langhorne, P., Bernhardt, J. & Kwakkel, G. Stroke rehabilitation. Lancet 377, 1693–1702 (2015). 54. Hesse, S. Locomotor therapy in neurorehabilitation. NeuroRehabilitation 16, 133–9 (2001). 55. Lewis, G. N. & Rosie, J. A. Virtual reality games for movement rehabilitation in neurological conditions: How do we meet the needs and expectations of the users. Disabil. Rehabil. 34, 1880–1886 (2012). 56. Giggins, O. M., Persson, U. & Caulfield, B. Biofeedback in rehabilitation. J. Neuroeng. Rehabil. 10, 60 (2013). 57. Colborne, G. R. R. et al. Feedback of triceps surae EMG in gait of children with cerebral palsy: a controlled study. Arch Phys Med Rehabil. 75, 40–5 (1994). 58. Dursun, E., Dursun, N. & Alican, D. Effects of biofeedback treatment on gait in children with cerebral palsy. Disabil. Rehabil. 26, 116–120 (2004).. 1. 19.

(34) Chapter 1. 1. 20. 59. Bolek, J. E. A Preliminary Study of Modification of Gait in Real-Time Using Surface Electromyography. Appl. Psychophysiol. Biofeedback 28, 129–138 (2003). 60. Hamed, N. S., Abd-elwahab, M. S., NS, H. & MS, A. Pedometer-based gait training in children with spastic hemiparetic cerebral palsy: a randomized controlled study. Clin. Rehabil. 25, 157–165 (2011). 61. Baram, Y. & Lenger, R. Gait improvement in patients with cerebral palsy by visual and auditory feedback. 2009 Virtual Rehabil. Int. Conf. (2009). doi:10.1109/ICVR.2009.5174222 62. Pu, F. et al. Feedback System Based on Plantar Pressure for Monitoring Toe-Walking Strides in Children with Cerebral Palsy. Am. J. Phys. Med. Rehabil. 93, 122–129 (2014). 63. MacIntosh, A., Lam, E., Vigneron, V., Vignais, N. & Biddiss, E. Biofeedback interventions for individuals with cerebral palsy: a systematic review. Disabil. Rehabil. 6, 1–23 (2018). 64. Lepage, C., Noreau, L. & Bernard, P. M. Association between characteristics of locomotion and accomplishment of life habits in children with cerebral palsy. Phys. Ther. 78, 458–469 (1998). 65. Sigrist, R., Rauter, G., Riener, R. & Wolf, P. Augmented visual, auditory, haptic, and multimodal feedback in motor learning: a review. Psychon. Bull. Rev. 20, 21–53 (2013). 66. Richards, R., van den Noort, J. C., Dekker, J. & Harlaar, J. Gait Retraining with real-time Biofeedback to reduce Knee adduction moment: systematic review of effects and methods used. Arch Phys Med Rehabil (2016). doi:10.1016/j.apmr.2016.07.006 67. Loudon, D. et al. Developing visualisation software for rehabilitation: Investigating the requirements of patients, therapists and the rehabilitation process. Health Informatics J. 18, 171–180 (2012)..

(35)

(36)

(37) Chapter 2. How normal is normal? Consequences of not accounting for stride-to-stride variability in paediatric reference gait data. Laura Oudenhoven, Adam Booth, Annemieke Buizer, Jaap Harlaar, Marjolein van der Krogt. Gait & Posture, 2019, 70:289-297.

(38) Chapter 2. 2. 24. Abstract Background: In the process of 3D-gait analysis interpretation, gait deviations in children with cerebral palsy are identified through comparison with reference data of typically developing children (TD). Generally, TD-data are presented based on averaged normalized curves of numerous strides for different ages and walking velocities. In patients however, often only a limited number of strides are available which are compared to group-averaged reference curves. Aim: To investigate the impact of ignoring stride-to-stride variation when averaged normalized curves are used as a reference paediatric dataset. To illustrate the implications for clinical practice, we investigated how many individual strides of TD children would be classified as abnormal, when compared to averaged normalized curves from the reference group, and how this is affected by age and treadmill versus overground walking. Methods: Ninety TD datasets were collected. Children (4-18y) walked on a 10m-walkway (n=49) or instrumented treadmill (n=41). Joint kinematic and kinetic curves and clinically relevant outcome parameters were established. Individual strides were considered abnormal if they exceeded the group average more than 2SD. In addition, the Edinburgh Visual Gait Score, Gait Profile Score (GPS) and GPS stride-to-stride variability were calculated. Generalized estimation equation analyses were used to investigate effects of age, overground/treadmill and their interaction. Results: Of all 2532 analysed strides, on average 28% were classified as abnormal for joint kinematic curves, 50% for moments, and 51% for powers. Younger children showed a greater percentage of abnormal strides, higher GPS and greater variability (p < 0.001). The effect of age was similar between treadmill and overground, but variability was lower on the treadmill. Significance: Our findings indicate that due to stride-to-stride variability, even in TD children a substantial number of strides can be classified as abnormal, when compared to group averaged normalized curves. Consequently, in patients, comparing a single stride to such a reference curve may lead to potential overestimation of gait deviations..

(39) How normal is normal? INTRODUCTION Three-dimensional clinical gait analysis is often performed in children with cerebral palsy (CP), to guide treatment decisions and improve surgical outcomes. In one aspect of this process, the measured gait function of children with CP is compared to a set of reference gait data of typically developing (TD) children. These reference datasets are often presented as grouped averages 1, for children across different ages and walking velocities, visualized as a grey band in the gait report, representing the mean plus group standard deviation 1–2. In clinical practice, a limited number of strides of a child with CP is collected for comparison to this reference dataset. Clinicians then use this to highlight pathological gait features that may fall outside the ‘normal’ walking curve 1-4. Findings from gait analysis, along with clinical measures are combined, providing indications for intervention 1, 4. Overall gait function can be also quantified relative to this normative dataset, using scales such as the gait profile score (GPS) 2 and gait deviation index (GDI) 5, 6.. 2. Although this process is accepted as standard clinical practice, and described in methods such as the impairment focused approach to interpretation 1, some basic statistical assumptions are violated in this approach. First, reference curves are usually based on grouped averages; multiple strides are first averaged within an individual and then averaged across the group. Therefore, the variability presented by the grey band (i.e. plus and minus one or two standard deviation) does not reflect variability of underlying individual strides, but variability between the averaged strides of subjects. Consequently, stride-to-stride variability within paediatric gait is masked by averaging over many strides. While several studies investigated effects of stride-to-stride variability in unimpaired adults and patient groups 7, little is known about these effects in children. Based on theories of motor development, it can be suggested that age is an important factor, since movements in younger children are more variable than in older individuals 8-11. Secondly, one of the assumptions of averaging gait curves, after time normalizing to 0-100%, is that all variation is in the amplitudes, not in timing of gait events. Although this holds to some extent for kinematics while walking at a fixed velocity, walking speed has a strong effect on the timing of gait events 12, 13. Therefore, temporal misalignment of peaks is likely to occur 14, 15. Since walking speed is more variable in younger children 11, the consequences of this assumption may be exacerbated in younger children to a larger extent than in older children. In addition, it may play a bigger role for children walking in a conventional overground gait lab at a free walking speed, compared to walking on a treadmill, since treadmill walking allows for a more continuous walking pattern. To address these issues, the aim of the present study was to investigate the consequences of ignoring stride-to-stride variation when averaged normalized curves are used as a reference paediatric dataset for individual strides, and how this is affected by age and treadmill versus overground walking. To illustrate the implications for clinical practice, we investigated how many individual strides of TD children could be classified as abnormal, when compared to their own reference group. Due to stride-to-stride variability, we expected this number of abnormal strides to be higher in younger children and in overground walking. Since previous 25.

(40) Chapter 2. 2. studies showed that gait is highly affected by walking speed speed were also considered.. 12. , effects of walking. METHODS Participants Ninety datasets were collected of TD children walking either in an overground gait lab or on a treadmill. Children were included if they were between 4-18 years of age and had no impairments or conditions that could interfere with walking ability. Exclusion criteria were any injuries, physical conditions or cognitive disabilities that might affect walking ability. Written informed consent was given by parents and children (if >12 years) before the start of the experiment. The study was approved by the local ethics committees of the VU University Medical Centre and the VU University, both in line with the declaration of Helsinki. Study design and materials The overground gait laboratory consisted of a 10m walkway with two embedded force plates (AMTI, Watertown, MA, USA). After familiarization, children were asked to walk up and down the walkway at a self-chosen comfortable walking speed. During the measurements, parents were asked to confirm whether children showed their typical daily life gait pattern. Measurements were continued until sufficient representative strides were collected, aiming for at least five force plate hits for each leg. The treadmill laboratory consisted of a dual-belt instrumented treadmill with speed-matched virtual reality environment projected on a surrounded screen (Gait Real-time Analysis Interactive Laboratory (GRAIL) system, Motek Medical B.V., Amsterdam, the Netherlands). After a familiarization period of at least 6 minutes, children walked at their self-chosen walking velocity. This was determined either by the child with a self-paced algorithm or using a fixed speed condition. In the latter case, the speed was increased until the children felt comfortable, at this stage the speed was increased until it was considered too fast for walking, speed was then reduced until the participants suggested it was their comfortable walking speed. The speed in the middle of the two speeds suggested as comfortable was taken as the self-selected walking speed and fixed throughout the rest of data collection, which was at least 1 minute. 3D motion data was captured with passive retroreflective and infrared cameras (Vicon, Oxford, UK) in both labs, following an updated version of the Human Body Model (HBM) 16,17. Most relevant updates to the previous reported version of HBM concerned the knee and ankle axis, which are no longer pose dependent due to the use of medial markers, making it more robust for clinical use in patients. Data was processed using the Gait Offline Analysis Tool (version 4.0, Motek Medical B.V., Amsterdam, The Netherlands). Data analysis. 26.

(41) How normal is normal? 3D kinematic and kinetic data were analysed the right leg only. Gait events were calculated according to Zeni et al. 18. Strides that were clear outliers (e.g. running, standing still, skipping) were manually excluded. Walking speed, stride length, and step length were non-dimensionalised through normalization to leg length (maximal distance ASIS to medial malleolus during the static trial), following the methods of Hof et al. 19. Both overground and treadmill data sets are shared in the supplementary material, as this paper may also serve as a paediatric reference base for HBM 16.. 2. To assess the normality of individual strides, four different approaches were used: Method A: Kinematic and kinetic curves of individual strides were compared to the group averaged curves of the same parameter. Strides were considered ‘abnormal’, if individual strides of a child exceeded the group averaged curves by more than 2 SD (group SD) for more than 10% of the gait cycle. Method B: A set of clinically relevant outcome parameters (CROPs) were extracted for individual strides of all children (see Table 2). Based on these CROPs, abnormal strides were identified as individual strides where CROPs exceeded the group averaged CROP by more than 2 SD (group SD) in either amplitude or timing. Method C: As a measure for overall gait abnormality, the gait profile score (GPS) and Edinburgh Visual Gait Score (EVGS) were calculated. GPS was calculated using the root mean square (RMS) difference between gait kinematic curves for individual children, compared to the group average, following the methods described by Baker 5 . In addition to the GPS for kinematics, GPS was also calculated for joint moments and powers, using the same methods as applied for kinematics. As several video-based measures may have additional clinical relevance on top of 3D kinematics, EVGS score was calculated for the overground data, based on deviations at clinically meaningful events of the gait cycle 20. EVGS was assessed from frontal and sagittal videos, and averaged over both legs. All videos were analysed by the same experienced investigator, using custom-made video analysis software 21. Method D: For spatiotemporal parameters, GPS and CROPs, the stride-to-stride variability was quantified as the SD between strides within each individual child. Statistics To assess the effect of age, condition (treadmill/overground), their interaction (age X condition) and non-dimensional walking speed on all outcome measures, linear generalized estimation equation (GEE) analyses were performed. This method was chosen, as it can correct for differences of walking speed between children and between the conditions. For each outcome parameter evaluated, an individual linear equation was evaluated, with β-values representing coefficients of the individual regression terms and p-values evaluating whether terms contributed significantly to the constructed equation. Due to the large number of parameters evaluated, significance level was set at p<0.01. The term correcting for non-dimensional walking speed was included only if it was significant at p<0.05. In addition, independent sample t-tests were performed to compare participant characteristics between both conditions, with a significance level of p<0.05.. 27.

(42) Chapter 2 RESULTS. 2. Data of 49 children (28 girls, 21 boys) were included for the overground 3D gait analyses and of 48 children for 2D video analysis. Data were excluded due to a lack of complete video recordings (n=2, excluded from video analysis only) and absence of kinetic data (n=1, excluded from all analyses). Forty-one datasets (17 girls, 24 boys) were collected on the treadmill, data of three children were excluded from kinetic analysis due to hardware issues. For individuals walking overground, for kinematics, on average 17 strides were included (min: 10, max: 38), whereas for kinetics on average 6 strides were included for analysis (min: 1, max:12). For treadmill walking, for kinematics on average 34 were included (min: 11, max: 54), whereas for kinetics on average 27 strides were included (min: 11, max: 44). No differences were found between the overground and treadmill group for participant’s age or mass (overground: mean age 9.6y, range 4-17; mass 36.8±16kg vs. treadmill: mean age 10y, range 5-15; mass 40.1±13.2kg). However, children in the treadmill group were slightly taller (overground: 1.38±0.21m vs. treadmill: 1.48±0.19m; p=.044) and absolute as well as non-dimensional comfortable walking speed were slower on the treadmill compared to overground walking (overground: 1.22±0.14m/s vs. treadmill: 1.05±0.20m/s; p<0.01). Kinematic, kinetic and CROP data are presented in Figures 1 & 2 and Tables 1 & 2. Complete results, kinematics/kinetics curves, and spatiotemporal parameters are available in the supplementary material (available online). For method A, on average 28% were classified as abnormal for joint kinematic curves, 50% for moments, and 51% for powers, as a mean of the different joints. The least strides were classified as abnormal for pelvic tilt (9%) and most for hip rotation (41%) (Table 1). For joint moments, least strides were classified as abnormal for hip flexion moments (43%) and most for hip rotational moments (59%). For joint powers, least strides were classified as abnormal for hip power (49%) and most for ankle power (53%). For most kinematic curves, significant effects were found for age on the number of abnormal strides, where younger children showed a higher percentage of abnormal strides (Table 1). For kinetics, the number of abnormal strides was related to walking speed for most parameters, where more abnormal strides were classified for faster walking speeds. Overall, the number of abnormal strides was higher on the treadmill than overground for some kinetic parameters. No interaction effects were found between condition and age for both kinematic and kinetic curves. For method B, the number of strides classified abnormal based on CROPs was on average 15% for kinematics. The least number of abnormal strides were found for pelvis tilt (5%) and most for peak abduction in swing phase (25%). For kinetics, on average 15% of strides were classified abnormal, where least abnormal strides were classified for peak ankle plantarflexion moment (6%) and most for peak hip moments (23%) (Table 2). The number of abnormal strides based on CROPs was related to children’s age for most parameters, even after correction for non-dimensional walking speed,. 28.

(43) How normal is normal? again with more abnormal strides for younger children. For kinetics, effects were mainly related to non-dimensional walking speed. For most outcome parameters, no difference was found between overground and treadmill walking (Tables 1,2, Figures 1-3).. 2. Concerning method C, mean GPS score was 6.4° for kinematics (Table 2). Alternative GPS scores calculated for kinetics were 0.12 Nm/kg for joint moments and 0.35 J/ kg for powers. Mean EVGS was 1.7 (only calculated for children who walked in the overground condition). GPS was significantly related to age, where older children. Figure 1. Kinematic curves averaged over all typically developing children for the overground (red) and treadmill (blue) condition, with two standard deviations for each group (light shaded; darker colours indicate 1SD). Crosses indicate a selection of the clinically relevant outcome parameters (CROPS), i.e. peak values, with two standard deviations in amplitude (vertical) and timing (horizontal). As an example of inter-individual variation, individual strides of one boy are presented (thin black lines) as well as his individual averaged curve (dashed red line), while walking in the overground gait lab.. 29.

(44) 2. 30. Plflx-dflx (% abnormal). Knee. Hip. 50 ± 34. 55 ± 30. 60 ± 31. Flex-ext (% abnormal). AbAdd (% abnormal). Rot (% abnormal). 28 ± 31. 33 ± 31. 24 ± 27. Rotation (% abnormal). Flex-ext (% abnormal). 26 ± 30. Abd-Add (% abnormal). 41 ± 27. Rotation (% abnormal). 14 ± 23. 28 ± 31. Flex-ext (% abnormal). 10 ± 22. Tilt (% abnormal). Mean ± SD. Obliquity (% abnormal). Kinetics - Moments. Ankle. Knee. Hip. Pelvis. Kinematics. Outcome measure. -2.0. 47.1. -41.3. 0.4. 1.0 3.2. -64.4. -2.0. -66.6. 51.4. -5.0. -3.4. 110.2. 84.0. -2.6. -3.1. 73.6 38.9. -1.5 -3.0. 55.6. Β2. 0.726. 0.005. 0.526. 0.111. <0.001. 0.012. 0.001. <0.001. 0.001. 0.006. 0.011. p. Age Effect. 23.9. Β1. Intercept. 0.052 0.002. 48.3. 0.039. 0.414. 0.330. 0.877. 0.864. 0.604. 0.698. 0.678. 0.632. p. 35.7. 44.1. -15.7. -17.7. 2.5. -3.2. 9.1. -6.2. -8.8. 7.5. Β3. Condition Effect TM1/OG0. -0.4. -1.6. -1.3. 0.6. 1.3. -1.0. -0.8. -0.8. 0.2. 1.2. -0.8. Β4. 0.80. 0.36. 0.50. 0.76. 0.45. 0.45. 0.63. 0.58. 0.91. 0.55. 0.51. p. Age x Condition. 183.2. 189.8. 213.5. -107.6. Β5. <0.001. <0.001. <0.001. 0.007. p. Walking Speed Effect. Table 1. Percentage of abnormal strides based on individual gait curves exceeding 2-SD during &10% of the gait cycle (method A). Mean/SD represent group average, expressed as the ratio between the number of abnormal strides of an individual and his/her number of strides analyzed. Evaluated results in GEE concern the effect of age, condition (treadmill/overground), their interaction, and walking speed on percentage of abnormal strides. GEE according to: percentage abnormal strides= β1 + (β2*age) + β3*(condition) + β4*(age X condition) β5*(non-dimensional walking speed). Condition was entered as a categorical variable: treadmill =1, overground =0. Individual strides were classified abnormal if curves exceeded group mean +/- 2SD for > 10% of the gait cycle (method A). Note that non-dimensional walking speed was only included p < 0.05. Results significant at p < 0.01.. Chapter 2.

(45) 56 ± 36. 53 ± 35. Ankle (% abnormal). Bold numbers indicate significant effects (p < 0.01).. 55 ± 35. Hip (% abnormal). Knee (% abnormal). 54 ± 34. Rot (% abnormal). Plflx-dflx (% abnormal). 64 ± 29. AbAdd (% abnormal). Kinetics - Powers. Ankle. 57 ± 34. 53 ± 30. Flex-ext(% abnormal). Table 1 continued.... <0.001. -1.8 -3.5. -47.0. 0.101. -1.0. 0.337. 0.520. -69.4. -0.8. 0.748. 0.066. 0.858. -110.8. -28.8. -0.3. 2.2. -35.8. -0.3. -31.3 -78.7. 38.7. 34.0. 39.5 0.017. 0.053. 0.007. 0.033. <0.001. 46.2 43.0. 0.003. 0.086. 51.7. 37.0. 1.1. 0.9. 0.2. -0.2. 0.1. -1.7. -0.2. 0.91. 0.48. 0.60. 0.90. 0.94. 0.91. 0.34. 262.6. 289.1. 367.1. 170.0. 192.4. 221.1. 177.4. <0.001. <0.001. <0.001. <0.001. <0.001. <0.001. <0.001. How normal is normal?. 2. 31.

(46) 2. 32. Pelvis. Kinematics. 1.7 ± 1.3. EVGS. 2 ± 1. 0.35 ± 0.09. SD. SD. 0.35 ± 0.09. GPS powers. 10 ± 7. 0.02 ± 0.01. SD. Mean tilt (°). 0.12 ± 0.02. GPS moments. 6 ± 2. 0.9 ± 0.5. GPS (°). SD. Overall gait score. 0.03 ± 0.01. 1.1 ± 0.3. SD. SD. 65 ± 1. Stance/swing (%). 0.63 ± 0.09. 1.2 ± 0.2. ND walking speed. Stance time (s). 1.2 ± 0.2. Mean/SD. Walking speed (m/s). Spatiotemporal. Outcome measure. 4.49. 11.86. 2.64. 0.04. 0.10. 0.03. 0.09. 1.91. 8.50. 0.06. 0.85. 1.82. 69.17. 0.50. 1.01. B1. Intercept. 0.001. −0.01. 0.166 <0.001. −0.32 −0.26. 0.023. 0.003. −0.01 −0.10. 0.469. 0.854. 0.00. 0.00. <0.001. <0.001. 0.00 <0.001. <0.001. 0.01. −0.10. <0.001. −0.07. −0.20. 0.417. 0.021. 0.004. p. 0.002. 0.001. 0.02. B2. Age Effect. −1.63. 3712.00. n.a.. −0.03. 0.01. 0.00. 3. −0.53. 0.61. −0.02. −0.01. −0.16. 0.50. −0.07. −0.17. B3. 0.005. 0.350. n.a.. 0.268. 0.912. 0.984. 0.042. 0.007. 0.421. 0.026. 0.737. 0.364. 0.236. 0.135. 0.165. p. Condition Effect TM1/OG0. 0.13. −0.14. n.a.. 0.00. 0.00. 0.00. 0.00. 0.03. −0.09. 0.00. 0.00. 0.02. −0.03. 0.00. 0.00. B4. 0.010. 0.690. n.a.. 0.381. 0.469. 0.267. 0.509. 0.045. 0.251. 0.148. 0.591. 0.144. 0.483. 0.948. 0.988. P. Age x Condition. 0.19. 0.73. 0.06. <0.001. <0.001. 0.047. <0.001. <0.001. −11.20 −0.72. n.a.. n.a.. p. n.a.. n.a.. B5. Walking Speed Effect. Table 2. Mean values and results of GEE analysis for clinically relevant outcome parameters (CROPS). Outcome measures presented in the first column represent: CROP parameters, individual standard deviations (within-subject SD) and percentage number of abnormal strides for individuals, based on values exceeding the group value +/− 2 SD (method B). Evaluated results in GEE concern the effect of age, condition (treadmill/overground) as well as their interaction, where B-values represent coefficients for each of these parameters. Note that non-dimensional walking speed was only included if in contributed significantly to the regression formula (p < 0.05). GEE according to: outcome = β1+(β2*age) + β3*(condition) + β4*(age X condition) β5*(non-dimensional walking speed).. Chapter 2.

(47) Ankle. Knee. Hip. 2 ± 1. 65 ± 5. 3 ± 2. 15 ± 22. 66 ± 6. 3 ± 2. 15 ± 22. % abnormal. Range of motion (°). SD. % abnormal. 3 ± 1. SD. SD. 0 ± 5. Peak flexion swing (°). 14 ± 22. % abnormal. Peak ext stance (°). 17 ± 23. 47.32. 3 ± 1. SD. % abnormal. 5.03. 38 ± 6. Peak flexion stance (°). 6.92. 17.90. 6.76. 53.77. 33.98. 5.81. 59.04. 30.12. 5.23. 0.17. 38.92. 17.74. 3 ± 2. 10 ± 18. SD. 12.95. 45.94. 3.16. −0.27. 39.37. 5.01. −1.36. 11.07. % abnormal. 8 ± 5. Flexion at ic (°). 24 ± 28. SD. % abnormal. 1 ± 3. 18 ± 20. 2 ± 1. −11 ± 8. 5 ± 18. Peak abd swing(°). % abnormal. SD. Minimal flexion (°). % abnormal**. Table 2 continued.... <0.001 0.006. <0.001. −1.96. −0.97. 0.001. −0.20. −0.28. <0.001. −0.69. 0.010. <0.001. −2.06. −0.45. <0.001. −0.26. −3.03 0.134. <0.001. −0.18 −0.22. 0.001 <0.001. −0.39. 0.561. <0.001. −0.36 −0.63. <0.001. −0.65. 0.154. <0.001. −0.15 −1.88. 0.198. 0.001. −2.25 0.15. 0.606 <0.001. −0.13. 0.025. −0.25. −0.72. −4.47. −2.65. 7.05. −21.83. −2.46. 6.62. −10.39. −0.84. 1582.00. −36.70. −0.54. 8.52. 4.65. −2.32. −5.19. −8.60. −0.33. −1.36. 5.58. −0.84. 10.14. 9.20. 0.575. <0.001. 0.042. 0.275. <0.001. 0.016. 0.513. 0.176. 0.660. 0.007. 0.348. 0.018. 0.561. <0.001. 0.096. 0.655. 0.321. 0.545. 0.629. 0.190. 0.037. 0.468. 0.21. 0.11. −0.52. 2.16. 0.08. −0.22. 2.93. 0.04. 0.18. 2.80. 0.03. −0.15. −0.69. 0.16. 0.77. −0.01. 0.04. 0.06. −0.39. 0.06. −0.40. −0.72. 0.754. 0.364. 0.102. 0.253. 0.213. 0.398. 0.066. 0.422. 0.580. 0.021. 0.552. 0.649. 0.595. 0.006. 0.008. 0.996. 0.226. 0.761. 0.707. 0.303. 0.356. 0.441. 36.77. 24.30. −24.98. <0.001. 0.001. 0.024. How normal is normal?. 2. 33.

(48) 34. 2 ± 1. SD. % abnormal. Knee. Hip. 0.1 ± 0.0. 17 ± 23. SD. % abnormal. −7.70. 0.09. 0.18. −7.70. 17 ± 23. 0.0 ± 0.1. % abnormal. 0.1 ± 0.1. Mean ext (Nm/kg). 0.08. 0.5 ± 0.2. 0.24. −3.68. SD. 9 ± 19. % abnormal. 0.11. 0.01. 37.20. −0.07. −0.39. 15.10. 5.47. −4.43. 9.81. 2.88. Peak ext (Nm/kg). −1 ± 0. 0.1 ± 0.0. SD. 23 ± 22. Peak hip flexion (Nm/kg). 0.1 ± 0.1. SD. % abnormal. Peak extension (Nm/kg). 1 ± 0. 7 ± 15. SD. Kinetics - Moments. −6 ± 7. 3.4 ± 1.2. Mean progression (°). Foot. 15 ± 26. 5 ± 4. % abnormal. 50.97. 21 ± 27. % abnormal. Peak dflx swing (°). 3.76. 3.06. 2 ± 1. SD. 9.83. 12 ± 4. Peak dflx stance (°). Table 2 continued.... 0.608. 0.005. 0.00 −0.63. 0.242. 0.608. 0.158. 0.724. 0.218. −0.01. −0.63. −0.01. 0.00. 1.36. 0.611. <0.001. −0.03 0.00. 0.373. 0.199. <0.001. 0.036. −0.85. 0.00. 0.02. −0.76. 0.344 <0.001. −0.19 −0.19. 0.382. <0.001. 1.07. −0.11. 0.014 0.634. 0.06. −2.80. 0.449 <0.001. 0.12 −0.14. 6.91. −11.60. −0.04. −0.06. −11.60. −0.09. −0.32. −7.17. 2. 2. 0.00. −0.01. 0.18. 1.38. −0.35. 9.13. 10.52. −0.46. 3.59. −25.76. −0.45. 0.494. 0.019. 0.429. 0.494. 0.027. 0.011. 0.575. 0.421. 0.805. 1.000. 0.901. 0.060. 0.872. 0.557. 0.330. 0.588. 0.191. 0.180. 0.126. 0.283. 0.023. 1.44. 0.00. 0.01. 1.44. 0.01. 0.03. 0.44. 0.00. 0.00. −1.30. 0.00. −0.01. −0.22. 0.00. −0.78. −2.09. 0.01. −0.13. 1.99. 0.04. −0.37. 0.396. 0.071. 0.199. 0.396. 0.063. 0.011. 0.772. 0.361. 0.815. 0.329. 0.849. 0.239. 0.751. 0.986. 0.043. 0.264. 0.655. 0.610. 0.208. 0.321. 0.199. 69.13. 0.015. 0.015 0.006. 69.13 −0.40. 0.022 <0.001. 0.61 0.26. <0.001. <0.001. 0.21 −0.88. <0.001. 1.86. Chapter 2.

(49) 1.3 ± 0.2. 0.1 ± 0.1. 6 ± 16. SD. % abnormal. 17 ± 23. % abnormal. Peak pflx (Nm/kg). 0.1 ± 0.0. SD. 2.1 ± 0.7. 0.4 ± 0.2. 9.1 ± 18.8. SD. % abnormal. 22 ± 23. Peak ankle (W/kg). 0.3 ± 0.2. SD. % abnormal. 16 ± 21. 0.9 ± 0.4. 0.3 ± 0.2. SD. % abnormal. Peak knee (W/kg). 1.3 ± 0.4. Peak hip (W/kg). 0.16. 19.16. 0.05. −1.28. 37.45. 0.22. −0.42. 45.02. 0.26. −0.33. 0.13. 0.14. 0.30. −7.70. 0.05. −0.99. −0.01. 0.13. <0.001. −0.02. 0.316. 0.308. <0.001. 0.006. 0.315. −0.01. 0.004. <0.001. −0.03 −2.49. 0.226. 0.512. 0.467. <0.001. −0.02. 0.55. 0.00. 0.05. 0.608. 0.023. −0.63. 0.002. 0.01 0.00. −0.01. 7. −19.51. 0.13. 0.76. 7.23. 0.155. 0.287. 0.041. 0.637. 0.687. 0.492. 0.003. 25. 0.003. −0.34. 0.102. 0.461. 0.253. 0.006. 0.494. 0.020. 0.886. −34.92. −0.31. 9.36. −0.04. 0.22. −11.60. −0.02. 1.85. −0.01. −0.07. 0.26. −0.01. −0.01. 2.41. 0.03. 0.03. −0.77. 0.00. −0.01. 1.44. 0.00. 0.00. 0.171. 0.350. 0.039. 0.860. 0.732. 0.674. 0.025. <0.001. 0.033. 0.558. 0.791. 0.165. 0.396. 0.018. 0.858. <0.001 <0.001. 0.87. 0.002. <0.001. <0.001. <0.001. <0.001. 0.015. 0.020. 4.81. 0.64. 3.26. 0.92. 3.99. 0.97. 69.13. 0.05. Condition was entered as a categorical variable: treadmill =1, overground =0. Results significant at p < 0.01. Bold numbers indicate significant effects (p < 0.01). **Percentage of abnormal strides for pelvis tilt is not reported, since only 1 individual presented abnormal strides > 0.. Ankle. Knee. Hip. Kinetics - Powers. Ankle. 0.2 ± 0.1. Peak abd (Nm/kg). Table 2 continued.... How normal is normal?. 35.

(50) Chapter 2. 2. Figure 2. Kinetic curves averaged over all typically developing children for the overground (red) and treadmill (blue) condition, with two standard deviations for each group (shaded). Crosses indicate clinically relevant outcome parameters (CROPS), i.e. peak values, with two standards in amplitude (vertical) and timing (horizontal). Positive values indicate internal abduction (coronal plane) and extension/plantar flexion (sagittal plane).. showed lower GPS scores (less deviations) (p<0.001, Figure 3). These effects were similar between overground and treadmill walking and not affected by walking speed. A comparable trend of a decrease with age was found for the EVGS, although this did not reach significance (p=0.023). Stride-to-stride variability (method D), expressed by children’s individual SD, decreased with age, where younger children presented higher SDs for most parameters (Table 2). Overall, stride-to-stride variability was higher for overground compared to treadmill walking for most outcome parameters, including SD of GPS (p=0.007, Figure 3). Effects of age were comparable between overground/treadmill walking for most parameters, as indicated by the non-significant interaction effects between age and condition (Table 2).. 36.

(51) How normal is normal?. 2. Figure 3. Progression with age of the Edinburg Visual Gait Score (EVGS) (upper panel), gait profile score (GPS) (middle panel) and individuals standard deviation (lower panel), for the overground group (OG, red) as well as the treadmill group (TM, blue). Higher scores on GPS / EVGS indicate a more ‘abnormal’ gait pattern, based on deviations of the group average. Note that EVGS was only calculated for children walking in the overground gait laboratory, based on video analysis.. Figure 4. Effect of incrementally averaging over 20 randomly selected strides for a subgroup of children walking in the treadmill (n=23). Results are visualized for effects on root mean square errors (RMSE) as a sum score for joint kinematics (left), moments (middle) and kinetics (right).. 37.

No results found