The study included 50 patients at a chronic period after stroke (18 females, 32 males); mean age 60.9 ± 11.2 years (range 30-75); mean time from stroke 42 months (range 8-120); 15 patients with right hemisphere lesions, 35 patients with left hemisphere lesions. Inclusion criteria were: ischaemic stroke, time from stroke at least 6 months, independent walking (use of a cane, crutches or AFO orthosis was permitted), Brunnström recovery stage 3–4. Stroke was confirmed by computed tomography or magnetic resonance imaging. Exclusion criteria were: unstable haemodynamic state, peripheral vascular disease, cognitive impairment (Mini Mental Scale < 20) and mobility deficits significantly limiting and disrupting the patients’ ability to walk. Table 1 (in the Supplementary Files) describes the laboratory characteristics of study participants’ gait.
This prospective observational study was conducted among patients treated at Rehabilitation Clinic of Provincial Hospital No. 2 in Rzeszow, Poland. All qualified patients were fully informed about the procedure and signed informed consent to participate in the study. The research protocol was approved by the local Bioethics Commission of the Medical Faculty (5/2/2017) and the study was registered at the Australian New Zealand Clinical Trials Registry (ACTRN12617000436370).
Primary outcome measure: spatiotemporal and kinematic parameters of gait. The patients’ walking abilities were assessed at the Laboratory of Biomechanics of the Institute of Physiotherapy, University of Rzeszów. A 3-dimensional gait analysis was carried out using the SMART system (6 cameras, 120 Hz), manufactured by BTS Bioengineering (BTS Bioengineering, Milan, Italy). The internal protocol of the system (Helen Hayes (Davis) Marker Placement) was applied in selecting locations for reference markers; these were placed on the sacrum, pelvis (anterior posterior iliac spine), femur (lateral epicondyle, great trochanter and in lower one-third of the shank), fibula (lateral malleolus, lateral condyle end in lower one-third of the shank), foot (metatarsal head and heel) . The patients were asked to walk at a comfortable self-selected speed and were allowed to use auxiliary equipment, such as canes and elbow crutches, during the examination. During one trial, six passes of the patient were recorded. At the next stage spatiotemporal and kinematic parameters were calculated with the use of Tracker and Analyzer programs (BTS Bioengineering), averaging the results to a single session. The analysis took into account: 1) spatiotemporal parameters including stance time [s], stance phase (% of gait cycle), step length [m], stride length [m] of the paretic and of the non-paretic limb; 2) kinematic parameters including hip flexion/extension range of motion (Hip FE ROM) and knee flexion/extension range of motion (Knee FE ROM) of the paretic and of the non-paretic limb.
The video recording and 3D recording were carried out concurrently. Positioning of the two video cameras (BTS Vixta, BTS Bioengineering Corp.), working in synchronicity, was selected in such a way as to obtain images in the frontal and the sagittal plane. The walking path was 10 metres long. One camera was set in line with the direction of the gait in the frontal plane, the other camera, recording sagittal plane view, was positioned halfway along the walking path, two metres away from the path. The cameras were programmed to allow visualization of three walking trials examining the paretic and the non-paretic sides for a total of six ambulation trials. The subjects were asked to walk the specified distance at a self-selected (comfortable) speed, with the support of orthopaedic aids if used on a regular basis.
Secondary outcomes: The recordings and WGS-based gait assessment were reviewed and interpreted by a physical therapist with over 10 years of experience in working with patients post-stroke, and with expertise in using the WGS and interpreting the scores. The WGS allows to assess 14 observable gait parameters (as described in the Introduction above). The total score, in the range of 13.35 - 42 points, is calculated for all the items. The points assigned to items 2–10 and 12-14 are added up. Responses to items 1 and 11 are weighted by 3/5 and 3/4, respectively and then the points are added to the total score. Higher scores correspond to poorer overall walking performance and more visible gait deviations [5.8]. Good intra- and inter-rater reliability of the WGS was demonstrated by a number of studies [6,8,15-17].
The analyses took into account six gait symmetry indexes calculated based on 3D assessment involving 50 study participants. Measurement of gait symmetry applied the most commonly used method – absolute index proposed by Robinson , which is calculated as a quotient of the absolute difference between the measures for both legs and the mean of these measures, multiplied by 100. The absolute value of the difference between the affected and the unaffected side was taken into account because gait defects are reflected by the disparity between the results identified for both legs, regardless the fact whether higher result is found for the right or the left leg. Given the method applied in determining the symmetry index (it can only assume positive values), we can expect that its distribution will be concentrated around values approaching 0. Zero value of the above index reflects perfect symmetry . SI was calculated from the formula:
SI =2(xn - xi)/(xn + xi) * 100
where: 𝑥𝑛 – the value of the variable obtained from the non-paretic limb, 𝑥𝑖 – the value of the corresponding variable obtained from the paretic limb .
The results of measurements obtained in 3D assessment for the non-paretic and paretic limb as well as the specific SIs were presented in the form of descriptive statistics and histograms. Analysis of the correlations between 3D symmetry indexes and the specific components of as well as the total score in WGS-based assessment is shown in the form of a correlation matrix presenting values of Spearman's rank correlation coefficients. The strength of all the correlations were interpreted as: 0.3£|R|<0.5 low correlation; 0.5£|R|<0.7 moderate correlation; 0.7£|R|<0.9 strong correlation; 0.9£|R|<1 very strong correlation .
At the next stage, regression analysis was applied to investigate whether a simple observational tool may be a substitute to the time-consuming and costly 3D analysis. Symmetry indexes based on 3D assessment of the relevant parameters were adopted as dependent variables for the specific models, while scores in items 1-14 of the WGS were applied as independent variables. Subsequently stepwise regression with forward selection was applied to find a model combining two desired characteristics: it would only contain statistically significant factors and would most successfully describe variability of the indexes. Determined based on such calculations, a regression model for each 3D symmetry index allows to estimate its value using selected WGS scores .
Statistical significance was assumed to be p<0.05. Statistical analyses were conducted with the use of Statistica 10.0 program (StatSoft, Poland).