Repeatability in the assessment of multi-segment foot kinematics

Abstract

A recently published systematic review on 3D multi-segment foot models has illustrated the lack of repeatability studies providing evidence for appropriate clinical decision making. The aim of the current study was to assess the repeatability of the recently published model developed by Leardini et al. [10]. Foot kinematics of six healthy adults were analyzed through a repeated-measures design including two therapists with different levels of experience and four test sessions.

For the majority of the parameters moderate or good repeatability was observed for the within-day and between-day sessions. A trend towards consistently higher within- and between-day variability was observed for the junior compared to the senior clinician. The mean inter-session variability of the relative 3D rotations ranged between 0.9–4.2° and 1.6–5.0° for respectively the senior and junior clinician whereas for the absolute angles this variability increased to respectively 2.0–6.2° and 2.6–7.8°. Mean inter-therapist standard deviations ranged between 2.2° and 6.5° for the relative 3D rotations and between 2.8° and 7.6° for the absolute 3D rotations. The ratio of inter-therapist to inter-trial errors ranged between 1.8 and 5.5 for the relative 3D rotations and between 2.4 and 9.7 for the absolute 3D rotations. Absolute angle representation of the planar angles was found to be more difficult.

Observations from the current study indicate that an adequate normative database can be installed in gait laboratories, however, it should be stressed that experience of therapists is important and gait laboratories should therefore be encouraged to put effort in training their clinicians.

Introduction

Stereophotogrammetry has been used for two decades to represent shank and foot multi-segment kinematics. Review of the literature demonstrates that several 3D multi-segment foot models (3DMFMs) have been published [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], embodying the international consensus that the foot should be modeled as a number of segments [11]. However, a recent review highlighted a lack of standardization as well as adequate repeatability studies [12]. As a consequence, clinical utility of several 3DMFMs still has to be determined.

Leardini et al. [10] recently described an interesting 3D multi-segment foot model (Leardini Foot Model: LFM) protocol. Noteworthy differences with some other 3DMFMs are the specific marker placement at the calcaneus, the indirect calculation of landmarks (by mid-point calculation), the fact that the midfoot is considered as one rigid segment and the calculation of several planar angles. The latter two components enforce the clinical relevance of the LFM, especially with respect to foot pathologies [8], [12].

With respect to the LFM protocol, reference waveforms for the stance phase of gait have been published for a group of 10 subjects, illustrating ‘typical’ patterns for all 3D rotations and planar angles [10]. Furthermore, the effect of walking cadence on the kinematic waveforms [13], as well as repeatability indices from a repeated measure design have been reported [14]. Results of the latter indicated that most of the 3D rotations can be measured in a repeatable way by experienced clinicians.

Although several studies contributed to an improved understanding of the LFM protocol, it should be stressed that a full picture of the clinical features has not yet been provided in literature. Moreover, kinematic waveforms have only been published for the stance phase of gait despite the fact that other studies have demonstrated the relevance to report the swing phase [15], [16], [17]. Still, replication of the study by a group independent from the authors of the LFM protocol is lacking, thereby limiting the strength of the conclusions on the reproducibility of the protocol [18]. Finally, intrinsic and extrinsic errors have only been quantified for the 3D rotations and not for the planar angles. Knowing consistency and variability of these planar angles is helpful, as it may guide the use and interpretation [19].

Consequently, the aim of the present study was to further explore the kinematic waveforms obtained from the LFM [10], to provide further insight in the repeatability of the protocol in the presence of different expertise levels, as well as to quantify the repeatability of all parameters (3D rotations and planar angles).