Concurrent related validity of the GAITRite® walkway system for quantification of the spatial and temporal parameters of gait

Abstract

The GAITRite® is a portable gait analysis tool for automated measurement of spatiotemporal gait parameters. Although frequently used for clinical and research purposes, the concurrent validity of GAITRite® has not been validated against a criterion measure. The aim of this experiment was to investigate the concurrent validity and test retest reliability of the GAITRite® carpet walkway system for quantification of spatial and temporal parameters of the footstep pattern. Twenty-five healthy adults aged 21–71 years (mean 40.5 years, S.D. 17.2) performed three walk trials at self-selected pace, three at fast pace and three at slow pace. For each trial, data were simultaneously collected from the GAITRite® and a Clinical Stride Analyzer®, which has established reliability and validity. At preferred, slow and fast walking pace there were very high correlations between the two measurement systems for gait speed (ICC (2,1)=0.99), stride length (ICC (2,1)=0.99) and cadence (ICC (2,1)=0.99). Correlations between the electronic carpet and the stride analyser were moderate to high for single limb support (SLS) time (ICC (2,1)=0.69–0.91) and weak for the proportion of the gait cycle spent in double limb support (ICC (2,1)=0.44–0.57). The reliability of repeated measures for the GAITRite® was good at preferred and fast speed for speed (ICC (3,1)=0.93–0.94), cadence (ICC (3,1)=0.92–0.94), stride length (ICC (3,1)=0.97), single support (ICC (3,1)=0.85–0.93) and the proportion of the gait cycle spent in double limb support (ICC (3,1)=0.89–0.92). The repeatability of the GAITRite® measures were more variable at slow speed (ICC (3,1)=0.76–0.91). These results indicate that the GAITRite® system has strong concurrent validity and test retest reliability, in addition to being a portable, simple clinical tool for the objective assessment of gait.

Introduction

Measurements of the spatial and temporal parameters of footstep patterns are frequently obtained to identify gait deviations, make diagnoses, determine appropriate therapy and monitor patient progress. Although visual observation is the most common method for assessing gait disorders, it has poor inter-rater and retest reliability [1], [2], [3] as well as poor criterion validity against kinematic analysis [4]. Timed measures of gait using a stopwatch have moderate to high reliability [5], [6], [7] yet limit data collection to mean values for gait speed, stride length and cadence. Measurement of footstep patterns using chalk [8], [9] and ink-pad methods [10], [11], [12] has the advantage of yielding data on step length and step width, yet is time consuming and cumbersome.

Clinicians have recently used commercially available gait measurement tools to quantify the spatial and temporal parameters of the footstep pattern [13], [14], [15]. Most commonly used is the Clinical Stride Analyser® (CSA). The CSA® consists of footswitches placed inside the shoes and attached to a portable data logger worn at the subject's waist. Studies on elderly adults [16] and people with movement disorders arising from stroke [13], [17], Parkinson's Disease (PD), [14], [18], [19], [20], [21], [22] and Huntington's Disease [23] have established high retest reliability, inter-rater reliability and predictive validity. For example, Morris et al. [20] examined the repeatability of CSA® measures of speed, cadence, stride length and percentage of the gait cycle in double limb support (DS% GC) in 15 subjects with PD and 15 matched control subjects. Repeatability was high for measures taken 30 min apart with intraclass correlation coefficient (ICC) values ranging from 0.92 to 0.99. Hill and colleagues [17] quantified the retest reliability of spatiotemporal parameters measured with the CSA® in 22 subjects with stroke. Data collected for two walking trials measured 5 min apart were highly consistent (ICC (2,1) 0.92–0.96). These and other studies [13], [14], [18] indicate that the CSA® can be used as a tool against which to judge the reliability and criterion related validity of other measurement tools.

There are two components of criterion related validity; (i) concurrent validity and (ii) predictive validity. Concurrent validity is established when simultaneous measurements are made by the tool to be validated and the criterion [24]. Predictive validity refers to the ability of the target test to predict the outcome of the criterion test [24]. The aim of this study is to examine the concurrent validity of the GAITRite® with the CSA® for the measurement of footstep patterns in adults. The GAITRite® system is a portable ‘carpet’ walkway that has pressure sensors embedded within its length. Subjects walk over the carpet without the encumbrance of wires or markers and data can be quickly and easily obtained for each step within the entire gait trial. Variables that can be quantified include walking speed, cadence, step length, single and double limb support duration and stride width for individual steps and the step sequence.

A single case report by McDonough et al. [25] explored the concurrent validity of GAITRite® measures of footstep parameters against chalked footsteps markings on the floor. Temporal parameters were validated against a hand-held stopwatch. For the spatial parameters, high ICC's were obtained for right and left step length (ICC (2,1)=0.97, 0.99, respectively) despite the potential for measurement error inherent in the process of using chalk to mark steps. Poor to moderate correlations were obtained between the GAITRite® and stopwatch measures of temporal parameters such as cadence (ICC (2,1)=0.31) and right step time (ICC (2,1)=0.67). A possible source of error from the hand-held stopwatch arises from subjective judgements of the timing of gait cycle events, as well as errors introduced by delays in stopwatch activation.

In a similar study, Selby-Silverstein and Besser [9] compared GAITRite® data with powdered footstep patterns obtained from 22 strides. Although ICC values of 0.99 were reported for step and stride length, the mean absolute error was 0.51 cm for step length and 0.67 cm for stride length. This level of error probably arises from error inherent in attempting to measure from chalk imprints. In the same study Selby-Silverstein et al. correlated outputs from the GAITRite® with the Parotec® system, an insole pressure measurement tool. Two adults with diabetes mellitus were instructed to take different sized steps, yielding moderate correlations between the two systems for the measurement of step and stride time (ICC (2,1)=0.83, 0.77, respectively). The lower than expected correlations may have arisen because of the difference in the temporal resolution of the GAITRite® and the pressure insole system. The differences between the systems in switch and closing time may have reduced the strength of the correlations.

Cutlip et al. [26] recently correlated GAITRite® footstep measures with those obtained on the Peak Performance Technologies Motus 3.1®, a video based movement analysis system. Ten healthy adults were tested simultaneously on both systems at preferred, fast and slow speeds. Step duration and stance and swing duration were strongly correlated (r=0.97, 0.98, 0.94, respectively). The correlation for step length was also high at r=0.93, despite the potential for parallax error due to using only a single camera positioned parallel to the walkway [26]. Whereas the Peak® system calculated spatiotemporal data from a single stride, the GAITRite® calculated data over the entire length of a 4.8 m electronic walkway. The intra-subject variability in the subject footstep patterns over these distances might have reduced the strength of correlation between a single stride and the mean value of several strides. In the presence of intra-subject variability the true variability between the measurement devices cannot be calculated.

The concurrent validity of the spatial and temporal measures of gait obtained by the GAITRite® remain inconclusive because the sample sizes in previous studies have been small (one to ten subjects) and the criterion measures, such as chalked footsteps, have not always been appropriate. The aim of this study is to further evaluate the concurrent validity of the GAITRite® in a large adult sample using the CSA® as the criterion measure. The CSA® was chosen as a criterion measure, because it has been widely used for the measurement of spatiotemporal parameters of gait. It was hypothesised that data would demonstrate validity of the GAITRite® in the measurement of spatiotemporal parameters with a clinically acceptable level of measurement error.