1. Gait & Posture 22 (2005) 1–9 www.elsevier.com/locate/gaitpost Reliability of upper and lower limb three-dimensional kinematics in children with hemiplegia Anna H. Mackeya, Sharon E. Waltb, Glenis A. Lobbc, N. Susan Stottd,* a University of Auckland Gait Laboratory, Tamaki Campus, Merton Road, Auckland, New Zealand b Department of Sport and Exercise Science, University of Auckland Gait Laboratory, Tamaki Campus, Merton Road, Auckland, New Zealand c Starship Children’s Hospital, Park Road, Grafton, Auckland, New Zealand d Department of Surgery, Faculty of Medicine and Health Sciences, University of Auckland, Room 3432, Park Road, Auckland, New Zealand Received 15 December 2003; accepted 1 June 2004Abstract The repeatability of both 3D kinematic measurements of arm movement during simple upper limb tasks and lower limb movement duringgait analysis was evaluated in 10 children with hemiplegic cerebral palsy. All tasks were completed on two separate occasions, 1 week apart.The 3D lower limb gait analysis showed high levels of repeatability in the sagittal plane measures, with mean coefﬁcient of multiplecorrelations (CMCs) greater than 0.92 between sessions. Transverse and frontal plane measures had mean CMCs greater than 0.7 betweensessions. A 3D analysis of shoulder and elbow ﬂexion/extension during the two functional tasks (hand to head and hand to mouth) was highlyrepeatable between sessions (mean CMCs, 0.87 to 0.95). Rotational measures at the shoulder and the elbow during the same tasksdemonstrated moderate levels of inter-sessional repeatability (mean CMCs shoulder 0.49 to 0.63; elbow 0.63 to 0.74). Overall, the lower limb3D kinematic analysis had similar levels of repeatability in both the hemiplegic limb and the limb with normal tone. The 3D kinematicanalysis of movement of the hemiplegic upper limb during simple upper limb tasks also had moderate to good repeatability, suggesting it maybe able to be used as an outcome measure in the hemiplegic upper limb.# 2004 Elsevier B.V. All rights reserved.Keywords: Kinematics; Hemiplegia; Paediatric; Three-dimensional analysis; Upper limb1. Introduction everyday functional tasks that involve reaching, grasping and manipulating objects [2–4]. The term cerebral palsy encompasses a broad group of Therapy and orthopaedic management are considered thedisorders of movement and posture resulting from a static mainstay of interventions for children with cerebral palsy.insult to the immature brain within the ﬁrst 2 years of life . Unfortunately one of the major problems in the managementChildren with hemiplegic cerebral palsy have varying com- of children with cerebral palsy is the paucity of reliable,binations of weakness, sensory loss, and spasticity, invol- valid and objective measures to monitor treatment out-ving the arm and leg on one side of the body . In this comes, particularly for the upper limb. Frequently usedgroup of children, treatment and intervention is often pri- clinical measures, such as passive range of motion [5,6],marily focused on the lower limb, to develop and improve measurement of muscle tone  and even classiﬁcation ofthe walking ability of the child, with intervention for the types of cerebral palsy have all shown poor reproducibilityupper limb dysfunction being secondary. However, the between observers and sessions, even when tested underupper limb dysfunction can be equally disabling, affecting standardised conditions . Two recently developed upper limb therapy measures, Quality Upper Extremity Skills Test * Corresponding author. Tel.: +64 9 3737599x82861; fax: +64 9 3677159. (QUEST) and Melbourne Assessment of Unilateral Upper E-mail address: email@example.com (N.S. Stott). Limb function have demonstrated high levels of reliability in0966-6362/$ – see front matter # 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.gaitpost.2004.06.002
2. 2 A.H. Mackey et al. / Gait & Posture 22 (2005) 1–9the cerebral palsy population [9,10]. However, both of these 2.2. Upper limb tasksmeasures rely on the therapist making a subjective visualassessment of the child’s range and quality of upper limb The upper limb tasks chosen were required to be bothmovement. simple and functional, to represent everyday upper limb In the lower limb, visual assessments of videoed gait are movement that would have been practised by the child manynot as accurate as 3D gait analysis in detecting and deﬁning times. The two upper limb tasks of taking the ‘hand to head’gait deviations in children with spastic diplegia [11–13]. and ‘hand to mouth’ met the above criteria and have bothThus, 3D kinematic gait analysis is considered the gold been described in previous studies of Rau et al.  and Rabstandard in movement analysis, providing clinicians with a et al. . In the ﬁrst task of ‘hand to head’ the child startsreliable and objective measurement tool that can be used to with the hand on the ipsilateral knee, and is asked to reachquantify changes in gait in children with cerebral palsy . and touch the top of their head and return the arm to theCurrently this technology is not routinely applied to upper ipsilateral knee. The second task of ‘hand to mouth’ had thelimb movement analysis. Potential problems encountered in same starting position, with the child asked to take their handobtaining upper limb kinematics include the large degrees of to their mouth and then return the hand to their knee.freedom at the shoulder; the complexities of deﬁning theshoulder joint centre using external markers and the lack of a 2.3. Upper limb modelcyclical movement task akin to gait in the lower limb[15,16]. However, several 3D upper limb kinematic models Twenty-one retro-reﬂective markers were placed on thehave recently been described in the literature [17,18]. Pre- child’s trunk and upper limbs to create a 3D mathematicalliminary assessments of 3D upper limb kinematic function model for the examination of upper limb movements (Fig. 1)have been made with two clinical populations, including the . We have previously published details of a comparisonassessment of movement limitations following burn scar of this upper limb model with goniometric measures with acontracture  and the assessment of a child with brachial strong correlation (r = 0.93) found between the mean elbowplexus palsy . angle measured with the goniometer and the same angle However, the use and reliability of 3D upper limb kine- measured by the 3D upper limb model . A similarmatics in the hemiplegia, cerebral palsy population has not comparison of frontal and sagittal plane measures at thebeen previously studied. The purpose of this study was shoulder has also shown a good correlation (r = 0.74)therefore to assess the repeatability of 3D joint kinematics between the goniometric measures and the 3D measuresfor the upper limb during gait and during two functional (unpublished laboratory data). The 3D upper limb modeltasks of taking hand to mouth and hand to head compared to consisted of seven segments, including right/left trunk; right/repeatability in the lower limb during gait for hemiplegic left upper arm; right/left forearm and pelvis (Table 1). Eachcerebral palsy. segment is assumed to be a rigid body deﬁned by three markers, generally representing proximal and distal ends of the segment plus a third non-collinear marker to allow for2. Methods rotational orientation . A joint coordinate system was implemented to describe relative angles between segments2.1. Subjects . The coordinate system deﬁning each segment is shown in Fig. 1, with joint ﬂexion-extension measured about the Ethical approval for the study was granted from the medial-lateral axis (y-axis); joint rotation about the long-Auckland Ethics Committee, New Zealand. Informed con- itudinal axis (x-axis) and the ﬁnal perpendicular axis (z-axis)sent was obtained from all participants and their guardians. determining abduction–adduction at the joint .The inclusion criteria included ambulatory children with a Upper limb joint centres, at the shoulder, elbow, wrist anddiagnosis of spastic hemiplegia, cerebral palsy aged between neck were deﬁned as virtual markers calculated from offsets5 and 16 years. Exclusion criteria included any other form of of two external marker positions (Table 1). In accordance withcerebral palsy or progressive spasticity; any casting or other upper limb models described in the literature [17,18],botulinum toxin A injections within the last 12 months; assumptions were made for the assessment of shoulder move-previous upper limb surgery; elbow ﬂexion contracture ment, with both scapulo-thoracic and acromion-calviculargreater than 208; lack of informed consent and any disabil- motions being discounted. The shoulder joint was thereforeities that would make it difﬁcult for the child to understand assumed to have only three degrees of freedom. Previousor cooperate fully with the study. Ten children with hemi- upper limb models have calculated the shoulder joint centreplegia (seven left and three right) were recruited from as an estimated offset from a single external marker on theorthopaedic and neurological clinic lists and local schools acromion [17,18]. For this upper limb model, the shoulderwith physical therapy units (six male subjects, mean age 9 Æ joint centre was calculated as the mid point between two4 years and four female subjects, mean age 12 Æ 3 years). external markers placed on the anterior and posterior aspect ofAll subjects were independent ambulators with a Gross the gleno-humeral joint (A1 and A2) (Fig. 1). Marker place-Motor Functional Classiﬁcation Scale of level I or II . ment was determined by palpation of bony landmarks, on the
3. A.H. Mackey et al. / Gait & Posture 22 (2005) 1–9 3Fig. 1. Schematic diagrams (frontal and lateral view) of the upper limb model used for three-dimensional kinematic analysis. (*): Retro-reﬂective externalmarkers placed on the skin. ( ): Virtual markers created at the neck; shoulder; elbow and wrist. A1: acromion back; A2: acromion front; C7: cervical spine 7;SN: sternal notch; UW: upper arm wand; LW: lower arm wand; EL: elbow lateral; EM: elbow medial; WL: wrist lateral; WM: wrist medial; S: sacral; ASIS:right and left anterior superior iliac spine. The insert of the dorsum scapulae details bony landmarks used for palpation to guide the placement of shouldermarkers (A1 and A2) .inferior aspect of the acromion, anteriorly and the posterior The elbow joint centre was deﬁned as the mid pointaspect of acromion, as it joins the spine of scapula, posteriorly between markers on the outside edge of the medial and(Fig. 1). The aim of using two bony landmarks to deﬁne lateral condyles of the humerus and the wrist joint centrethe shoulder joint centre was to standardise further the joint set as the mid point between medial and lateral wristcentre position in children of different ages, rather than joint markers [17,18]. A rotational wand was placed onestimating a set offset from one external marker. both the upper and lower arm segments of each arm. The upper arm wand was placed in the middle of the upper armTable 1 segment, in line with the shoulder joint centre proximallySegment deﬁnitions and joint centre (JC) offsets for the upper limb model and lateral epicondyle of the elbow distally. The forearmSegment Markers wand was placed on the distal third of the pronated forearm,Right upper arm Right Shoulder JC Right UW Right elbow JC in line with the lateral epicondyle of the elbow, proximallyRight forearm Right elbow JC Right LW Right wrist JC and wrist joint centre, distally. The distal placement of theRight trunk Neck JC Right shoulder JC C7 forearm wand was used to represent forearm rotation.Left upper arm Left shoulder JC Left UW Left elbow JC Marker location for this upper limb model differedLeft forearm Left elbow JC Left LW Left wrist JCLeft trunk Neck JC Left shoulder JC C7 slightly from those previously described by Schmidt et al.Pelvis Sacral Right ASIS Left ASIS ; Rau et al.  and Rab et al. . Rab et al. includedDeﬁning joint centres (JC) Markers (Fig. 1) external markers on the head and did not use rotational wand markers . An external head marker was not included inShoulder JC A1 A2 MidpointNeck JC C7 SN Midpoint this upper limb model, meaning that speciﬁc information onElbow JC EL EM Midpoint individual head motion was not obtained from this model.Wrist JC WL WM Midpoint However, markers placed on the trunk (C7 and sternum) canASIS: anterior superior iliac spine; UW: upper arm wand; LW: lower arm provide information on forward or posterior trunk lean,wand. which may occur during upper limb tasks. Rau et al. 
4. 4 A.H. Mackey et al. / Gait & Posture 22 (2005) 1–9and earlier work from colleagues Schmidt et al.  using the method described by Kadaba et al.  anddescribed an upper limb model with marker triads on Steinwender et al. . The adjusted R-squared (R2 ) statistic ahand, forearm and upper arm segments, with additional or coefﬁcient of determination was used to assess themarkers placed at the elbow and wrist joint to determine similarity between two representative waveforms forjoint centres. Marker triads were not utilised for our model, within-session analysis, and between mean waveforms foras preliminary testing in our laboratory with paediatric between-session analysis. Similarity in the waveformssubjects found that they prevented full comfortable elbow results in the adjusted R-squared value tending towards 1;movement. dissimilar waveforms result in the adjusted R-square tending towards 0. The positive square root of the adjusted R-squared2.4. Testing procedures value, being the r value, or adjusted coefﬁcient of multiple correlation (CMC), is reported in the text. Comparisons were Testing was performed in the University of Auckland made in all three planes, frontal, sagittal and transverse andGait Laboratory on two occasions, 1 week apart. Children at seven anatomical levels, ankle, knee, hip, pelvis, trunk,were seen on the same day and time, 1 week apart, to shoulder and elbow.minimise environmental changes that might affect range For within-session analysis, we compared the third andof movement or dynamic muscle tone. The same testing fourth walking trials collected during the gait analysis andprotocol was used in both sessions. The 21 markers the ﬁnal two repetitions of the upper limb tasks. For thedescribed above in the upper limb model were ﬁrst applied between-session analysis, the mean waveform of the fourto the subject’s upper limbs. Each child then completed two walking trials from week 1 was compared to the meanupper limb functional tasks in sitting (hand to mouth and waveform of the four walking trials at week 2. Correspond-hand to head). To perform these tasks the children were ingly for the upper limb tasks between session analyses, theseated on a stool in the centre of the video capture volume mean of the three repetitions from week 1 was compared toarea. A small box was placed in front of the children to rest the mean of the three task repetitions in week 2. The meantheir feet on and ensure they felt secure sitting on the stool. values were utilised for between-session analyses due to theIn an attempt to standardise the upper limb movement, the small variance found within one session.therapist stood in front of the child demonstrating the task to For the 3D gait analysis, the mean absolute differencebe performed. Each task was carried out three times in a both within and between sessions was determined for stepsingle session. A 3D gait analysis was then undertaken, length, stride length, cadence and forward velocity for eachusing a Cleveland clinic marker set (OrthoTrak 4.2 Refer- subject. The mean absolute difference (degrees) between theence Manual, MotionAnalysis Corporation, Santa Rosa, CA, gait kinematic graphs at week 1 and at week 2 was alsoUSA) to examine lower limb kinematics during gait. The 21 calculated for each subject. This additional informationmarkers applied to the upper limbs were left in place to highlights the testing variation that can be expected fromobtain a 3D analysis of the arm swing during gait. The child week to week, and is useful for the interpretation of gaitwas asked to walk at a self-selected speed along a 10-m analysis results. This information was not determined for thewalkway. At least four walking trials were collected for two upper limb tasks as the starting position for the uppereach participant. All data were collected with an 8-camera limb tasks had not been fully constrained, resulting inMotionAnalysis video system at 60 Hz (MotionAnalysis variation in the self-selected starting position between ses-Corporation). sions in some subjects, particularly at the shoulder joint. The repeatability of the pattern of movement of these tasks was2.5. Data analysis the main interest of this study, which was best represented by the previously described CMC statistic. Processing of the 3D gait analysis data was completed withEvA software version 6.15 and OrthoTrak software version5.1 (MotionAnalysis Corporation). OrthoTrak gait analysis 3. Resultssoftware produces graphs normalised to the gait cycle. Theadditional markers used to evaluate the 3D position of the arm 3.1. Temporal-spatial parametersduring gait required separate analysis using KinTrak softwareversion 6 (MotionAnalysis Corporation). Kinematic infor- Kinematic gait patterns may be affected by walkingmation for the two upper limb tasks in sitting was collected velocity . We therefore assessed the inter-sessional meanusing EvA software version 6.15 and analyzed using KinTrak absolute differences in the temporal-spatial parameters tosoftware version 6 (MotionAnalysis Corporation). deﬁne the level of repeatability between walking trials. The highest variability was found for walking velocity, with a2.6. Statistical analysis mean absolute difference of 6.8 cm/s within session and 10.3 cm/s between sessions. The lowest variability was Statistical analysis of the repeatability of the pattern of found for step length, both in the affected and unaffectedupper and lower limb movement waveforms was carried out limb (Table 2).
5. A.H. Mackey et al. / Gait & Posture 22 (2005) 1–9 5Table 2Mean absolute difference in temporal-spatial parameters within and between sessionsTemporal/spatial parameters Within session Between sessions Mean absolute difference (S.D.) Range Mean absolute difference (S.D.) RangeStep length (cm) Affected 2.7 (2.3) 1.0–7.3 2.3 (1.1) 1.0–4.4 Unaffected 2.5 (2.2) 0.0–7.4 2.0 (1.6) 0.2–6.0 Stride length (cm) 4.9 (4.1) 0.0–12.4 3.8 (2.6) 1.0–9.0 Cadence (steps/min) 4.2 (2.9) 0.0–9.5 8.9 (11.3) 0.1–35.0 Velocity (cm/s) 6.8 (6.3) 0.0–20.0 10.3 (11.3) 0.5–34.0Table 3Mean adjusted coefﬁcient of multiple correlation (CMC) for lower limb measures during gait analysisKinematics Within session CMC (S.D.) Between session CMC (S.D.) Normal limb Hemiplegic limb Normal limb Hemiplegic limbSagittal plane Ankle dorsi/plantar ﬂexion 0.96 (0.0) 0.98 (0.0) 0.98 (0.0) 0.96 (0.0) Knee ﬂexion/extension 0.99 (0.0) 0.98 (0.0) 0.99 (0.0) 0.99 (0.0) Hip ﬂexion/extension 0.98 (0.0) 0.99 (0.0) 0.99 (0.0) 0.99 (0.0) Pelvic tilt 0.88 (0.2) 0.82 (0.3) 0.92 (0.2) 0.93 (0.1) Trunk tilt 0.84 (0.2) 0.82 (0.2) 0.92 (0.1) 0.94 (0.1)Frontal plane Foot progression 0.87 (0.2) 0.89 (0.2) 0.76 (0.2) 0.81 (0.3) Knee varus/valgus 0.86 (0.2) 0.89 (0.2) 0.70 (0.2) 0.85 (0.1) Hip abduction/adduction 0.93 (0.1) 0.92 (0.2) 0.91 (0.1) 0.95 (0.1) Pelvic obliquity 0.95 (0.1) 0.88 (0.2) 0.92 (0.2) 0.91 (0.2) Trunk obliquity 0.86 (0.2) 0.91 (0.1) 0.93 (0.1) 0.89 (0.1)Transverse plane Foot rotation 0.89 (0.1) 0.92 (0.1) 0.91 (0.1) 0.92 (0.1) Knee rotation 0.91 (0.1) 0.89 (0.2) 0.80 (0.1) 0.82 (0.2) Hip rotation 0.89 (0.1) 0.88 (0.2) 0.78 (0.2) 0.86 (0.1) Pelvis rotation 0.91 (0.1) 0.90 (0.1) 0.95 (0.1) 0.87 (0.1) Trunk rotation 0.88 (0.2) 0.81 (0.3) 0.97 (0.0) 0.83 (0.3)CMC: coefﬁcient of multiple correlation.3.2. Lower limb kinematics normal limb and the hemiplegic limb (Fig. 2). The frontal and transverse plane measures had slightly lower levels of High levels of repeatability were observed across all three repeatability; however the mean CMC was still 0.7 or greaterplanes in the lower limb (Table 3). Sagittal plane kinematics for these measures. The lowest CMCs between sessionsat the hip, knee and ankle were the most reliable, with CMC were found for foot progression and knee varus/valgusvalues of 0.96–0.99 within and between sessions for both the position in the frontal plane and correspondingly for hipFig. 2. Representative frontal, sagittal and transverse plane kinematic graphs of hip range of motion during gait in a child with hemiplegia, derived from datacollected at time 0 (session one) and 1 week later (session two). The x-axis represents 0–100% of the gait cycle, beginning with Right Heel Strike (RHS).Laboratory normative data is shown as a broad grey band, while the solid line indicates data from session one and the dotted line indicates data from session two,1 week later. The waveforms are very similar for the two testing sessions (LTO, Left Toe Off; LHS, Left Heel Strike; RTO, Right Toe Off).
6. 6 A.H. Mackey et al. / Gait & Posture 22 (2005) 1–9Table 4 Table 4. 3D kinematics in the frontal and sagittal plane wereMean absolute differences in lower limb kinematics between sessions most repeatable with mean absolute differences of fourKinematics Mean absolute differences in degrees (S.D.) degrees or less from week 1 to week 2. Normal limb Hemiplegic limbSagittal plane 3.3. Upper limb position during gait Ankle dorsi/plantar ﬂexion 2 (1) 1 (0) Knee ﬂexion/extension 4 (2) 2 (1) A 3D analysis of the upper limb position in gait found Hip ﬂexion/extension 4 (1) 3 (1) moderate repeatability both within and between sessions Pelvic tilt 2 (1) 2 (1) Trunk tilt 2 (1) 3 (1) (Table 5). Repeatability was highest at the shoulder in the sagittal and frontal planes with CMC values of 0.52 to 0.87.Frontal plane Foot progression 1 (0) 1 (0) Transverse plane elbow supination and pronation had the Knee varus/valgus 4 (2) 2 (1) lowest repeatability within and between sessions, with CMC Hip abduction/adduction 3 (1) 3 (1) values of 0.42 to 0.61. The repeatability of the elbow ﬂexion Pelvic obliquity 2 (1) 2 (1) position during gait was similar for the affected (CMC, 0.58) Trunk obliquity 3 (1) 3 (1) and unaffected upper limb (CMC, 0.60) between sessions.Transverse plane Foot rotation 8 (1) 5 (1) 3.4. Upper limb kinematics—hand to mouth and hand to Knee rotation 10 (3) 7 (2) head tasks Hip rotation 7 (2) 5 (2) Pelvis rotation 3 (1) 3 (2) Trunk rotation 3 (1) 4 (2) Simple hand to head and hand to mouth tasks were used to assess the repeatability of a simple motor task in the upperTable 5Mean adjusted coefﬁcient of multiple correlation for upper limb position during gaitKinematics Within Session CMC (S.D.) Between Session CMC (S.D.) Unaffected limb Affected limb Unaffected limb Affected limbSagittal plane Elbow ﬂexion 0.58 (0.3) 0.60 (0.2) 0.60 (0.3) 0.58 (0.3) Shoulder ﬂexion 0.80 (0.2) 0.67 (0.3) 0.87 (0.1) 0.68 (0.3)Frontal plane Shoulder abduction/adduction 0.52 (0.3) 0.65 (0.3) 0.66 (0.3) 0.54 (0.3)Transverse plane Elbow supination/pronation 0.48 (0.2) 0.61 (0.3) 0.49 (0.3) 0.42 (0.3)Shoulder rotation 0.58 (03) 0.67 (0.3) 0.57 (0.3) 0.57 (0.2)CMC: coefﬁcient of multiple correlation.and knee rotation in the transverse plane. Similar values of limb (Fig. 3). Table 6 shows the mean and standard deviationrepeatability were found for both the normal and hemiplegic of the CMC values comparing waveforms within andlimb, and in the within-day and between-day variability. The between sessions. Moderate to high levels of repeatabilitymean absolute differences, in degrees, between gait kine- were found for the upper limb kinematics of these twomatic graphs from session one to session two are shown in functional tasks. Sagittal plane elbow and shoulder kine- Elbow Flexion / Extension Elbow Supination / Pronation 60 Joint Angle (degrees) 160 Joint Angle (degrees) Sup / Pron 120 40 Ext / Flex 80 20 40 0 0 0 100 0 100Fig. 3. Representative kinematic graphs of elbow ﬂexion/extension and forearm supination/pronation for one subject carrying out hand to mouth task using thehemiplegic arm. The x-axis represents time, normalised to 100% from the start to the end of the movement. Three repetitions of elbow movement during thesame testing session are shown with the solid line representing the ﬁrst trial and the two dotted lines indicating the second and third trials. Although the pattern ofmovement is similar in the three trials, there are changes in the starting point between the three trials.
7. A.H. Mackey et al. / Gait & Posture 22 (2005) 1–9 7Table 6Mean adjusted coefﬁcient of multiple correlation (CMC) for affected upper limb movements during hand to head and hand to mouth taskKinematics Task 1: Hand to head Task 2: Hand to mouth Within session CMC (S.D.) Between session Within session CMC (S.D.) Between session CMC (S.D.) CMC (S.D.) Session 1 Session 2 Session 1 Session 2Sagittal plane Elbow ﬂexion/extension 0.96 (0.0) 0.94 (0.0) 0.92 (0.1) 0.93 (0.1) 0.96 (0.1) 0.95 (0.0) Shoulder ﬂexion/extension 0.95 (0.0) 0.90 (0.1) 0.87 (0.2) 0.90 (0.1) 0.94 (0.1) 0.91 (0.0)Frontal plane Shoulder abduction/adduction 0.71 (0.2) 0.77 (0.1) 0.76 (0.2) 0.74 (0.2) 0.74 (0.1) 0.62 (0.3)Transverse plane Elbow supination/pronation 0.90 (0.1) 0.88 (0.1) 0.63 (0.3) 0.76 (0.2) 0.90 (0.1) 0.74 (0.3) Shoulder rotation 0.72 (0.2) 0.63 (0.3) 0.63 (0.3) 0.67 (0.3) 0.72 (0.2) 0.49 (0.3)CMC: coefﬁcient of multiple correlation.matics had the highest levels of repeatability for both tasks and extension. Three-dimensional upper limb kinematic ana-within and between sessions, with CMC values ranging from lysis has been used by several pilot studies as an assessment0.87 to 0.96. However, some variation was seen in the tool, particularly for burns patients and a patient with brachialstarting position of the arm when the task was repeated plexus palsy [15,17–19]. However, this is the ﬁrst reportedboth within and between sessions (see Fig. 3). Lower levels study to assess the reliability of 3D upper limb kinematics in aof repeatability were found for frontal and transverse plane paediatric population with hemiplegia, cerebral palsy.kinematics, with shoulder rotation having the lowest values We found a high degree of reproducibility in the patternof repeatability in both tasks with CMC values of 0.49–0.63. of sagittal plane movement during hand to head and hand to mouth tasks across the two testing sessions, suggesting that 3D kinematic analysis can be used reliably to assess sagittal4. Discussion plane motion at the elbow and shoulder in an individual carrying out such a task. However, there was some varia- In this study we were interested in the reliability of 3D bility in the starting position of the arm used to achieve thekinematic analysis in the upper limb compared to 3D task, both between subjects and between sessions. This iskinematic analysis of lower limbs in children with hemi- similar to the data found by Rab et al. which showedplegia, cerebral palsy, to understand better the usefulness of between subject standard deviations of up to 25 degrees3D kinematic analysis as an outcome measure in the hemi- for elbow and shoulder motion in normal children during aplegic upper limb. As expected, the lower limb 3D kine- hand to head task . Greater standardisation of both thematics had excellent repeatability, both within and between starting and the ﬁnishing points for the task would besessions. Similar to previous studies [14,24], the repeatabil- required to improve the repeatability of the starting andity was highest in the sagittal plane and lower in the frontal ending points of this type of movement.plane (foot and knee) and transverse plane (hip and knee). The repeatability of upper limb motion in the frontal andSmall changes in marker placement at the knee and ankle transverse planes was lower than that in the sagittal plane.can result in signiﬁcant changes in transverse plane kine- Even within one session, the repeatability was less suggest-matics , and this may have accounted for the lower ing that this was not related to variations in marker place-repeatability in the transverse plane seen in this study. We ment or camera position but rather to differences in patternsfound that 3D kinematic analysis of arm movement during of movements between each trial. In the lower limb, thesimple functional tasks also had moderate to high levels of motor strategies that can be used to complete a task arerepeatability within and between measurement sessions, limited. In contrast, a large number of different motorparticularly in the sagittal plane. Surprisingly, arm swing strategies can be used to achieve the same motor task induring gait had only moderate levels of repeatability in both the upper limb. Bernstein has described the anatomicalthe normal limb and the hemiplegic limb. redundancy present in the upper limb as the ‘degrees of Two-dimensional (2D) forms of movement analysis have freedom’ or ‘motor equivalence’ problem . The largebeen widely used to investigate the motor control strategies number of degrees of freedom in the upper limb exceeds thatinvolved in reaching and grasping in both normal paediatric required for performance of a task and thus the position ofpopulations [26,27] and in the paediatric hemiplegic popula- the hand in space can be determined by an inﬁnite number oftion [28–30]. However, the kinematic information obtained joint angles [32–34]. The explanation for this joint redun-from these studies is limited to sagittal plane elbow motion dancy is not fully understood, but it has been proposed thatonly, with measures more accurately reﬂecting relative posi- this ﬂexibility permits a better response to random changestion of arm and forearm segments rather than elbow ﬂexion in the ﬁnal target position .
8. 8 A.H. Mackey et al. / Gait & Posture 22 (2005) 1–9 During gait, the arm swings reciprocally to act as a  Autti-Ramo I, Larsen A, Taimo A, von Wendt L. Management of thecounter force to minimise the rotational displacement of upper limb with botulinum toxin type A in children with spastic type cerebral palsy and acquired brain injury: clinical implications. Eur Jthe body during gait . Considerable variation has been Neurol 2001;8(Suppl 5):136–44.shown amongst individuals in the amount of elbow and  Fehlings D, Rang M, Glazier J, Steele C. An evaluation of botulinum-shoulder ﬂexion and extension used, with an increased A toxin injections to improve upper extremity function in childrenwalking velocity leading to increased arc of motion at the with hemiplegic cerebral palsy. J Pediatr 2000;137:331–7.  LaStayo PC, Wheeler DL. Reliability of passive wrist ﬂexion andshoulder and elbow . The posturing of the hemiplegic extension goniometric measurements: a multicenter study. Phys Therarm into an increased ﬂexion or high guard pattern during 1994;74:162–76.gait is well known but not often recorded in gait analysis  Harris S, Harthun Smith LihL, Krukowski L. Goniometric reliability. We found only moderate levels of repeatability for the for a child with spastic quadriplegia. J Pediatr Orthop 1985;5:shoulder position during gait, with slightly lower CMC 348–51.values for the elbow position. Closer analysis of our data  Pandyan A, Johnson G, Price C, Curless R, Barnes M, Rodgers H. A review of the properties and limitations of the Ashworth and modiﬁedsuggested that one reason for the lower CMC values could be Ashworth Scales as measures of spasticity. Clin Rehabil 1999;13:due to the reduced range of motion of the elbow joint during 373–83.gait compared to the shoulder. Such a problem has been  Blair E, Stanley F. Interobserver agreement in the classiﬁcation ofnoted before with this type of statistical analysis . cerebral palsy. Dev Med Child Neurol 1985;27:615–22.Kadaba et al. found that pelvic tilt measures had lower  Randall M, Carlin JB, Chondros P, Reddihough D. Reliability of the Melbourne assessment of unilateral upper limb function. Dev MedCMC values than, for example, sagittal plane knee joint Child Neurol 2001;43:761–7.motion due to both the small range of pelvic tilt measures  DeMatteo C, Law M, Russell D, Pollock N, Rosenbaum P, Walter S.recorded and the lack of a well deﬁned pattern of movement The reliability and validity of the Quality of Upper Extremity Skills. Test. Pediatr Phys Occup Ther 1993;13(2):1–18.  Mackey AH, Lobb GL, Walt SE, Stott NS. Reliability and validity of In summary, 3D kinematic analysis of upper limb move- the Observational Gait Scale in children with spastic diplegia. Devments during a simple task can have moderate to high Med Child Neurol 2003;45:4–11.repeatability between sessions in the hemiplegic population.  Krebs DE, Edelstein JE, Fishman S. Reliability of observationalSimilar to the lower limb, the repeatability is highest in the kinematic gait analysis. Phys Ther 1985;65:1027–33.sagittal plane and lower in the transverse and frontal planes.  Eastlack ME, Arvidson J, Snydner-Mackler L, Danoff JV, McGarveyThis level of repeatability suggests that 3D kinematic CL. Interrater reliability of videotaped observational gait-analysis assessments. Phys Ther 1991;71:465–72.analysis may be able to be used as an outcome measure,  Steinwender G, Saraph S, Scheiber S, Zwick EB, Uitz C. Intrasubjectat least in measuring some aspects of a simple upper limb repeatability of gait analysis in normal and spastic children. Clintask such as elbow ﬂexion/extension or shoulder ﬂexion/ Biomech 2000;15:134–9.extension. Clearly this is only a small patient population  Rau G, Disselhorst-Klug C, Schmidt R. Movement biomechanics goes upwards: from the leg to the arm. J Biomech 2000;33:1207–16.and such ﬁndings need to be validated by further studies  van der Helm FC... A ﬁnite element musculoskeletal model of thein different patient groups and different functional tasks. The shoulder mechanism. J Biomech 1994;27:551–69.sensitivity of such a measure to change caused by a speciﬁed  Schmidt R, Disselhorst-Klug C, Silny J, Rau G. A marker-basedintervention is also not known and remains to be studied. measurement procedure for unconstrained wrist and elbow motions. J Biomech 1999;32:615–21.  Rab G, Petuskey K, Bagley A. A method for determination of upper extremity kinematics. Gait Posture 2002;15:113–9.Acknowledgements  Palmieri TL, Petuskey K, Bagley A, Takashiba S, Greenhalgh DG, Rab GT. Alterations in functional movement after axillary burn scar contracture: a motion analysis study. J Burn Care Rehabil The New Zealand Neurological Foundation and the 2003;24:104–8.School of Medicine Foundation, University of Auckland  Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi BE.assisted with the funding of this study. We would like to Development and reliability of a system to classify gross motorthank the children and families who participated in this function in children with cerebral palsy. Dev Med Child Neurolstudy; Joanna Stewart from the Health Research Council 1997;39:214–23.  Mackey AH, Walt SE, Lobb G, Stott NS. Intra-observer reliability ofBiostatistics Unit at the University of Auckland; Nicola the modiﬁed Tardieu scale in the upper limb of children with hemi-Reynolds for her assistance with data collection; and Chris- plegia. Dev Med Child Neurol 2004;46:267–72.tine Ganly for secretarial support in preparing this paper.  Nigg BM, Cole GK, Wright IC. Optical methods. In: Nigg BM, Herzog W, editors. In: Biomechanics of the musculo-skeletal system. Chichester: John Wiley & Sons Ltd.; 1999. p. 302–31.  Grood ES, Suntay WJ. A joint coordinate system for the clinicalReferences description of three-dimensional motions: application to the knee. J Biomech Eng 1983;105:136–44.  Rang M. Cerebral palsy. In: Morrissy RT, editor. In: Pediatric ortho-  Kadaba MP, Ramakrishnan HK, Wootten ME, Gainey J, Gorton G, paedics. Philadelphia: J.B. Lippincott; 1990. p. 465–506. Cochran GVB. Repeatability of kinematic, kinetic and electromyo-  Boyd R, Morris ME, Graham HK. Management of upper limb graphic data in normal adult gait. J Orthop Res 1989;7:849–60. dysfunction in children with cerebral palsy: a systematic review.  van der Linden ML, Alison MK, Hazelwood ME, Hillman SJ, Robb Eur J Neurol 2001;8(Suppl 5):150–67. JE. Kinematic and kinetic gait characteristics of normal children
9. A.H. Mackey et al. / Gait & Posture 22 (2005) 1–9 9 walking at a range of clinically relevant speeds. J Pediatr Orthop  Bernstein N. The coordination and regulation of movements. Oxford: 2002;22:800–6. Pergamon; 1967. Konczak J, Dichgans J. The development toward stereotypic arm  Grea H, Desmurget M, Prablanc C. Postural invariance in three- kinematics during reaching in the ﬁrst 3 years of life. Exp Brain dimensional reaching and grasping movements. Exp Brain Res Res 1997;117:346–54. 2000;134:155–62. Schneiberg S, Sveistrup H, McFaydyen BJ, McKinley P, Levin M. The  Gielen CCAM. van Bolhuis BM, Theeuwen M. On the control of development of coordination for reach-to-grasp movements in chil- biologically and kinematically redundant manipulators. Hum Mov Sci dren. Exp Brain Res 2002;146:142–54. 1995;14:487–509. Hurvitz EA, Conti GE, Brown SH. Changes in movement character-  Robertson EM, Miall RC. Multi-joint limbs permit a ﬂexible response istics of the spastic upper extremity after botulinum toxin injection. to unpredictable events. Exp Brain Res 1997;117:148–52. Arch Phys Med Rehabil 2003;84:444–54.  Perry J. Gait analysis: normal and pathological function. New Jersey: Kluzik J, Fetters L, Coryell J. Quantiﬁcation of control: a preliminary Thorofare; 1992. study of effects of neurodevelopmental treatment on reaching in  Carmick J. Use of neuromuscular electrical stimulation and a dorsal children with spastic cerebral palsy. Phys Ther 1990;70:65–78. wrist splint to improve the hand function of a child with spastic Steenbergen B, van Thiel E, Hulstijn W, Meulenbroek R. The coor- hemiparesis. Phys Ther 1997;77:661–71. dination of reaching and grasping in spastic hemiparesis. Hum Mov  Anderson JE, editor. In: Grant’s atlas of anatomy. Baltimore: Williams Sci 2000;19:75–105. & Wilkins; 1978.