- Email: [email protected]
Local dynamic stability during treadmill walking can detect children with developmental coordination disorder
Gait & Posture 59 (2018) 99–103
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Full length article
Local dynamic stability during treadmill walking can detect children with developmental coordination disorder
MARK
⁎
Merete B. Speedtsberga,b, , Sofie B. Christensenb, Jan Stenumb,e, Thomas Kallemosef, Jesper Benckea, Derek J. Curtisa,d, Bente R. Jensenb,c a
Laboratory of Human Movement Analysis, Department of Orthopaedic Surgery, Copenhagen University Hospital Hvidovre, Copenhagen, Denmark Biomechanics and Motor Control Lab., Integrated Physiology, Department of Nutrition, Exercise and Sport, University of Copenhagen, Copenhagen, Denmark c Department of Neurology, Odense University Hospital, University of Southern Denmark, Odense, Denmark d Department of Physical and Occupational Therapy, Metropolitan University College, Copenhagen, Denmark e Locomotion Neuromechanics Laboratory, Department of Kinesiology, University of Massachusetts Amherst, Amherst, MA, USA f Clinical Orthopaedic Research Hvidovre, Department of Orthopaedic Surgery, Copenhagen University Hospital Hvidovre, Copenhagen, Denmark b
A R T I C L E I N F O
A B S T R A C T
Keywords: Developmental coordination disorder Local dynamic stability Gait stability Treadmill walking Trunk accelerations
Objective: Developmental coordination disorder (DCD) is an innate impairment of motor coordination that affects basic locomotion and balance. This study investigated local dynamic stability of trunk accelerations during treadmill walking as an objective evaluation of gait stability and the sensitivity and specificity of this measure to discriminate children with DCD from typically developing children. Method: Eight children with DCD and ten age- and gender-matched typically developing children (TD) walked four minutes on a treadmill. Trunk accelerations in vertical, medio-lateral and anterior-posterior directions were recorded with a sternum mounted accelerometer at 256 Hz. Short term local dynamic stability (λs), root mean square (RMS) and relative root mean square (RMSR) were calculated from measures of orthogonal trunk accelerations. Receiver operating characteristic curve (ROC) analysis was performed to discriminate between groups based on short term local dynamic stability. Results: λs was significantly greater in children with DCD in the main movement direction (AP) (DCD: 1.69 ± 0.17 λs; TD:1.41 ± 0.17 λs; p = 0.005), indicating reduced local dynamic stability. RMS and RMSR accelerations showed no difference between children with DCD and TD children in any direction. The ROC analysis of λs in separate directions and in two dimensions showed an excellent accuracy of discriminating between children with DCD and TD children. Anterior-posterior direction in combination with medio-lateral or vertical showed best performance with an area under the curve (AUC) of 0.91. Conclusion: We have shown that children with developmental coordination disorder have general reduced local dynamic stability and that the short term Lyapunov exponent has good power of discrimination between DCD and TD.
1. Introduction Developmental coordination disorder (DCD) is an innate impairment of motor coordination that may affect basic locomotion, balance and acquisition of motor skills. Due to great variability of motor performance between children with DCD [1] and diagnostic criteria that leave room for clinical interpretation [2], DCD is under-recognised and difficult to diagnose by health care professionals [3]. Standardised norm-referenced tests are available to assess motor function (e.g. mABC-2 & BOTMP), but they are mainly skill-based and may be influenced by experience in the specific task [4]. In summary, there could be
a need for objective measures to identify children with DCD based on deficits in basic locomotion. Gait is a basic, highly automatised motor function and less affected by experience, gender and movement culture. As instability in gait may play an important role in the motor impairments associated with DCD [5], objective evaluation of gait stability could be a relevant addition to the diagnostic toolbox and an important method to evaluate improvements in proprioception and functional motor control. Prior research on gait function in children with DCD has mainly been focused on identifying characteristics and variability in discrete spatiotemporal parameters and kinematics. Even though children with
⁎ Corresponding author at: Laboratory of Human Movement Analysis (247), Department of Orthopaedic Surgery (333), Hvidovre University Hospital, Kettegård Allé 30, 2650 Hvidovre, Denmark. E-mail address: [email protected] (M.B. Speedtsberg).
http://dx.doi.org/10.1016/j.gaitpost.2017.09.035 Received 10 January 2017; Received in revised form 10 August 2017; Accepted 27 September 2017 0966-6362/ © 2017 Published by Elsevier B.V.
Gait & Posture 59 (2018) 99–103
M.B. Speedtsberg et al.
2.2. Procedure and data acquisition
DCD exhibit a more asymmetrical and variable kinematic gait pattern [6], and a greater stride-to-stride variability [7] (normally interpreted as an expression of impaired gait control), Woodruff and colleagues concluded in their study on classification of gait patterns in children with DCD, that there was no solid systematic gait pattern explaining the gait deviation from typically developing children [8]. It remains unclear, however, how the previously reported discrete measures of variability relate to time-dependent measures of stability [9] in children with DCD. In the present study we decided to investigate local dynamic stability of trunk accelerations during treadmill walking. Local dynamic stability is a method of quantifying the body’s resilience to small perturbations naturally inherent during walking [10,11]. It is quantified over numerous sequential steps, and is a measure of how well a person responds to their internal variability and the environment’s external variability (e.g., a slippery or uneven walking surface). If small perturbations are not appropriately attenuated, the deviations from the normal gait pattern will accumulate. During gait, proprioceptive, vestibular and visuo-motor feedback contribute to an online control to attenuate the natural kinematic and kinetic variability [12,13]. This continuous processing and response requires an efficient, online motor adjustment, sensory organization and efficient neuromuscular response. These are mechanisms that are commonly reported to be impaired in children with DCD [5]. If children with DCD respond less efficiently to their naturally occurring, but increased variability, their local dynamic stability will be reduced. This may be reflected in reduced stability during gait or even increased risk of falling [9,10]. We therefore hypothesised that children with DCD have lower local dynamic stability during gait than typically developing children (TD). To establish clinical relevance, we further explored whether local dynamic stability has the sensitivity and specificity to discriminate between children with DCD and TD children.
Trunk accelerations were recorded with a sternum mounted accelerometer (MQ16, MarqMedical, Denmark) during four minutes walking at preferred walking speed. The test was performed on treadmill to control for walking speed, and each participant was accustomed to treadmill walking for maximum five minutes at different speeds comfortable to the child. Familiarisation ended at four min. if the child felt comfortable. Relatively short familiarisation was chosen to ensure full attentional focus during testing. Preferred walking speed was determined according to the recognised protocol by Dingwell and Marin [11,16], in which preferred walking speed is selected 6 times by the subject while alternating between increasing and decreasing speeds from 0.42 m/s above or below the previously selected speed. The preferred walking speed is subsequently calculated as the average of the six self-selected speeds. The participants were instructed, to walk without support from treadmill handles and were informed of time progress every minute. Upper body accelerations in vertical (VT), medio-lateral (ML) and anteror-posterior (AP) directions were sampled at 256 Hz (MQ16, Marq Medical, Denmark). Recordings began when the treadmill reached a constant speed and the participant was comfortable with the selected speed. 2.3. Data analysis All data processing and calculations were performed using customised MATLAB (The MathWorks, Inc., Natick, MA, USA) scripts [11]. Accelerometer based assessment of gait parameters has been validated in children [17]. The primary outcome was short term local dynamic stability in each VT, ML and AP directions. Root mean square was calculated for each direction of acceleration as a measure of variability. The method for calculating local dynamic stability has been described extensively in the literature [10,11,16]. Local dynamic stability was quantified using nonlinear time-series methods, and expressed as the short term Lyapunov exponent (λs) which has shown good reliability, good intrasession and fair intersession repeatability and a correlation to falls risk [10,18–20]. The first 195 strides were identified from the vertical accelerations, and the unfiltered time-series were time-normalised to 19,500 data points using cubic spline interpolation. Five-dimensional state spaces were reconstructed from each acceleration direction using the method of delays [21]. From the reconstructed state spaces Euclidean distances between nearest neighbours in state space were calculated as a function of time and averaged over all nearest neighbours. λs was calculated as the average rate of logarithmic divergence of the distance between nearest neighbours in state space from 0 to 0.5 stride (ln(div) > /stride-time from 0–0.5 stride) [11,22]. A higher λs expresses lower local dynamic stability as the rate of divergence in state space is faster. Acceleration Root Mean Square (RMS) was calculated as the dispersion of the acceleration data relative to zero and quantifies the average magnitude of accelerations in each direction during a complete walking trial [23]. To normalise RMS for effect of walking speed, the ratios between RMS accelleration in each direction and the RMS vector magnitude (RMSR) were calculated [24].
2. Method 2.1. Participants Eight children with DCD and ten age-, gender- and anthropometrically matched TD children with normal motor proficiency were recruited (Table 1). The DCD group was recruited through paediatric physiotherapists, occupational therapists and the Danish Parents Association for DCD. Motor performance was assessed using the Movement Assessment Battery for children (MABC-2), which is a norm-referenced basic motor abilities assessment tool containing fine motor, ball handling, and balance tasks [14]. Inclusion criteria for children with DCD were a score below the 15th percentile in the MABC-2 and to meet the official criteria for DCD according to the DSM-V [15] based on evaluation by a qualified health care professional. To avoid confounding factors, we did not include children with co-morbidities. Inclusion criteria for the TD-children were a score above the 16th percentile in the MABC-2, and no diagnosed or suspected neurodevelopment disorders. Informed written consent was obtained from a legal guardian. The experimental protocol was approved by the Capital Region Committee on Health Research Ethics, Denmark (ref.: H-4-2013-144).
2.4. Statistical analysis The distribution of the data was analysed for normality using the Shapiro-Wilk test and Q-Q plots in IBM SPSS statistics 22 (IBM Corp. Released 2013. IBM SPSS Statistics for Windows). Non-parametric (Mann–Whitney U) tests were used to test for group differences in age, height, body mass and MABC-2 percentile. A chi-squared test was used to evaluate sex differences between groups. Non-parametric (Mann–Whitney U) tests were performed to compare the short term λs, RMS and RMSR between children with DCD and TD children in separate directions of acceleration. The discriminative power of λs to
Table 1 Participant Characteristics (mean ± SD).
Age, years MABC-2 percentile Gender (♂/♀), n Height, cm Body mass, kg
DCD group (n = 8)
TD group (n = 10)
p-value
8.8 ± 1.5 2.6 ± 3.93 6/2 139.5 ± 8.1 33.6 ± 7.3
9.1 ± 1.4 73.3 ± 5.2 7/3 141.1 ± 3.0 33.7 ± 1.8
0.885 < 0.001 0.658 0.763 0.839
100
Gait & Posture 59 (2018) 99–103
M.B. Speedtsberg et al.
differentiate children with DCD and TD was assessed using receiver operating characteristic curve (ROC) analysis based on logistic regression. Area under the curve (AUC) and the optimal cut off were calculated, based on Youden’s index [25]. The ROC analyses were performed on separate directions of acceleration and also two-dimensionally taking two directions of acceleration and interaction between them into account. Since the cut off values of the variables in a two-dimensional ROC analyses are interdependent, a cutoff value for one of the variables can only be calculated given the value of the other variable and vice versa. As such a single value for each variable can’t be specified, rather a spectrum of values given by the value of the other variable. Because of this the cutoff values for the 2 dimensional ROC analyses are represented graphically. ROC analysis was done in R 3.2.2(R Foundation for Statistical Computing, Vienna, Austria). Results are presented as group means ± SD and significance level was set at p = 0.05. Effect sizes were estimated using the Hedges’g calculation (g) and presented with the results.
Table 2 Detailed information about the performance in the ROC analysis in each direction separately and the 2 dimensional analyses. Accuracy is defined as the sum of true negatives and positives divided by the population.
3. Results
with DCD in the ML (DCD:1.95 ± 0.27 (λs); TD:1.74 ± 0.18λs; p = 0.105; g = 0.94) or VT direction (DCD:2.27 ± 0.28; TD:2.09 ± 0.2 λs; p = 0.165; g = 0.97). When excluding the two children with DCD that held onto the handlebars during treadmill walking, the difference in ML λs between children with DCD and TD children reached statistical significance (p = 0.034). The ROC analysis of λs in separate directions and in combination showed an accuracy of discrimination between the full group of children with DCD from TD children ranging from fair to good (Table 2). AP λs performed best of all separate directions with an AUC of 0.875, which is classified as good. An optimal cut off of 1.57 λs was associated with a specificity of 0.9 and a sensitivity of 0.88 with one false positive and one false negative. The AUC of the ML (AUC = 0.750) and vertical (AUC = 0.713) λs showed fair accuracy. Our 2 dimensional ROC analysis revealed that factoring in ML along with AP λs (ML_AP) (Fig. 2a), the AUC increased to 0.91, which is rated excellent, and with a sensitivity of 1 and specificity of 0.8 (Table 2). Including VT along with AP λs (VT_ML) (Fig. 2b) also increased the AUC to 0.91, but did not improve sensitivity and specificity compared to AP λs separately (Table 2). A graphical representation of the cut off spectrums of the two- dimensional ROC analyses are presented in Fig. 2a–c.
All eight children with DCD and all ten TD children completed the four minutes treadmill walking. We found no significant difference in preferred walking speed between the children with DCD and TD (DCD:0.92 ± 0.06m/s; TD:1.06 ± 0.06m/s; p = 0.122; g = 2.33), but two children with DCD were not comfortable walking without handle support. We included these children in all analyses, as this might happen in a clinical setting, and indicate in the results if excluding these two children would change the results. The normalised RMS acceleration (RMSR) showed no difference between children with DCD and TD children in either VT (DCD: 0.63 ± 0.11; TD: 0.69 ± 0.08; p = 0.315; g = 0.64), ML (DCD: 0.54 ± 0.14; TD: 0.50 ± 0.07; p = 0.146; g = 0.38) or AP (DCD: 0.50 ± 0.17; TD: 0.51 ± 0.1; p = 0.460; g = 0.07) directions. Neither did the raw acceleration RMS (VT (DCD: 0.19 ± 0.04g; TD: 0.21 ± 0.05 g; p = 0.897; g = 0.44), ML (DCD: 0.16 ± 0.03 g; TD: 0.15 ± 0.04 g; p = 0.274; g = 0.28) or AP (DCD: 0.17 ± 0.11 g; TD: 0.15 ± 0.04 g; p = 0.897; g = 0.25). However, our analyses of λs revealed significant group differences (Fig. 1). λs was significantly greater (indicating reduced local dynamic stability) in children with DCD than that of TD children in the AP direction, i.e. the main movement direction (DCD: 1.69 ± 0.17 λs; TD:1.41 ± 0.17 λs; p = 0.005; g = 1.65) (Fig. 1a). We found no significant difference in local dynamic stability on group level in children
AnteriorPosterior (AP)
MedioLateral (ML)
Vertical (VT)
VT_ML
VT_AP
ML_AP
AUC Optimal cut-off
0.875 1.57 λs
0,750 1.79 λs
0.713 2.30 λs
0.8 –
0.91 –
0.91 –
Specificity False positive True negative
0.9 1 9
0.8 2 8
0.9 1 9
0.9 1 9
0.9 1 9
0.8 2 8
Sensitivity False negative True positive
0.88 1 7
0.75 2 6
0.62 3 5
0.75 2 6
0.875 1 7
1 0 8
Accuracy
0,89
0,78
0,78
0,83
0,89
0,89
Fig. 1. Box-plots of Lyapunov exponents during treadmill walking. The full DCD-group (n = 8) is marked in grey. The TD group (n = 10) is marked in black/grey. P-values from the Mann–Whitney U test of group difference are noted above each direction of acceleration on the graph. Outliers are presented as ●.
101
Gait & Posture 59 (2018) 99–103
M.B. Speedtsberg et al.
Fig. 2. Plot of the 2-dimensional receiver operations curve analyses (ROC). Optimal Cut-off Values are represented by the lines. The TD group (n = 10) is presented as ▲. The DCD group(n = 8) is presented as ●. a)Plot of the anterior-poster and mediolateral 2-dimensional ROC analysis (ML_AP). b)Plot of the anterior-poster and vertical 2-dimensional ROC analysis. (VT_AP). c) Plot of the vertical and mediolateral 2-dimensional ROC analysis (VT_ML).
4. Discussion
adjustments through efficient organization of sensory feed-forward and feed-back [13]. Impairments in the cerebellum, as suggested in children with DCD [1], would make automatised predictive locomotor control less efficient [13]. This could explain a less efficient control of local variability and perturbations, as the constant recalibration of walking pattern would be impaired. Previous DCD research may also suggest an involvement of impairments on a spinal level in decreased local dynamic stability. Our group recently showed increased trembling in bipedal standing [28], together with longer pre-motor time with an evoked patellar reflex and a slower motor time [29], this indicates impaired spinal and peripheral control of balance in children with DCD. So on a spinal level, children with DCD seem to have a slower reflex control than TD children, which would increase the latency of adaptive corrections of internal perturbations. Also, children with DCD have been shown to have a greater patella reflex response indicating a higher muscle spindle gain [30]. This could induce less precise corrections, and affect the proprioceptive feedback during gait [13]. Interestingly, the λs in the movement direction showed the best discriminative power between DCD and TD of the three orthogonal directions. Control of gait in this direction is less affected by visualvestibular disturbance but is highly reliant on the proprioceptive feedback information and spinal control [12]. This supports the notion that impairments influencing the proprioception and less efficient highlevel processes regulated through proprioception may explain why children with DCD have reduced local dynamic stability.
We investigated local dynamic stability during gait in children with DCD and TD during treadmill walking, and explored the discriminative power of the Lyapunov exponent (λs) between the two groups. We have shown that λs can be used to discriminate between children with DCD and TD children and that λs reveals information that may not otherwise be obtained from the dispersion of acceleration. Our results indicated that local dynamic stability could be a responsive measure when quantifying possible gait stability impairments in children with DCD. We did not show any statistical difference in the RMSR of acceleration, though higher RMS accelerations are typically associated with walking instability [23]. Several authors have suggested that measures of variability are insufficient to quantify dynamic stability [16,26]. Conversely, λs is potentially an early predictor of fall risk as it increases before global fall risk increases [19,27]. This indicates that the local dynamic stability is very sensitive to finitely small differences that may have great functional importance. It could therefore be argued that local dynamic stability is a more responsive outcome measure to evaluate the minor but functionally important effects of treatment interventions than the more skill based tests. Lower λs in children with DCD indicates a fundamental impairment of local stability of normal gait. Woodruff and colleagues concluded that there was no solid systematic gait pattern explaining the gait deviation of children with DCD from typically developing children [8]. But our results showed that local dynamic stability detects an impairment that has a high occurrence in a notoriously heterogeneous group. This could indicate an underlying impairment that may, regardless of apparent gait problems, influence stability in general ambulation or other functional tasks. This makes it a feasible discriminating biomarker between children with DCD and TD children. The main movement direction (anterior-posterior) performed best of the three directions on the ROC analysis, with good sensitivity identifying seven out of eight children with DCD. Taking medio-lateral λs into account in a twodimensional ROC analysis, we were able to identify all children with DCD and increase the area under the curve to 0.91 (excellent). These results indicate that the local dynamic stability in the movement direction (AP) is where children with DCD differ mostly from TD children. Including the ML or VT direction appear to provide additional information relevant to identifying children with DCD based on gait stability To our knowledge, λs has not yet been used to differentiate between groups in this manner, and comparison of optimal cut offs in other patient groups has not been possible. Further analysis of a larger population is therefore needed to fully determine the potential of λs in discriminating between DCD and TD during gait. The finding of reduced local dynamic stability indicates a fundamental impairment in the neural response to naturally occurring internal perturbations during gait. Gait control relies on automatic processes such as regulation of muscle tone and online postural
5. Conclusion We have shown that children with developmental coordination disorder have reduced local dynamic stability, indicating impaired attenuation of local naturally occurring perturbations during gait. We also showed that, the short term Lyapunov exponent in anterior posterior direction has good power of discrimination between DCD and TD, and excellent power when either medio-lateral or vertical direction is factored in. We therefore conclude that it is possible to discriminate between children with DCD and TD using local dynamic stability, and that it is feasible to use local dynamic stability as a responsive measure of gait stability in children with DCD. Based on the results, we further speculate that the integration of proprioceptive feedback information and spinal control could play a significant role in explaining the gait instability of children with DCD. 6. Clinical perspectives The present study proposes a method to evaluate the small but important differences in gait stability between children with DCD and typically developing children. As mentioned in the introduction, the diagnostic criteria of DCD are open to clinical interpretation, adding a test on basic locomotion in the diagnostic toolbar, could be a way to 102
Gait & Posture 59 (2018) 99–103
M.B. Speedtsberg et al.
(Accessed 8 July, 2016). [9] P.E. Roos, J.B. Dingwell, Influence of simulated neuromuscular noise on movement variability and fall risk in a 3D dynamic walking model, J. Biomech. 43 (2010) 2929–2935, http://dx.doi.org/10.1016/j.jbiomech.2010.07.008. [10] S.M. Bruijn, D.J.J. Bregman, O.G. Meijer, P.J. Beek, J.H. van Dieën, Maximum Lyapunov exponents as predictors of global gait stability: a modelling approach, Med. Eng. Phys. 34 (2012) 428–436, http://dx.doi.org/10.1016/j.medengphy. 2011.07.024. [11] J. Stenum, S.M. Bruijn, B.R. Jensen, The effect of walking speed on local dynamic stability is sensitive to calculation methods, J. Biomech. 47 (2014) 3776–3779, http://dx.doi.org/10.1016/j.jbiomech.2014.09.020. [12] C.E. Bauby, A.D. Kuo, Active control of lateral balance in human walking, J. Biomech. 33 (2000) 1433–1440. [13] K. Takakusaki, Neurophysiology of gait: from the spinal cord to the frontal lobe: neurophysiology of Gait, Mov. Disord. 28 (2013) 1483–1491, http://dx.doi.org/10. 1002/mds.25669. [14] S.E. Henderson, D. Sugden, A.L. Barnett, Movement Assessment Battery for Children-2, Harcourt Assessment, London, 2007. [15] American Psychiatric Association (Ed.), Diagnostic and Statistical Manual of Mental Disorders: DSM-5, 5th ed., American Psychiatric Association, Washington D.C, 2013. [16] J.B. Dingwell, L.C. Marin, Kinematic variability and local dynamic stability of upper body motions when walking at different speeds, J. Biomech. 39 (2006) 444–452, http://dx.doi.org/10.1016/j.jbiomech.2004.12.014. [17] M. Brandes, W. Zijlstra, S. Heikens, R. van Lummel, D. Rosenbaum, Accelerometry based assessment of gait parameters in children, Gait Posture 24 (2006) 482–486, http://dx.doi.org/10.1016/j.gaitpost.2005.12.006. [18] H.G. Kang, J.B. Dingwell, Intra-session reliability of local dynamic stability of walking, Gait Posture 24 (2006) 386–390, http://dx.doi.org/10.1016/j.gaitpost. 2005.11.004. [19] F. Reynard, P. Vuadens, O. Deriaz, P. Terrier, Could local dynamic stability serve as an early predictor of falls in patients with moderate neurological gait disorders? a reliability and comparison study in healthy individuals and in patients with paresis of the lower extremities, PLoS One 9 (2014) e100550, http://dx.doi.org/10.1371/ journal.pone.0100550. [20] F. Reynard, P. Terrier, Local dynamic stability of treadmill walking: intrasession and week-to-week repeatability, J. Biomech. 47 (2014) 74–80, http://dx.doi.org/ 10.1016/j.jbiomech.2013.10.011. [21] F. Takens, Detecting strange attractors in turbulence, in: D. Rand, L.-S. Young (Eds.), Dyn. Syst. Turbul. Warwick 1980, Springer, Berlin, Heidelberg, 1981, pp. 366–381 http://www.springerlink.com/index/10.1007/BFb0091924 (Accessed 26 November 2015). [22] M.T. Rosenstein, J.J. Collins, C.J. De Luca, A practical method for calculating largest Lyapunov exponents from small data sets, Phys. Nonlinear Phenom. 65 (1993) 117–134, http://dx.doi.org/10.1016/0167-2789(93)90009-P. [23] H.B. Menz, S.R. Lord, R.C. Fitzpatrick, Acceleration patterns of the head and pelvis when walking on level and irregular surfaces, Gait Posture 18 (2003) 35–46. [24] M. Sekine, T. Tamura, M. Yoshida, Y. Suda, Y. Kimura, H. Miyoshi, Y. Kijima, Y. Higashi, T. Fujimoto, A gait abnormality measure based on root mean square of trunk acceleration, J. Neuroeng. Rehabil. 10 (2013) 118, http://dx.doi.org/10. 1186/1743-0003-10-118. [25] M.D. Ruopp, N.J. Perkins, B.W. Whitcomb, E.F. Schisterman, Youden index and optimal cut-point estimated from observations affected by a lower limit of detection, Biom. J. 50 (2008) 419–430, http://dx.doi.org/10.1002/bimj.200710415. [26] J.B. Dingwell, J.P. Cusumano, P.R. Cavanagh, D. Sternad, Local dynamic stability versus kinematic variability of continuous overground and treadmill walking, J. Biomech. Eng. 123 (2001) 27, http://dx.doi.org/10.1115/1.1336798. [27] P.E. Roos, J.B. Dingwell, Influence of simulated neuromuscular noise on the dynamic stability and fall risk of a 3D dynamic walking model, J. Biomech. 44 (2011) 1514–1520, http://dx.doi.org/10.1016/j.jbiomech.2011.03.003. [28] M.B. Speedtsberg, S.B. Christensen, K.K. Andersen, J. Bencke, B.R. Jensen, D.J. Curtis, Impaired postural control in children with developmental coordination disorder is related to less efficient central as well as peripheral control, Gait Posture 51 (2017) 1–6, http://dx.doi.org/10.1016/j.gaitpost.2016.09.019. [29] A.J. Raynor, Fractioned reflex and reaction time in children with developmental coordination disorder, Motor Control 2 (1998) 114–124. [30] H.G. Williams, J.R. Burke, Conditioned pateallar reflex function in children with and without developmental coordination disorders, Adapt. Phys. Act. Q. 1995 (1995) 250–261.
quantify basic movement impairments of children with DCD. Furthermore local dynamic stability is also sensitive to changes after intervention in other patient groups. This gives it great potential as a tool to measure improvements in proprioception and basic locomotion after interventions. The method of measuring trunk accelerations during treadmill walking requires only a tri-axial accelerometer along with a treadmill, which is basic equipment in rehabilitation. Data processing for one patient is not time consuming, but will require a large sample size of TD to determine a cut off/spectrum of normalcy. 7. Limitations DCD is an impairment many medical professionals find difficult to diagnose, even though the criteria are well-defined [3]. Thus the diagnosed population of children with DCD is difficult to recruit. Furthermore, we were very stringent in the exclusion criteria, choosing only children diagnosed by a medical professional. It should therefore be considered that the results presented in this study are based on a small sample size and should be repeated in a larger population. We chose a short 5 min. familiarization period to treadmill walking so as not to exhaust the children’s attentional focus. Since preferred walking speed was not different between groups, we do not believe that the short familiarization period introduced confounding effects on ʎs or RMS. Conflict of interest No conflict of interests to report. References [1] J. Bo, C.-M. Lee, Motor skill learning in children with developmental coordination disorder, Res. Dev. Disabil. 34 (2013) 2047–2055, http://dx.doi.org/10.1016/j. ridd.2013.03.012. [2] R. Blank, B. Smits-Engelsman, H. Polatajko, P. Wilson, European Academy for Childhood Disability (EACD): recommendations on the definition, diagnosis and intervention of developmental coordination disorder (long version), Dev. Med. Child Neurol. 54 (2012) 54–93, http://dx.doi.org/10.1111/j.1469-8749.2011. 04171.x. [3] B.N. Wilson, K. Neil, P.H. Kamps, S. Babcock, v Awareness and knowledge of developmental co-ordination disorder among physicians, teachers and parents, Child Care Health Dev. 39 (2013) 296–300. [4] A. Kirby, D. Sugden, C. Purcell, Diagnosing developmental coordination disorders, Arch. Dis. Child. 99 (2014) 292–296, http://dx.doi.org/10.1136/archdischild2012-303569. [5] P.H. Wilson, S. Ruddock, B. Smits-Engelsman, H. Polatajko, R. Blank, Understanding performance deficits in developmental coordination disorder: a meta-analysis of recent research, Dev. Med. Child Neurol. 55 (2013) 217–228, http://dx.doi.org/10.1111/j.1469-8749.2012.04436.x. [6] K.S. Rosengren, F.J.A. Deconinck, L.A. Diberardino, J.D. Polk, J. Spencer-Smith, D. De Clercq, M. Lenoir, Differences in gait complexity and variability between children with and without developmental coordination disorder, Gait Posture 29 (2009) 225–229, http://dx.doi.org/10.1016/j.gaitpost.2008.08.005. [7] K. Wilmut, W. Du, A.L. Barnett, Gait patterns in children with developmental coordination disorder, Exp. Brain Res. 234 (2016) 1747–1755, http://dx.doi.org/10. 1007/s00221-016-4592-x. [8] S.J. Woodruff, C. Bothwell-Myers, M. Tingley, W.J. Albert, Gait pattern classification of children with developmental coordination disorder, Hum. Kinet. J. (2010) http://journals.humankinetics.com/apaq-back-issues/apaqvolume19issue3july/ gaitpatternclassificationofchildrenwithdevelopmentalcoordinationdisorder
103
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Full length article
Local dynamic stability during treadmill walking can detect children with developmental coordination disorder
MARK
⁎
Merete B. Speedtsberga,b, , Sofie B. Christensenb, Jan Stenumb,e, Thomas Kallemosef, Jesper Benckea, Derek J. Curtisa,d, Bente R. Jensenb,c a
Laboratory of Human Movement Analysis, Department of Orthopaedic Surgery, Copenhagen University Hospital Hvidovre, Copenhagen, Denmark Biomechanics and Motor Control Lab., Integrated Physiology, Department of Nutrition, Exercise and Sport, University of Copenhagen, Copenhagen, Denmark c Department of Neurology, Odense University Hospital, University of Southern Denmark, Odense, Denmark d Department of Physical and Occupational Therapy, Metropolitan University College, Copenhagen, Denmark e Locomotion Neuromechanics Laboratory, Department of Kinesiology, University of Massachusetts Amherst, Amherst, MA, USA f Clinical Orthopaedic Research Hvidovre, Department of Orthopaedic Surgery, Copenhagen University Hospital Hvidovre, Copenhagen, Denmark b
A R T I C L E I N F O
A B S T R A C T
Keywords: Developmental coordination disorder Local dynamic stability Gait stability Treadmill walking Trunk accelerations
Objective: Developmental coordination disorder (DCD) is an innate impairment of motor coordination that affects basic locomotion and balance. This study investigated local dynamic stability of trunk accelerations during treadmill walking as an objective evaluation of gait stability and the sensitivity and specificity of this measure to discriminate children with DCD from typically developing children. Method: Eight children with DCD and ten age- and gender-matched typically developing children (TD) walked four minutes on a treadmill. Trunk accelerations in vertical, medio-lateral and anterior-posterior directions were recorded with a sternum mounted accelerometer at 256 Hz. Short term local dynamic stability (λs), root mean square (RMS) and relative root mean square (RMSR) were calculated from measures of orthogonal trunk accelerations. Receiver operating characteristic curve (ROC) analysis was performed to discriminate between groups based on short term local dynamic stability. Results: λs was significantly greater in children with DCD in the main movement direction (AP) (DCD: 1.69 ± 0.17 λs; TD:1.41 ± 0.17 λs; p = 0.005), indicating reduced local dynamic stability. RMS and RMSR accelerations showed no difference between children with DCD and TD children in any direction. The ROC analysis of λs in separate directions and in two dimensions showed an excellent accuracy of discriminating between children with DCD and TD children. Anterior-posterior direction in combination with medio-lateral or vertical showed best performance with an area under the curve (AUC) of 0.91. Conclusion: We have shown that children with developmental coordination disorder have general reduced local dynamic stability and that the short term Lyapunov exponent has good power of discrimination between DCD and TD.
1. Introduction Developmental coordination disorder (DCD) is an innate impairment of motor coordination that may affect basic locomotion, balance and acquisition of motor skills. Due to great variability of motor performance between children with DCD [1] and diagnostic criteria that leave room for clinical interpretation [2], DCD is under-recognised and difficult to diagnose by health care professionals [3]. Standardised norm-referenced tests are available to assess motor function (e.g. mABC-2 & BOTMP), but they are mainly skill-based and may be influenced by experience in the specific task [4]. In summary, there could be
a need for objective measures to identify children with DCD based on deficits in basic locomotion. Gait is a basic, highly automatised motor function and less affected by experience, gender and movement culture. As instability in gait may play an important role in the motor impairments associated with DCD [5], objective evaluation of gait stability could be a relevant addition to the diagnostic toolbox and an important method to evaluate improvements in proprioception and functional motor control. Prior research on gait function in children with DCD has mainly been focused on identifying characteristics and variability in discrete spatiotemporal parameters and kinematics. Even though children with
⁎ Corresponding author at: Laboratory of Human Movement Analysis (247), Department of Orthopaedic Surgery (333), Hvidovre University Hospital, Kettegård Allé 30, 2650 Hvidovre, Denmark. E-mail address: [email protected] (M.B. Speedtsberg).
http://dx.doi.org/10.1016/j.gaitpost.2017.09.035 Received 10 January 2017; Received in revised form 10 August 2017; Accepted 27 September 2017 0966-6362/ © 2017 Published by Elsevier B.V.
Gait & Posture 59 (2018) 99–103
M.B. Speedtsberg et al.
2.2. Procedure and data acquisition
DCD exhibit a more asymmetrical and variable kinematic gait pattern [6], and a greater stride-to-stride variability [7] (normally interpreted as an expression of impaired gait control), Woodruff and colleagues concluded in their study on classification of gait patterns in children with DCD, that there was no solid systematic gait pattern explaining the gait deviation from typically developing children [8]. It remains unclear, however, how the previously reported discrete measures of variability relate to time-dependent measures of stability [9] in children with DCD. In the present study we decided to investigate local dynamic stability of trunk accelerations during treadmill walking. Local dynamic stability is a method of quantifying the body’s resilience to small perturbations naturally inherent during walking [10,11]. It is quantified over numerous sequential steps, and is a measure of how well a person responds to their internal variability and the environment’s external variability (e.g., a slippery or uneven walking surface). If small perturbations are not appropriately attenuated, the deviations from the normal gait pattern will accumulate. During gait, proprioceptive, vestibular and visuo-motor feedback contribute to an online control to attenuate the natural kinematic and kinetic variability [12,13]. This continuous processing and response requires an efficient, online motor adjustment, sensory organization and efficient neuromuscular response. These are mechanisms that are commonly reported to be impaired in children with DCD [5]. If children with DCD respond less efficiently to their naturally occurring, but increased variability, their local dynamic stability will be reduced. This may be reflected in reduced stability during gait or even increased risk of falling [9,10]. We therefore hypothesised that children with DCD have lower local dynamic stability during gait than typically developing children (TD). To establish clinical relevance, we further explored whether local dynamic stability has the sensitivity and specificity to discriminate between children with DCD and TD children.
Trunk accelerations were recorded with a sternum mounted accelerometer (MQ16, MarqMedical, Denmark) during four minutes walking at preferred walking speed. The test was performed on treadmill to control for walking speed, and each participant was accustomed to treadmill walking for maximum five minutes at different speeds comfortable to the child. Familiarisation ended at four min. if the child felt comfortable. Relatively short familiarisation was chosen to ensure full attentional focus during testing. Preferred walking speed was determined according to the recognised protocol by Dingwell and Marin [11,16], in which preferred walking speed is selected 6 times by the subject while alternating between increasing and decreasing speeds from 0.42 m/s above or below the previously selected speed. The preferred walking speed is subsequently calculated as the average of the six self-selected speeds. The participants were instructed, to walk without support from treadmill handles and were informed of time progress every minute. Upper body accelerations in vertical (VT), medio-lateral (ML) and anteror-posterior (AP) directions were sampled at 256 Hz (MQ16, Marq Medical, Denmark). Recordings began when the treadmill reached a constant speed and the participant was comfortable with the selected speed. 2.3. Data analysis All data processing and calculations were performed using customised MATLAB (The MathWorks, Inc., Natick, MA, USA) scripts [11]. Accelerometer based assessment of gait parameters has been validated in children [17]. The primary outcome was short term local dynamic stability in each VT, ML and AP directions. Root mean square was calculated for each direction of acceleration as a measure of variability. The method for calculating local dynamic stability has been described extensively in the literature [10,11,16]. Local dynamic stability was quantified using nonlinear time-series methods, and expressed as the short term Lyapunov exponent (λs) which has shown good reliability, good intrasession and fair intersession repeatability and a correlation to falls risk [10,18–20]. The first 195 strides were identified from the vertical accelerations, and the unfiltered time-series were time-normalised to 19,500 data points using cubic spline interpolation. Five-dimensional state spaces were reconstructed from each acceleration direction using the method of delays [21]. From the reconstructed state spaces Euclidean distances between nearest neighbours in state space were calculated as a function of time and averaged over all nearest neighbours. λs was calculated as the average rate of logarithmic divergence of the distance between nearest neighbours in state space from 0 to 0.5 stride (ln(div) > /stride-time from 0–0.5 stride) [11,22]. A higher λs expresses lower local dynamic stability as the rate of divergence in state space is faster. Acceleration Root Mean Square (RMS) was calculated as the dispersion of the acceleration data relative to zero and quantifies the average magnitude of accelerations in each direction during a complete walking trial [23]. To normalise RMS for effect of walking speed, the ratios between RMS accelleration in each direction and the RMS vector magnitude (RMSR) were calculated [24].
2. Method 2.1. Participants Eight children with DCD and ten age-, gender- and anthropometrically matched TD children with normal motor proficiency were recruited (Table 1). The DCD group was recruited through paediatric physiotherapists, occupational therapists and the Danish Parents Association for DCD. Motor performance was assessed using the Movement Assessment Battery for children (MABC-2), which is a norm-referenced basic motor abilities assessment tool containing fine motor, ball handling, and balance tasks [14]. Inclusion criteria for children with DCD were a score below the 15th percentile in the MABC-2 and to meet the official criteria for DCD according to the DSM-V [15] based on evaluation by a qualified health care professional. To avoid confounding factors, we did not include children with co-morbidities. Inclusion criteria for the TD-children were a score above the 16th percentile in the MABC-2, and no diagnosed or suspected neurodevelopment disorders. Informed written consent was obtained from a legal guardian. The experimental protocol was approved by the Capital Region Committee on Health Research Ethics, Denmark (ref.: H-4-2013-144).
2.4. Statistical analysis The distribution of the data was analysed for normality using the Shapiro-Wilk test and Q-Q plots in IBM SPSS statistics 22 (IBM Corp. Released 2013. IBM SPSS Statistics for Windows). Non-parametric (Mann–Whitney U) tests were used to test for group differences in age, height, body mass and MABC-2 percentile. A chi-squared test was used to evaluate sex differences between groups. Non-parametric (Mann–Whitney U) tests were performed to compare the short term λs, RMS and RMSR between children with DCD and TD children in separate directions of acceleration. The discriminative power of λs to
Table 1 Participant Characteristics (mean ± SD).
Age, years MABC-2 percentile Gender (♂/♀), n Height, cm Body mass, kg
DCD group (n = 8)
TD group (n = 10)
p-value
8.8 ± 1.5 2.6 ± 3.93 6/2 139.5 ± 8.1 33.6 ± 7.3
9.1 ± 1.4 73.3 ± 5.2 7/3 141.1 ± 3.0 33.7 ± 1.8
0.885 < 0.001 0.658 0.763 0.839
100
Gait & Posture 59 (2018) 99–103
M.B. Speedtsberg et al.
differentiate children with DCD and TD was assessed using receiver operating characteristic curve (ROC) analysis based on logistic regression. Area under the curve (AUC) and the optimal cut off were calculated, based on Youden’s index [25]. The ROC analyses were performed on separate directions of acceleration and also two-dimensionally taking two directions of acceleration and interaction between them into account. Since the cut off values of the variables in a two-dimensional ROC analyses are interdependent, a cutoff value for one of the variables can only be calculated given the value of the other variable and vice versa. As such a single value for each variable can’t be specified, rather a spectrum of values given by the value of the other variable. Because of this the cutoff values for the 2 dimensional ROC analyses are represented graphically. ROC analysis was done in R 3.2.2(R Foundation for Statistical Computing, Vienna, Austria). Results are presented as group means ± SD and significance level was set at p = 0.05. Effect sizes were estimated using the Hedges’g calculation (g) and presented with the results.
Table 2 Detailed information about the performance in the ROC analysis in each direction separately and the 2 dimensional analyses. Accuracy is defined as the sum of true negatives and positives divided by the population.
3. Results
with DCD in the ML (DCD:1.95 ± 0.27 (λs); TD:1.74 ± 0.18λs; p = 0.105; g = 0.94) or VT direction (DCD:2.27 ± 0.28; TD:2.09 ± 0.2 λs; p = 0.165; g = 0.97). When excluding the two children with DCD that held onto the handlebars during treadmill walking, the difference in ML λs between children with DCD and TD children reached statistical significance (p = 0.034). The ROC analysis of λs in separate directions and in combination showed an accuracy of discrimination between the full group of children with DCD from TD children ranging from fair to good (Table 2). AP λs performed best of all separate directions with an AUC of 0.875, which is classified as good. An optimal cut off of 1.57 λs was associated with a specificity of 0.9 and a sensitivity of 0.88 with one false positive and one false negative. The AUC of the ML (AUC = 0.750) and vertical (AUC = 0.713) λs showed fair accuracy. Our 2 dimensional ROC analysis revealed that factoring in ML along with AP λs (ML_AP) (Fig. 2a), the AUC increased to 0.91, which is rated excellent, and with a sensitivity of 1 and specificity of 0.8 (Table 2). Including VT along with AP λs (VT_ML) (Fig. 2b) also increased the AUC to 0.91, but did not improve sensitivity and specificity compared to AP λs separately (Table 2). A graphical representation of the cut off spectrums of the two- dimensional ROC analyses are presented in Fig. 2a–c.
All eight children with DCD and all ten TD children completed the four minutes treadmill walking. We found no significant difference in preferred walking speed between the children with DCD and TD (DCD:0.92 ± 0.06m/s; TD:1.06 ± 0.06m/s; p = 0.122; g = 2.33), but two children with DCD were not comfortable walking without handle support. We included these children in all analyses, as this might happen in a clinical setting, and indicate in the results if excluding these two children would change the results. The normalised RMS acceleration (RMSR) showed no difference between children with DCD and TD children in either VT (DCD: 0.63 ± 0.11; TD: 0.69 ± 0.08; p = 0.315; g = 0.64), ML (DCD: 0.54 ± 0.14; TD: 0.50 ± 0.07; p = 0.146; g = 0.38) or AP (DCD: 0.50 ± 0.17; TD: 0.51 ± 0.1; p = 0.460; g = 0.07) directions. Neither did the raw acceleration RMS (VT (DCD: 0.19 ± 0.04g; TD: 0.21 ± 0.05 g; p = 0.897; g = 0.44), ML (DCD: 0.16 ± 0.03 g; TD: 0.15 ± 0.04 g; p = 0.274; g = 0.28) or AP (DCD: 0.17 ± 0.11 g; TD: 0.15 ± 0.04 g; p = 0.897; g = 0.25). However, our analyses of λs revealed significant group differences (Fig. 1). λs was significantly greater (indicating reduced local dynamic stability) in children with DCD than that of TD children in the AP direction, i.e. the main movement direction (DCD: 1.69 ± 0.17 λs; TD:1.41 ± 0.17 λs; p = 0.005; g = 1.65) (Fig. 1a). We found no significant difference in local dynamic stability on group level in children
AnteriorPosterior (AP)
MedioLateral (ML)
Vertical (VT)
VT_ML
VT_AP
ML_AP
AUC Optimal cut-off
0.875 1.57 λs
0,750 1.79 λs
0.713 2.30 λs
0.8 –
0.91 –
0.91 –
Specificity False positive True negative
0.9 1 9
0.8 2 8
0.9 1 9
0.9 1 9
0.9 1 9
0.8 2 8
Sensitivity False negative True positive
0.88 1 7
0.75 2 6
0.62 3 5
0.75 2 6
0.875 1 7
1 0 8
Accuracy
0,89
0,78
0,78
0,83
0,89
0,89
Fig. 1. Box-plots of Lyapunov exponents during treadmill walking. The full DCD-group (n = 8) is marked in grey. The TD group (n = 10) is marked in black/grey. P-values from the Mann–Whitney U test of group difference are noted above each direction of acceleration on the graph. Outliers are presented as ●.
101
Gait & Posture 59 (2018) 99–103
M.B. Speedtsberg et al.
Fig. 2. Plot of the 2-dimensional receiver operations curve analyses (ROC). Optimal Cut-off Values are represented by the lines. The TD group (n = 10) is presented as ▲. The DCD group(n = 8) is presented as ●. a)Plot of the anterior-poster and mediolateral 2-dimensional ROC analysis (ML_AP). b)Plot of the anterior-poster and vertical 2-dimensional ROC analysis. (VT_AP). c) Plot of the vertical and mediolateral 2-dimensional ROC analysis (VT_ML).
4. Discussion
adjustments through efficient organization of sensory feed-forward and feed-back [13]. Impairments in the cerebellum, as suggested in children with DCD [1], would make automatised predictive locomotor control less efficient [13]. This could explain a less efficient control of local variability and perturbations, as the constant recalibration of walking pattern would be impaired. Previous DCD research may also suggest an involvement of impairments on a spinal level in decreased local dynamic stability. Our group recently showed increased trembling in bipedal standing [28], together with longer pre-motor time with an evoked patellar reflex and a slower motor time [29], this indicates impaired spinal and peripheral control of balance in children with DCD. So on a spinal level, children with DCD seem to have a slower reflex control than TD children, which would increase the latency of adaptive corrections of internal perturbations. Also, children with DCD have been shown to have a greater patella reflex response indicating a higher muscle spindle gain [30]. This could induce less precise corrections, and affect the proprioceptive feedback during gait [13]. Interestingly, the λs in the movement direction showed the best discriminative power between DCD and TD of the three orthogonal directions. Control of gait in this direction is less affected by visualvestibular disturbance but is highly reliant on the proprioceptive feedback information and spinal control [12]. This supports the notion that impairments influencing the proprioception and less efficient highlevel processes regulated through proprioception may explain why children with DCD have reduced local dynamic stability.
We investigated local dynamic stability during gait in children with DCD and TD during treadmill walking, and explored the discriminative power of the Lyapunov exponent (λs) between the two groups. We have shown that λs can be used to discriminate between children with DCD and TD children and that λs reveals information that may not otherwise be obtained from the dispersion of acceleration. Our results indicated that local dynamic stability could be a responsive measure when quantifying possible gait stability impairments in children with DCD. We did not show any statistical difference in the RMSR of acceleration, though higher RMS accelerations are typically associated with walking instability [23]. Several authors have suggested that measures of variability are insufficient to quantify dynamic stability [16,26]. Conversely, λs is potentially an early predictor of fall risk as it increases before global fall risk increases [19,27]. This indicates that the local dynamic stability is very sensitive to finitely small differences that may have great functional importance. It could therefore be argued that local dynamic stability is a more responsive outcome measure to evaluate the minor but functionally important effects of treatment interventions than the more skill based tests. Lower λs in children with DCD indicates a fundamental impairment of local stability of normal gait. Woodruff and colleagues concluded that there was no solid systematic gait pattern explaining the gait deviation of children with DCD from typically developing children [8]. But our results showed that local dynamic stability detects an impairment that has a high occurrence in a notoriously heterogeneous group. This could indicate an underlying impairment that may, regardless of apparent gait problems, influence stability in general ambulation or other functional tasks. This makes it a feasible discriminating biomarker between children with DCD and TD children. The main movement direction (anterior-posterior) performed best of the three directions on the ROC analysis, with good sensitivity identifying seven out of eight children with DCD. Taking medio-lateral λs into account in a twodimensional ROC analysis, we were able to identify all children with DCD and increase the area under the curve to 0.91 (excellent). These results indicate that the local dynamic stability in the movement direction (AP) is where children with DCD differ mostly from TD children. Including the ML or VT direction appear to provide additional information relevant to identifying children with DCD based on gait stability To our knowledge, λs has not yet been used to differentiate between groups in this manner, and comparison of optimal cut offs in other patient groups has not been possible. Further analysis of a larger population is therefore needed to fully determine the potential of λs in discriminating between DCD and TD during gait. The finding of reduced local dynamic stability indicates a fundamental impairment in the neural response to naturally occurring internal perturbations during gait. Gait control relies on automatic processes such as regulation of muscle tone and online postural
5. Conclusion We have shown that children with developmental coordination disorder have reduced local dynamic stability, indicating impaired attenuation of local naturally occurring perturbations during gait. We also showed that, the short term Lyapunov exponent in anterior posterior direction has good power of discrimination between DCD and TD, and excellent power when either medio-lateral or vertical direction is factored in. We therefore conclude that it is possible to discriminate between children with DCD and TD using local dynamic stability, and that it is feasible to use local dynamic stability as a responsive measure of gait stability in children with DCD. Based on the results, we further speculate that the integration of proprioceptive feedback information and spinal control could play a significant role in explaining the gait instability of children with DCD. 6. Clinical perspectives The present study proposes a method to evaluate the small but important differences in gait stability between children with DCD and typically developing children. As mentioned in the introduction, the diagnostic criteria of DCD are open to clinical interpretation, adding a test on basic locomotion in the diagnostic toolbar, could be a way to 102
Gait & Posture 59 (2018) 99–103
M.B. Speedtsberg et al.
(Accessed 8 July, 2016). [9] P.E. Roos, J.B. Dingwell, Influence of simulated neuromuscular noise on movement variability and fall risk in a 3D dynamic walking model, J. Biomech. 43 (2010) 2929–2935, http://dx.doi.org/10.1016/j.jbiomech.2010.07.008. [10] S.M. Bruijn, D.J.J. Bregman, O.G. Meijer, P.J. Beek, J.H. van Dieën, Maximum Lyapunov exponents as predictors of global gait stability: a modelling approach, Med. Eng. Phys. 34 (2012) 428–436, http://dx.doi.org/10.1016/j.medengphy. 2011.07.024. [11] J. Stenum, S.M. Bruijn, B.R. Jensen, The effect of walking speed on local dynamic stability is sensitive to calculation methods, J. Biomech. 47 (2014) 3776–3779, http://dx.doi.org/10.1016/j.jbiomech.2014.09.020. [12] C.E. Bauby, A.D. Kuo, Active control of lateral balance in human walking, J. Biomech. 33 (2000) 1433–1440. [13] K. Takakusaki, Neurophysiology of gait: from the spinal cord to the frontal lobe: neurophysiology of Gait, Mov. Disord. 28 (2013) 1483–1491, http://dx.doi.org/10. 1002/mds.25669. [14] S.E. Henderson, D. Sugden, A.L. Barnett, Movement Assessment Battery for Children-2, Harcourt Assessment, London, 2007. [15] American Psychiatric Association (Ed.), Diagnostic and Statistical Manual of Mental Disorders: DSM-5, 5th ed., American Psychiatric Association, Washington D.C, 2013. [16] J.B. Dingwell, L.C. Marin, Kinematic variability and local dynamic stability of upper body motions when walking at different speeds, J. Biomech. 39 (2006) 444–452, http://dx.doi.org/10.1016/j.jbiomech.2004.12.014. [17] M. Brandes, W. Zijlstra, S. Heikens, R. van Lummel, D. Rosenbaum, Accelerometry based assessment of gait parameters in children, Gait Posture 24 (2006) 482–486, http://dx.doi.org/10.1016/j.gaitpost.2005.12.006. [18] H.G. Kang, J.B. Dingwell, Intra-session reliability of local dynamic stability of walking, Gait Posture 24 (2006) 386–390, http://dx.doi.org/10.1016/j.gaitpost. 2005.11.004. [19] F. Reynard, P. Vuadens, O. Deriaz, P. Terrier, Could local dynamic stability serve as an early predictor of falls in patients with moderate neurological gait disorders? a reliability and comparison study in healthy individuals and in patients with paresis of the lower extremities, PLoS One 9 (2014) e100550, http://dx.doi.org/10.1371/ journal.pone.0100550. [20] F. Reynard, P. Terrier, Local dynamic stability of treadmill walking: intrasession and week-to-week repeatability, J. Biomech. 47 (2014) 74–80, http://dx.doi.org/ 10.1016/j.jbiomech.2013.10.011. [21] F. Takens, Detecting strange attractors in turbulence, in: D. Rand, L.-S. Young (Eds.), Dyn. Syst. Turbul. Warwick 1980, Springer, Berlin, Heidelberg, 1981, pp. 366–381 http://www.springerlink.com/index/10.1007/BFb0091924 (Accessed 26 November 2015). [22] M.T. Rosenstein, J.J. Collins, C.J. De Luca, A practical method for calculating largest Lyapunov exponents from small data sets, Phys. Nonlinear Phenom. 65 (1993) 117–134, http://dx.doi.org/10.1016/0167-2789(93)90009-P. [23] H.B. Menz, S.R. Lord, R.C. Fitzpatrick, Acceleration patterns of the head and pelvis when walking on level and irregular surfaces, Gait Posture 18 (2003) 35–46. [24] M. Sekine, T. Tamura, M. Yoshida, Y. Suda, Y. Kimura, H. Miyoshi, Y. Kijima, Y. Higashi, T. Fujimoto, A gait abnormality measure based on root mean square of trunk acceleration, J. Neuroeng. Rehabil. 10 (2013) 118, http://dx.doi.org/10. 1186/1743-0003-10-118. [25] M.D. Ruopp, N.J. Perkins, B.W. Whitcomb, E.F. Schisterman, Youden index and optimal cut-point estimated from observations affected by a lower limit of detection, Biom. J. 50 (2008) 419–430, http://dx.doi.org/10.1002/bimj.200710415. [26] J.B. Dingwell, J.P. Cusumano, P.R. Cavanagh, D. Sternad, Local dynamic stability versus kinematic variability of continuous overground and treadmill walking, J. Biomech. Eng. 123 (2001) 27, http://dx.doi.org/10.1115/1.1336798. [27] P.E. Roos, J.B. Dingwell, Influence of simulated neuromuscular noise on the dynamic stability and fall risk of a 3D dynamic walking model, J. Biomech. 44 (2011) 1514–1520, http://dx.doi.org/10.1016/j.jbiomech.2011.03.003. [28] M.B. Speedtsberg, S.B. Christensen, K.K. Andersen, J. Bencke, B.R. Jensen, D.J. Curtis, Impaired postural control in children with developmental coordination disorder is related to less efficient central as well as peripheral control, Gait Posture 51 (2017) 1–6, http://dx.doi.org/10.1016/j.gaitpost.2016.09.019. [29] A.J. Raynor, Fractioned reflex and reaction time in children with developmental coordination disorder, Motor Control 2 (1998) 114–124. [30] H.G. Williams, J.R. Burke, Conditioned pateallar reflex function in children with and without developmental coordination disorders, Adapt. Phys. Act. Q. 1995 (1995) 250–261.
quantify basic movement impairments of children with DCD. Furthermore local dynamic stability is also sensitive to changes after intervention in other patient groups. This gives it great potential as a tool to measure improvements in proprioception and basic locomotion after interventions. The method of measuring trunk accelerations during treadmill walking requires only a tri-axial accelerometer along with a treadmill, which is basic equipment in rehabilitation. Data processing for one patient is not time consuming, but will require a large sample size of TD to determine a cut off/spectrum of normalcy. 7. Limitations DCD is an impairment many medical professionals find difficult to diagnose, even though the criteria are well-defined [3]. Thus the diagnosed population of children with DCD is difficult to recruit. Furthermore, we were very stringent in the exclusion criteria, choosing only children diagnosed by a medical professional. It should therefore be considered that the results presented in this study are based on a small sample size and should be repeated in a larger population. We chose a short 5 min. familiarization period to treadmill walking so as not to exhaust the children’s attentional focus. Since preferred walking speed was not different between groups, we do not believe that the short familiarization period introduced confounding effects on ʎs or RMS. Conflict of interest No conflict of interests to report. References [1] J. Bo, C.-M. Lee, Motor skill learning in children with developmental coordination disorder, Res. Dev. Disabil. 34 (2013) 2047–2055, http://dx.doi.org/10.1016/j. ridd.2013.03.012. [2] R. Blank, B. Smits-Engelsman, H. Polatajko, P. Wilson, European Academy for Childhood Disability (EACD): recommendations on the definition, diagnosis and intervention of developmental coordination disorder (long version), Dev. Med. Child Neurol. 54 (2012) 54–93, http://dx.doi.org/10.1111/j.1469-8749.2011. 04171.x. [3] B.N. Wilson, K. Neil, P.H. Kamps, S. Babcock, v Awareness and knowledge of developmental co-ordination disorder among physicians, teachers and parents, Child Care Health Dev. 39 (2013) 296–300. [4] A. Kirby, D. Sugden, C. Purcell, Diagnosing developmental coordination disorders, Arch. Dis. Child. 99 (2014) 292–296, http://dx.doi.org/10.1136/archdischild2012-303569. [5] P.H. Wilson, S. Ruddock, B. Smits-Engelsman, H. Polatajko, R. Blank, Understanding performance deficits in developmental coordination disorder: a meta-analysis of recent research, Dev. Med. Child Neurol. 55 (2013) 217–228, http://dx.doi.org/10.1111/j.1469-8749.2012.04436.x. [6] K.S. Rosengren, F.J.A. Deconinck, L.A. Diberardino, J.D. Polk, J. Spencer-Smith, D. De Clercq, M. Lenoir, Differences in gait complexity and variability between children with and without developmental coordination disorder, Gait Posture 29 (2009) 225–229, http://dx.doi.org/10.1016/j.gaitpost.2008.08.005. [7] K. Wilmut, W. Du, A.L. Barnett, Gait patterns in children with developmental coordination disorder, Exp. Brain Res. 234 (2016) 1747–1755, http://dx.doi.org/10. 1007/s00221-016-4592-x. [8] S.J. Woodruff, C. Bothwell-Myers, M. Tingley, W.J. Albert, Gait pattern classification of children with developmental coordination disorder, Hum. Kinet. J. (2010) http://journals.humankinetics.com/apaq-back-issues/apaqvolume19issue3july/ gaitpatternclassificationofchildrenwithdevelopmentalcoordinationdisorder
103