Elsevier

Gait & Posture

Volume 39, Issue 1, January 2014, Pages 354-358
Gait & Posture

Repeatability and validation of Gait Deviation Index in children: Typically developing and cerebral palsy

https://doi.org/10.1016/j.gaitpost.2013.08.001Get rights and content

Highlights

  • The repeatability and uncertainty of GDI (Gillette Deviation Index) were evaluated.

  • Repeatability coefficient obtained on GDI for typically developing children was ±10.

  • GDI is robust, not sensitive to the noise applied on its entries.

  • We found a moderate correlation between GDI and GMFCS (Gross Motor Function Classification System).

Abstract

The Gait Deviation Index (GDI) is a dimensionless parameter that evaluates the deviation of kinematic gait from a control database. The GDI can be used to stratify gait pathology in children with cerebral palsy (CP). In this paper the repeatability and uncertainty of the GDI were evaluated. The Correlation between the GDI and the Gross Motor Function Classification System (GMFCS) was studied for different groups of children with CP (hemiplegia, diplegia, triplegia and quadriplegia). Forty-nine, typically developing children (TD) formed our database. A retrospective study was conducted on our 3D gait data and clinical exams and 134 spastic children were included. Sixteen TD children completed the gait analysis twice to evaluate the repeatability of the GDI (test–retest evaluation). Monte Carlo simulations were applied for all groups (TD and children with CP) in order to evaluate the propagation of errors stemming from kinematics. The repeatability coefficient (2SD of test–retest differences), obtained on the GDI for the 16 TD children (32 lower limbs) was ±10. Monte Carlo simulations showed an uncertainty ranging between 0.8 and 1.3 for TD children and all groups with CP. The Spearman Rank correlation showed a moderate correlation between the GDI and the GMFCS (r = −0.44, p < 0.0001).

Introduction

Given the different gait patterns and pathologies in children with cerebral palsy (CP), a global analysis is essential in clinical practice. For this purpose, estimation of gait deviations from normative values is required and is helpful to improve therapeutic interventions.

In the CP pathology, three dimensional (3D) gait analysis for ambulant children and clinical evaluations are commonly used as assessments and are part of the international standard of CP care [1]. Even with 3D gait analysis, which is becoming an essential tool to assess ambulant children with CP [2], it is sometimes difficult to define objectively the amount of abnormalities and the degree by which an abnormal gait deviates from normal patterns. Three dimensional (3D) gait analysis provides a large amount of interdependent data and variables corresponding to different gait patterns [3]. The quantity and complexity of the data have pushed many authors to describe indices based on 3D gait analysis, developed primarily to evaluate clinical changes after a therapeutic intervention like the hip flexor index (HFI) [4] and the Gillette Gait Index (GGI) [5]. The GGI is calculated using discrete parameters incorporating 16 kinematic and temporal distance variables, chosen arbitrarily, allowing us to describe and to quantify the amount of pathology in an individual's gait pattern, and its repeatability has been evaluated [6]. These tools ignore the relation that exists between gait variables, contrary to other indices, described later, like the Gait Profile Score (GPS) [3] and the Gait Deviation Index (GDI) [7]. The GDI is an alternative to the GGI, measuring the subject's gait deviation from a normative database. It is calculated using kinematic variables, studied point by point, during the entire gait cycle. It is a scaled distance between 9 kinematic parameters of pathological gait and the average of 9 kinematic parameters of normal gait (group of typically developing (TD) children) [8]. The GDI is represented as a single number: when the number decreases, clinical involvement increases, and when the number increases, the gait profile is closer to a normal profile (≥100).

Many authors were interested in studying the correlation between indices calculated from the computerized gait analysis and clinical evaluation tools [9], [10], [11]. Molloy [12] studied the correlation between the GDI and clinical functional measures like the Gross Motor Function Measure (GMFM) and Gross Motor Function Classification system (GMFCS) and concluded that the GDI is a valid tool to describe motor impairments in CP.

However, for clinical use, it would be interesting to evaluate the repeatability of the GDI, by the test–retest method, since there is no previous study on this subject. Moreover, the method of calculation of the GDI involves kinematic curves and their transformation into vectors, matrices, euclidian distances and singular value decomposition [7]. Kinematics are subject to errors in 3D gait analysis and these errors can lead to uncertainties on the value of the GDI. It would be interesting to evaluate the propagation of errors during the calculation of the GDI.

The aim of this study is to evaluate the repeatability of the GDI within typically developing (TD) children. The error propagation during the calculation of the GDI was evaluated by applying Monte Carlo simulations on TD children and children with CP.

Section snippets

Samples

A retrospective study was conducted on our 3D gait data and clinical assessments performed between 2006 and 2012. One hundred ninety one (N = 191) children were referred to our gait laboratory for an orthopedic evaluation, an orthotic intervention or for a baseline gait assessment. Forty-nine (N = 49) TD children formed our asymptomatic database with a mean age of 10.3 years (SD = 3, minimum = 5 years, maximum = 15 years) [6]. One hundred thirty four (N = 134) children with the diagnosis of spastic

GDI for children with CP

The demographic characteristics of the children with CP and TD children are summarized in Table 1. The GDI was calculated separately for each lower limb. The GDI was normally distributed in each GMFCS level for children with CP and for TD children. The mean value of the GDI according to GMFCS levels is represented in Table 2. The mean value of the GDI decreases when the GMFCS increases. The histogram of the GDI is shifted toward 100 when the GMFCS decreases (Fig. 1 (left)). The mean values of

Discussion

In this study, the repeatability of the GDI for the 16 TD children who completed the exam twice was evaluated. The error propagation on GDI calculation for 49 TD subjects and 134 spastic children with CP was studied by applying Monte Carlo simulations. Correlation between the GDI and the GMFCS was assessed.

The test–retest study on TD subjects did not show any statistical differences between the mean values of the GDI in the first and second session (p > 0.05). This result gives further evidence

Conflict of interest statement

All authors disclose any financial and personal relationships with other people or organisations that could inappropriately influence this work.

Acknowledgment

The authors would like to thank Ziad El Bakouny for his assistance in the preparation for the manuscript.

Cited by (43)

  • Three decades of gait index development: A comparative review of clinical and research gait indices

    2022, Clinical Biomechanics
    Citation Excerpt :

    Statistically, GDI can be described by a normal curve, allowing the application of parametric models during data analysis (McMulkin and MacWilliams, 2015; Schwartz and Rozumalski, 2008). It correlates also with other functional scales, such as FAQ, GMFM-66, GMFCS and five-times-sit-to-stand-test (FTSST) (Ito et al., 2019; Massaad et al., 2014; Molloy et al., 2010; Schwartz and Rozumalski, 2008; Tsitlakidis et al., 2020). GDI, thus far, has established validity for spastic CP, rheumatoid arthritis, muscular dystrophy, idiopathic clubfoot, Batten disease, degenerative spinal pathologies, X-linked hypophosphatemia, cervical spondylotic myelopathy, Williams syndrome, knee and hip arthroplasty, knee osteoarthritis and post-stroke hemiparetic gait (Esbjörnsson et al., 2014; Galli et al., 2012a, 2012b; Guzik and Drużbicki, 2020; Haddas et al., 2021; Ito et al., 2020; Jensen et al., 2015; Lööf et al., 2019; Maanum et al., 2012; Mar et al., 2020; Mindler et al., 2020; Naili et al., 2019b, 2017b, 2017a; Ropars et al., 2016; Schwartz and Rozumalski, 2008; Sienko Thomas et al., 2010).

  • Gait deficits and dynamic stability in children and adolescents with cerebral palsy: A systematic review and meta-analysis

    2020, Clinical Biomechanics
    Citation Excerpt :

    The range of mean ages reported across all 68 studies on acaCP was 2 to 17 years, and for TDP was 3.2 to 15 years. The GMFCS levels of acaCP ranged from 1 to 3 across the 37 studies (Bakir et al., 2013; Barber et al., 2017; Bartonek et al., 2016; Bourgeois et al., 2014; Bulea et al., 2017; Cappellini et al., 2016; Choi et al., 2011; Degelaen et al., 2016; Delabastita et al., 2016; Eek et al., 2017; Goudriaan et al., 2018; Iosa et al., 2013; Ishihara and Higuchi, 2014; Kalsi et al., 2016; Klotz et al., 2014; Kruger et al., 2017; Kurz et al., 2012, 2015b; Malone et al., 2015; Malt et al., 2016; Massaad et al., 2014; Meyns et al., 2012a, 2016, 2014, 2017, 2012b; Pauk et al., 2016; Petersen et al., 2012; Rumberg et al., 2016; Saether et al., 2014; Tavernese et al., 2016; Taweetanalarp et al., 2011; Van de Walle et al., 2012; Wahid et al., 2015; Wallard et al., 2014; Wang et al., 2015; Zanin et al., 2018) which reported this measure. Overground walking was reported in 61 studies (Abel et al., 2003; Al-Abdulwahab and Al-Khatrawi, 2009; AlAbdulwahab, 2011; Bakir et al., 2013; Balaban et al., 2012; Barber et al., 2017; Bartonek et al., 2016; Bosmans et al., 2016; Bourgeois et al., 2014; Bruijn et al., 2011, 2013; Cappellini et al., 2016; Carriero et al., 2009; Chang et al., 2011; Chen et al., 2009; Choi et al., 2011; Degelaen et al., 2016; Delabastita et al., 2016; Eek et al., 2017; Feng et al., 2014; Goudriaan et al., 2018; Gross et al., 2013; Iosa et al., 2013; Ishihara and Higuchi, 2014; Kalsi et al., 2016; Kiernan et al., 2014; Kim et al., 2014; Klotz et al., 2014; Kruger et al., 2017; Kurz et al., 2015b,b; Langerak et al., 2008; Malone et al., 2015; Malt et al., 2016; Massaad et al., 2014; Meyns et al., 2012a, 2016, 2014, 2017; Mileti et al., 2016; Pauk et al., 2016; Rumberg et al., 2016; Saether et al., 2014; Steinwender et al., 2000; Svehlík et al., 2010; Tavernese et al., 2016; Taweetanalarp et al., 2011; Thompson et al., 1998; Tuzson et al., 2003; Van de Walle et al., 2012; Wahid et al., 2015; Wallard et al., 2014; Wang et al., 2015; White et al., 2005, 1999; Wong et al., 2004, 2005, 2000; Zanin et al., 2018; Zwick et al., 2010; Krzak et al., 2015; Meyns et al., 2012b) out of which one study (Tuzson et al., 2003) reported fast-walking speed, six studies (Bruijn et al., 2011; Delabastita et al., 2016; Kurz et al., 2015b; Meyns et al., 2012a, 2016, 2012b) reported both preferred and fast-walking speed conditions and one study (Gross et al., 2013) reported preferred, slow and fast-walking speed conditions.

View all citing articles on Scopus
View full text