INTRODUCTION AND PURPOSE
Developmental coordination disorder (DCD) is a neurodevelopmental disorder characterized by impairment in motor coordination1 leading to poor balance, delayed developmental characteristics, and motor coordination deficits.2 Children with DCD are considered a heterogeneous group as the motor coordination deficits tend to vary between individuals3 and consequently problems can arise when attempting to clinically evaluate the gait patterns of these children.
In a study looking at gait patterns in children with DCD (n = 7), Woodruff et al4 reported greater variance in the gait pattern of children with DCD when compared to a normative gait database for children who were typically developing, particularly for percentage of gait cycle for opposite toe off, single stance, toe off, and step length. Gait data for the children with DCD was derived from analysis using a 3-camera VICON motion analysis system (Oxford Metrics Ltd) and, despite the limitations of the work, the researchers reported no systematic pattern to the differences in parameters. A later study evaluating gait parameters in a sample of children with DCD (n = 10) and age matched controls (n = 10) during treadmill walking reported differences in temporal gait parameters for the children with DCD.5 A shorter stride time (P < .05), decreased double support duration (P < .05), decreased swing phase duration (P < .05), a higher cadence (P < .05), and decreased stride length (P < .05) were reported. These findings add to the clinical picture such that children with DCD can be described as walking with smaller but faster steps to maintain speed while minimizing the time when balance can be compromised. A recent study designed to characterize gait complexity and variability through the use of 3-dimensional gait analysis and elliptical Fourier analysis concluded that there was greater complexity in the movement characteristics of the lower limb in children with DCD suggesting difficulty with limb control during gait.6
Spatial gait parameters (including stride length and step length) and temporal gait parameters (including double and single support duration, stance, and swing phase duration) offer useful clinical data for assessment of the child, determining treatment objective(s) and also in monitoring the efficacy of intervention. Gait patterns in children are known for their variability when compared to adults and also when compared across age groups as gait develops. Greater trial-to-trial variability of gait in children has been reported when comparisons were drawn to adults.7 Comparison between a sample of children (mean age = 6.8 years) and adults (mean age = 27.5 years) revealed lower reliability in the children for gait velocity (m/s), stride length (cm), step length (cm), cadence (steps/min), step width (mm), foot angle (°), stance phase (ms), double support (ms), and gait cycle time (ms) but not swing phase (ms).7 Greater variability in the gait pattern was also reported for the children and considered to be due to poorer postural control.
Further work on the reliability of gait reported variable levels of reliability for spatiotemporal gait parameters in a sample of 57 children, stratified across 3 age groups.8 Reliability reported in the study differed across the 3 age groups with no clear pattern to the gait variables other than velocity and cadence which maintained good reliability across the age span. In a study evaluating spatiotemporal gait parameters in children with motor disabilities (including cerebral palsy, Angelman syndrome, and arthrogryphosis), excellent reliability was reported for cadence, velocity, cycle time, base width, and stride length.9 To the authors' knowledge there have been no studies looking at the reliability of gait parameters in children with DCD. Because of the gait variability typically observed in children with DCD it is unclear as to what use the clinical evaluation of spatiotemporal parameters might serve as an atypical gait pattern makes it difficult for the clinician to monitor and objectively evaluate the efficacy of intervention and rehabilitation. Therefore, the aim of this study was to evaluate the intra-session reliability of spatiotemporal gait parameters used in the clinical evaluation of a sample of children with DCD.
The study sample was recruited from children entering a multidisciplinary therapy program for those diagnosed with DCD at the Sanderson Child Development Centre, Medway Maritime Hospital and ethical approval was granted by University of East London Ethics Committee and Lewisham Local Research Ethics Committee. All children aged between 6 and 11 years and referred with a diagnosis of DCD were identified as potential participants for the study. Following entry to the rehabilitation programme children were invited to participate. Children with additional medical complications likely to affect gait were excluded from the study and this included any condition affecting neuromuscular integrity and orthopedic conditions (such as talipes equinovarus) leading to a gait disorder. Any child wearing foot orthoses in their shoes were excluded from the study and all children had their feet and footwear screened by a podiatrist prior to inclusion. All children wearing Velcro or laced-up trainers/school shoes were invited to participate. Parents/guardians of all children suitable for inclusion were required to give written consent before being enrolled in the study. Children were also asked for assent prior to participation.
The GAITRite walkway was used for the collection of spatiotemporal gait parameters. The 4.5-meter mat was placed in the center of a walkway and each participant was instructed to walk across the mat to become comfortable with the procedure. Following this each participant was asked to ambulate across the mat and 3 trials were captured, with an average of 4 to 5 steps per trial. All children were instructed to walk in their normal style at a self-selected speed from approximately 2 meters before the mat and for 2 meters after the mat, while looking ahead at a point on the wall at the end of the room. Any footsteps that did not fall within the active sensor area of the mat were deleted and any trials with obvious disruption to the typical gait pattern were deleted and the trial recaptured. All trials were conducted in the footwear identified prior to participation in the study. Only trials deemed by the parents/guardians as accurately representing their child's normal gait were accepted and the parameters recorded were: velocity (cm/sec), stance phase duration (% of gait cycle), swing phase duration (% of gait cycle), stride length (cm), double support (% of gait cycle), and single support (% of gait cycle).
Data for both limbs were extracted for all participants and statistical analysis was conducted using SPSS Version 15 for Windows (SPSS Inc, Chicago, Illinois). Data were initially explored for normality with the Kolmogorov-Smirnov test. All gait parameters were uniform and normally distributed (P > .05). Reliability was assessed across 3 repeated trials collected on the same day and evaluated with intraclass correlation coefficients (ICC) (3,1). Absolute measures of reliability were evaluated using coefficients of variation (CV), an expression of the standard deviation as a proportion of the mean.10 The interpretation of the CV is independent of units of measurement but without defined boundary and there are no recognized thresholds for poor, moderate, or good variability in the data other than the higher the percentage scores the greater the variability in the data. ICC values were interpreted in accordance with Portney and Watkins11 where values greater than 0.75 indicate good reliability, 0.5 to 0.75 moderate reliability, and values below 0.5 indicative of poor reliability. Systematic differences between the left and right limb (L limb and R limb, respectively) were evaluated with the paired t test with the level of significance set at 0.05. To satisfy clinical interest in the results, data from both limbs are presented.
Twenty participants were recruited into the study; all were male with an age range of 6 to 11 years and a mean age of 7.5 years. Coefficients of variation are presented in Table 1 and demonstrate a range of values for the gait parameters ranging from 2.2% to 12.1%, with stance phase duration (L limb) demonstrating the lowest variation (2.7%) in measurement and double support duration (R limb) demonstrating the greatest (12.7%).
As an index of reliability, the ICC was applied across a mean of 3 ratings values, presented in Table 1. Values ranged from 0.24 to 0.73 with good reliability achieved for one parameter (cadence = 0.73) and moderate reliability achieved for step length (right = 0.55 and left = 0.58), stride length (left = 0.57 and right = 0.61), and double support duration (left = 0.56 and right = 0.59). Poor and moderate reliability was found for swing phase duration (right = 0.32 and left = 0.57), and single support duration (left = 0.33 and right = 0.55), and poor reliability for velocity (0.24).
On comparison of gait parameters between left and right limb, significant differences were found for step length and single support duration (P ≤ .05) (Table 2).
Establishing the reliability of gait parameters is important to ensure confidence in monitoring and objectively evaluating the efficacy of therapies for intervention, rehabilitation, or both. Spatiotemporal gait parameters are often used as outcome measures but to date very few studies have considered the reliability of spatiotemporal gait parameters in children with DCD. This study was conducted to determine the intra-session reliability of spatiotemporal gait parameters in a sample of children aged 6 to 11 years of age with DCD and analysis of the results revealed a range of ICC scores representing poor to good reliability (see Table 1). In the context of this study spatiotemporal gait parameters demonstrating poor reliability (velocity) may be explained by the variable gait pattern typical of children with DCD4 where inconsistency in the repeated measures yielded a variable data and thus poor reliability, defined by a low ICC scores.12 Poor reliability could also be explained by very low variability within the data as evidenced by the standard deviations for swing and stance phase duration (see Table 1). The low standard deviation could be a result of the subject scores being too homogenous. On initial consideration this may appear to represent less variation in the parameter yet it results in a lowering of the ICC reliability score as the statistical calculation for the ICC requires a certain degree of variability within the parameter to be calculated.12
Parameters with moderate to good reliability indicate consistency in the measurement across participants and leads to confidence that the observed variance in measurements is due to the differences in gait patterns between the children. In this study step length, stride length, and double support duration demonstrated moderate reliability (see Table 1). The ICC is dependent on the heterogeneity of the data set and is also a unit-less value which supports grading of reliability (ie, from poor to excellent) but does not reflect absolute reliability and therefore, further consideration of the CV (%) data was warranted.
The CV is an expression of the standard deviation as a proportion of the mean10 and portrays the distribution of the data around the mean as a percentage. This allows comparison of the spread of each parameter to the next regardless of the value of the parameter and as observed in Table 1, the CV for most gait parameters was quite low which was deemed acceptable for the study. Comparison of the gait data between the left and right limbs (see Table 2) reached statistical significance for single support duration and step length. These differences between the limbs could be a result of poorer balance that is typical in children with DCD.
Direct comparison of the findings from the published literature is limited by the group recruited but nevertheless, an appreciation of the published data is warranted. Thorpe et al8 evaluated the reliability of spatiotemporal gait parameters in 57 children who are typically developing aged 1.3 to 10.9 years with participants stratified into 3 age groups. On comparison with the 2 age groups matched to those in this study (4 to < 8 years of age, 8 to < 11 years of age), the findings followed a similar trend where cadence was the most reliable parameter and single support and double support duration had moderate reliability, although moderate reliability for single support duration was only reported for the L limb in the study by Thorpe et al.8 In children who are typically developing velocity was reported to have good reliability (0.73–0.74),9 but this was not supported in this study where poor reliability (0.24) was reported. In the study by Wondra et al,9 good reliability for the chosen variables was established in the sample of children with motor disabilities (0.94 for cadence, 0.97 for velocity, 0.98 for stride length and 0.97 for stance). The differences between studies can be explained by the samples under investigation and future studies looking at parameters in a specific group of children need to identify reliable parameters specific to that group to evaluate changes appropriately.
The findings from our research highlight that some spatiotemporal parameters are more reliable than others and those demonstrating moderate to good reliability should be used in the evaluation of treatment, rehabilitation, or both in children with DCD. It must be appreciated that gait analysis is complex and challenging in children and it is important to acknowledge that gathering reliable data may be dependent on a number of variables such as participants' age and developmental status. In this study it was noted that the younger children struggled to maintain their concentration throughout the data collection and this may have introduced variability into the data set. Furthermore, instrumentation (technology and reliability and validity of instrumentation) and the environment where data are collected must also be considered to ensure rigor of the data. This study was conducted as part of routine assessment in the clinical setting and thus it was difficult to establish procedures for controlling all gait variables (such as gait velocity). This could explain some of the variability within the data set and explain the poorer reliability scores. Nevertheless, stringent protocols could influence the natural gait pattern of children with DCD and lead to gait changes not necessarily reflective of their true gait pattern. As reported, children with DCD have a variable gait pattern and a limitation of this research is that data were captured in 1 test session. Consequently, it was not possible to conclude on the temporal stability of these parameters by conducting analysis of test-retest reliability. Further work should consider the test-retest reliability to determine whether or not spatiotemporal gait parameters maintain their reliability across time. The extrapolation of the data to females with DCD is limited as only male participants entered into the study and further work is needed before conclusions can be drawn. Further work should also build upon the small sample size recruited for this study with further analysis across stratified age groups.
Reliability of spatiotemporal gait parameters in children with DCD yielded mixed reliability outcomes for the parameters selected in this study. The results suggest that the collection of reliable spatiotemporal gait parameters in children with DCD is possible and 1 parameter attained good reliability (cadence) while a number demonstrated moderate reliability across both limbs (step length, stride length, and double support duration). In light of the variable gait pattern typical in children with DCD good reliability may be difficult to achieve and therefore, parameters demonstrating moderate to good reliability should be considered in further studies measuring spatiotemporal gait parameters. The results presented can only apply to the sample under investigation and to the instrumentation used in this study, and further work is warranted to address the limitations of this study.
1. Cherng RJ, Liang LY, Chen YJ, et al. The effects of a motor and a cognitive concurrent task on walking in children with developmental coordination disorder
2. Polatajko HJ, Cantin N. Developmental coordination disorder
(dyspraxia): an overview of the state of the art. Semin Pediatr Neurol. 2006;13:250–258.
3. Dewey D, Wilson BN. Developmental coordination disorder
: what is it? Phys Occup Ther Pediatr. 2001;20:5–28.
4. Woodruff SJ, Bothwell-Myers C, Tingley M, Albert WJ. Gait
pattern classification of children with developmental coordination disorder
. Adapt Phys Activ Q. 2002;19:378–391.
5. Deconinck FJ, De Clercq D, Savelsbergh GJ, et al. Differences in gait
between children with and without developmental coordination disorder
. Motor Control. 2006;10:125–142.
6. Rosengren KS, Deconinck FJ, DiBeradino LA, et al. Differences in gait
complexity and variability with and without developmental coordination disorder
7. Stolze H, Kuhtz-buschbeck JP, Mondwurf C, et al. Retest reliability of spatiotemporal gait
parameters in children and adults. Gait
8. Thorpe DE, Dusing SC, Moore CG. Repeatability of temporospatial gait
measures in children using the GAITRite® electronic walkway. Arch Phys Med Rehabil. 2005;86:2342–2346.
9. Wondra VC, Pitetti KH, Beets MW. Gait
parameters in children with motor disabilities using an electronic walkway system: assessment of reliability. Pediatr Phys Ther. 2007;19:326–331.
10. Bruton A, Conway JH, Holgate ST. Reliability: what is it, and how is it measured? Physiotherapy. 2000;86:94–99.
11. Portney LG, Watkins MP. Foundations of Clinical Research Applications to Practice. London, United Kingdom: Prentice Hall International (UK) Limited; 2000.
12. Keating J, Matyas T. Unreliable inferences from reliable measurements. Aust Physiother. 1998;44:5–10.