Influence of normative data’s walking speed on the computation of conventional gait indices

Abstract

The pathology’s impact on gait pattern may be overestimated by conventional gait indices (Gillette Gait Index – GGI, Gait Deviation Index – GDI, Gait Profile Score – GPS), since impairments’ consequences on kinematics may be amplified by a change in walking speed. The objectives of this study were to evaluate the influence of walking speed on the computation of gait indices and to propose a corrective method to cancel the effects of walking speed. Spatiotemporal parameters and kinematics of fifty-four asymptomatic participants (30 M/24 W, 37.9 ± 13.7 years, 72.8 ± 13.3 kg, 1.74 ± 0.10 m) were collected at four speed conditions (C₁:[0,0.4] m s⁻¹, C₂:[0.4,0.8] m s⁻¹, C₃:[0.8,1.2] m s⁻¹, C₄:spontaneous). Four values of each index were computed for each trial using successively the four conditions as normative data repository. Mean values over all participants were statistically compared (paired t-tests, 95% confidence level). Indices values computed with normative at equivalent walking speed were not statistically different from reference values. Meanwhile, deviations appeared when the walking speed discrepancy between conditions and normative increased. These drifts related to walking speed mismatch have been quantified and fitting functions proposed. A correction was applied to indices. GGI was efficiently adjusted while GDI and GPS remain different from their reference values for C₁ and C₂. Gait indices must be interpreted cautiously in function of the normative data repository’s walking speed used for computation. Furthermore, a coupled use of conventional and corrected gait indices could lead to a better comprehension of the contribution of impairments and walking speed on gait deviations and overall gait quality.

Introduction

Nowadays, instrumented gait analysis is widely used to quantify movement patterns of individuals during walking, especially to understand the deficits related to a pathology with intricate gait deviations. This analysis provides joint kinematics, kinetics and ground reaction forces in three dimensions, and electrical muscular activity. However, the interpretation of this complex set of data is not trivial, and several gait indices have been introduced in the literature to summarise them and to assess treatment outcomes. The most common are the Gillette Gait Index (GGI) or Normalcy Index (Schutte et al., 2000), the Gait Deviation Index (GDI) (Schwartz and Rozumalski, 2008), and the Gait Profile Score (GPS) (Baker et al., 2009). The GGI, based on a principle components analysis, defines a distance between 16 discrete gait parameters (i.e. temporal, spatial and kinematic parameters) and averaged normative data (Schutte et al., 2000). Instead of using discrete variables, the GDI and GPS take into account 15 kinematic time-series along the whole gait cycle and tend to give a more overall measure of gait deviations (Baker et al., 2009, Schwartz and Rozumalski, 2008). In all cases, the computation of these indices is based on a comparison between gait characteristics of a participant and those of a normative data repository established on an asymptomatic population. These three indices have been validated and used in children with cerebral palsy (Baker et al., 2009, Massaad et al., 2014, Rasmussen et al., 2015), and to a lesser extent in children with various pathologies (McMulkin and MacWilliams, 2015, Romei et al., 2004). In adults, the GGI has been validated for an asymptomatic population (Cretual et al., 2010), and each of these indices has been used for various pathologies, such as spinal cord and brain injuries (GGI: (Cretual et al., 2010)), Parkinson’s disease (GDI and GPS: (Speciali et al., 2014)), spastic cerebral palsy (GDI: (Maanum et al., 2012)), and multiple sclerosis (GPS: (Pau et al., 2014)).

Whatever the investigated population, the spontaneous walking speed of participants (e.g. ranged between 0.18 and 1.03 m s⁻¹ for stroke (Olney et al., 1994)) is often slower than for asymptomatic participants (ranged between 1.04 and 1.60 m s⁻¹ (Salbach et al., 2015)). However, a modification of walking speed influence gait parameters, and several studies have described its effect on spatiotemporal parameters, kinematics, kinetics and muscle activity in asymptomatic children (Schwartz et al., 2008, van der Linden et al., 2002) and adults (Hanlon and Anderson, 2006, Kirtley et al., 1985, Kwon et al., 2015, Lelas et al., 2003, Murray et al., 1984). In particular, it has been shown that a decrease in walking speed implies a decrease in cadence and swing phase relative duration (i.e. expressed as percentage of gait cycle) (Kirtley et al., 1985, Murray et al., 1984, Schwartz et al., 2008), in the range of hip flexion–extension (Murray et al., 1984, Schwartz et al., 2008, van der Linden et al., 2002), in the maximum knee flexion during early stance and swing phase (Hanlon and Anderson, 2006, Kwon et al., 2015, Lelas et al., 2003, Schwartz et al., 2008, van der Linden et al., 2002), as well as in the maximum plantarflexion (Kwon et al., 2015, Schwartz et al., 2008, van der Linden et al., 2002). Gait indices, based on gait parameters or kinematic curves and being usually computed using normative data repositories established on asymptomatic participants walking at spontaneous walking speed (Baker et al., 2009, Romei et al., 2004), may thus also be influenced by walking speed. To our knowledge, only one study has been reported in a conference abstract to highlight the impact of the normative data’s walking speed on the computation of GDI in a child population (Rozumalski and Schwartz, 2012).

While an impairment highlighted during clinical examination may have a direct impact on the related joint kinematics, it can also affect walking speed and thus indirectly the kinematics of other joints. For example, in case of spasticity of the triceps surae, walking speed may be decreased willingly to avoid muscle spasms, and hip joint kinematics may thus be altered indirectly because of this reduced walking speed. Hence, since the impact of impairments on spatiotemporal parameters and kinematics may be amplified by the consequences of a change in walking speed, the impact of a pathology on gait may be overestimated by conventional gait indices. Not adjusting the walking speed of the two compared populations (i.e. pathological vs. asymptomatic) may thus influence the analysis of gait deviations, and lead to misinterpretations regarding the impact of the pathology on gait. Unfortunately, this levelling if often impossible in a clinical setting, since established normative data repositories has been obtained at a spontaneous walking speed (Pinzone et al., 2014), and the development of a new normative data repository requires a lot of resources and time. To overcome this issue, several authors have thus proposed some methods aiming to adapt existing normative data repositories for slow-walking participants. Lelas et al. proposed linear and quadratic regression equations between walking speed and 27 kinematic and kinetic parameters (Lelas et al., 2003). Hanlon and Anderson considered kinematic time-series along the whole gait cycle and confirmed correlations and linear regressions between walking speed and lower limb kinematics (Hanlon and Anderson, 2006). However, these correction methods have not been applied yet to the gait indices computation.

Based on a recently established data repository of asymptomatic participants walking at different speeds (Schreiber et al., 2016a, Schreiber et al., 2016b), the objectives of the present study were (1) to evaluate the impact of the normative data’s walking speed on the computation of GGI, GDI and GPS in asymptomatic adults and (2) to propose non dependent-velocity indices based on corrective methods of conventional indices to limit the effect of walking speed discrepancy.