Reliability of 3D gait data across multiple laboratories
Introduction
Gait studies performed on the same individual in presumably identical circumstances do not necessarily yield identical results. These differences may be attributed to errors inherent in every test procedure. Factors that affect the outcome of a gait study cannot all be completely controlled. Variability in 3D gait data can come from both intrinsic and extrinsic factors. Extrinsic factors that add to this variability come from differences between data collection sessions and differences between researchers. For example, extrinsic factors may result from differences in positioning the reflective markers [1]. Intrinsic factors that affect error may include intra-individual variations that arise naturally from trial-to-trial and subject-to-subject, neither of which can be reduced [2]. For example, age and walking speed can both contribute to the natural variability of gait between subjects [3], [4], [5]. In order to assess these factors, several walking trials are typically collected during the same session to provide a measure of the data variability. In healthy subjects it is generally thought that five trials are sufficient to obtain reliable data [6], [7].
Many different factors may contribute to the total variability in gait data including the operator, equipment used, and equipment calibration. Therefore, it is not surprising that recent studies have found that the largest sources of extrinsic variability come from tests across different laboratories. Noonan et al. [8] tested 11 patients with cerebral palsy at four separate gait analysis laboratories and found that at least 5% of gait cycle could vary among laboratories by at least 21°. The authors also found a 27° average difference in static range of motion across laboratories. Most importantly, the results of the gait analysis generated considerably different treatment recommendations across four sites. Gorton et al. [9] evaluated one healthy subject at twelve Shriner's Hospitals and found an average maximum difference of 15° across sites. Variability was also high between subsequent sessions at a single site with 5 of 8 gait parameters significantly different. Most recently, Benedetti et al. [10] assessed the consistency of one healthy subject's gait at seven different laboratories and concluded that joint angles and moments were generally consistent, with the exception of hip and knee joint rotations, but there were systematic errors in the data across sites, particularly in hip and knee powers. These studies raise questions about comparing gait data between different laboratories and, although some of the results are promising, concerns about inter-lab reliability still remain.
Data reliability is of utmost importance when attempting to share data across sites to evaluate patient progress or to pool research data for comparative effectiveness studies. Therefore, the goal of this study was to collect consistent 3D gait data across three motion analysis laboratories. Prior work has identified 5° as an upper end criterion of minimal detectable change for almost all kinematic measures [20]. There is not an established criterion for kinetic data. We hypothesized that with appropriate calibration of the equipment and uniformity of the underlying biomechanical model it would be possible to obtain reliable measurements. The results of this study are important because measurement consistency across sites participating in multi-center research studies directly influences required sample size, level of detectable difference, and statistical power.
Section snippets
Methods
Ten subjects (six male/four female, age 30 ± 6 years, BMI 24 ± 4 kg/m2) without any neuromuscular impairment were studied. All subjects were screened by a licensed physical therapist. Inclusion criteria were individuals between the ages of 10 and 45 years who had normal muscle strength and range of motion of the lower extremities with no neuromusculoskeletal deficits. Individuals were excluded from the study if they had a major orthopedic injury such as a meniscal tear, ligament tear, or recent
Results
The inter-trial, inter-session, and inter-laboratory errors in lower extremity kinematics and kinetics were calculated for every point in the gait cycle for each subject (Fig. 2). Reliability of each variable was expressed by the standard deviation of its measurement. The inter-trial error represents the variability of the subject's gait and serves as an appropriate reference for comparisons. The inter-session error was larger than the inter-trial error. The inter-lab error was larger than the
Discussion
The study results demonstrate that reliable gait data can be collected across multiple gait laboratories. In this study we compared the lower extremity kinematic and kinetic data collected on 10 healthy subjects who were studied in three gait laboratories. The three laboratories had different hardware configurations, marker configurations, and examiners that could have contributed to inconsistencies across the laboratories. The possible sources of variability were not controlled, simulating a
Disclaimer
The views expressed in the article are those of the authors and do not reflect the official policy of Department of the Navy, Department of the Army, Department of Defense, or the US Government.
Acknowledgements
Funding was provided by DOD 731743-1, DOD Defense Health Programs/Center for Rehabilitative Sciences Research, Grant # HU0001-11-1-0004. There are no commercial relationships which may lead to a conflict of interest with any of the authors.
References (30)
- et al.
Increasing the number of gait trial recordings maximises intra-rater reliability of the CODA motion analysis system
Gait Posture
(2007) - et al.
Measurement and management of errors in quantitative gait data
Gait Posture
(2004) The evolution of clinical gait analysis part III—kinetics and energy assessment
Gait Posture
(2005)The reliability of kinetic and kinematic variables used to analyse normal running gait
Gait Posture
(2001)- et al.
Intra-rater and inter-rater reliability of gait measurements with CODA mpx30 motion analysis system
Gait Posture
(2003) - et al.
Assessment of the kinematic variability among 12 motion analysis laboratories
Gait Posture
(2009) - et al.
Inter-laboratory consistency of gait analysis measurements
Gait Posture
(2013) - et al.
A proposed test to support the clinical movement analysis laboratory accreditation process
Gait Posture
(2003) - et al.
The accuracy of estimating proximal tibial translation during natural cadence walking: bone vs. skin mounted targets
Clin. Biomech.
(2003) - et al.
A comparison of the accuracy of several hip center location prediction methods
J. Biomech.
(1990)
Prediction of hip joint center location from external landmarks
Hum. Mov. Sci.
Repeated measures of adult normal walking using a video tracking system
Gait Posture
Repeatability of gait data using a functional hip joint centre and a mean helical knee axis
J. Biomech.
Reliability and minimal detectible change values for gait kinematics and kinetics in healthy adults
Gait Posture
The comparison of normative reference data from different gait analysis services
Gait Posture
Cited by (48)
Impact of deep brain stimulation on gait in Parkinson disease: A kinematic study
2024, Gait and PostureLaboratory based assessment of gait and balance impairment in patients with progressive supranuclear palsy
2021, Journal of the Neurological SciencesCitation Excerpt :This is an essential first step to gain a broad objective understanding of gait and balance impairment in PSP. While motion analysis labs are not present at every center, a high degree of reproducibility exists between labs [45] and the accurate data obtained in labs can be used to validate motion sensing devices that capture a more limited number of motion metrics. This study has important limitations.
Reliability and sources of variability of 3D kinematics and electromyography measurements to assess newly-acquired gait in toddlers with typical development and unilateral cerebral palsy
2021, Journal of Electromyography and KinesiologyInter-session repeatability of markerless motion capture gait kinematics
2021, Journal of BiomechanicsCitation Excerpt :The use of markerless motion capture to measure joint kinematics removed the reliance on skin-mounted markers, thereby removing the need for an experienced examiner to identify anatomical landmarks and accurately place markers and a dedicated laboratory, reduced the total data collection time, allowed subjects to wear the clothing of their choice, and resulted in kinematic data that was reliable between sessions on separate days. Inter-trial variability was generally larger in this study (average = 2.5°) than those previously reported (1.0°-2.4°), but by relatively small margins of 0.1°-1.5° and subjects’ session mean waveforms were found to be consistent between sessions despite the greater inter-trial variability (Caravaggi et al., 2011; Kaufman et al., 2016; Manca et al., 2010; Schwartz et al., 2004). The average inter-session variability measured in this study (2.8°) was the smallest across all other studies (3.0–3.6°), indicating that measuring gait kinematics across multiple sessions using markerless motion capture introduces less variability compared to marker-based motion capture systems.