Backward vs. Forward Gait Symmetry Analysis Based on Plantar Pressure Mapping

Abstract

Symmetry is one of the factors analysed in normal and pathological gaits. Backward gait is an area of interest to scientists, in terms of its physiology and therapeutic possibilities. This study aimed to analyse the symmetry of the pressure parameters of backward gait in comparison to forward gait using different symmetry indices. Eighty-one healthy people aged between 19 and 84 years took part in the study. Foot pressure distribution was analysed during forward and backward gaits at self-selected speeds. Mean and maximum pressure values were calculated after dividing the foot into four or ten areas. Delta, Ratio Index, Robinson Index, Gait Asymmetry, and Symmetry Angle were calculated for each area, separately for both forward and backward gaits. Higher ratios of asymmetry were found during backward than during forward gait. Larger ratios of asymmetry were found within toes II–V, forefoot, metatarsals I, II, and III, medial and lateral heel areas. No significant correlation between symmetry indices and age or BMI was found. Results suggested that the lower symmetry of backward gait is caused by a higher number of corrective movements that allow for the maintenance of body balance and global symmetry of gait. This can be realised by increased cortical control of the backward gait, which was a new movement task for all participants.

1. Introduction

Gait is a basic means of human locomotion and also one of the most common human activities. It has a complicated, three-dimensional, and cyclic nature. According to Voughan [1], any type of walking act (normal and pathological) has two mandatory requisites: movement periodicity and sufficient ground reaction forces, supporting the body. In the case of healthy individuals, we should observe symmetrical kinetic and spatiotemporal parameters. On the other hand, pathological gait reveals some asymmetrical movement patterns. Nevertheless, we cannot identify gait as simply as symmetrical = normal, asymmetrical = pathological. The problem is more complicated, and in many cases a pathological gait can be more symmetrical than the normal one. Prosser et al. [2] reported that symmetry ratios for the gait of healthy and cerebral palsy children were not statistically different between those two groups for any examined spatiotemporal measure. Seminati et al. [3] showed a positive correlation between anatomical and dynamical asymmetry in runners. It should also be noted that healthy human movement is not fully repeatable [4] and symmetrical [5]. From a mathematical point of view, we can call the normal gait quasi-repeatable (quasi-periodical) and quasi-symmetrical, whereas some short-time alterations are natural. Those “imperfections” are the results of interactions between many body segments, muscle groups, and their activation, and activity of a central nervous system. The human musculoskeletal system is redundant and operates in a changeable environment. Nevertheless, gait asymmetry is often used as one of the indicators of pathological gait or to monitor the rehabilitation process and clinical improvements [6]. It must be underlined that the level of asymmetry imposed in the literature does not exist, the exceeding of which a gait can be termed as pathological [7].

In addition to forward gait, areas of interest for researchers include backward, sideways, or crossed gait. By analysing the backward gait and comparing it with the forward gait, it has been observed that there are many points of convergence in these two activities. Some variables in backward gait can be considered identical to those seen in forward gait, only reversed in time [8]. Meyns et al. indicate that interlimb coordination is the same in walking in both directions [9,10]. However, there are also fundamental differences between these two activities. According to Grasso et al. [11] and Minetti and Ardigò [12], the anatomical and functional asymmetry between the anterior and posterior aspects of the limb may impose biomechanical constraints in forward and backward gait. Researchers highlight the challenging nature of backward gait activity, indicating that there could be a different spinal mechanism or different functional networks that control backward gait [13], showing greater activity within sensorimotor cortices in backward gait [14,15]. Considering the spatial and temporal parameters, backward gait is characterised by a lower speed, reduced stride length, lower step frequency, shorter swing phase, longer duration of the gait cycle, and a longer double support phase when compared to forward gait [11,16,17]. Minetti and Ardigò [12] also pointed out that backward walking has a higher metabolic cost and requires higher degree of mechanical work, both internally and externally.

The data sources used in gait analysis can be divided into kinematic and kinetic. The former concerns joint and body segments positions, velocities, accelerations, walking speed, step or stride length, gait phase duration, and many different aspects, mainly examined by optoelectronic or IMU systems [1]. The latter concerns forces and torques present in the musculoskeletal system or that act upon it. These parameters are mainly examined with the use of force or pedobarographic plates [1] or special force-sensing insoles [8]. Most of the data assessed during instrumented gait analysis, regardless of the system used, can be analysed using different symmetry indices.

Nowadays, symmetry indices are mainly focused on the magnitude and localisation of the asymmetry. The most popular are so-called discrete indices: Ratio, Robinson indices, or Symmetry Angle [9] or Gait Asymmetry [18] are some of the common choices among researchers. Meyns et al. [9] also pointed out the existence of other types of indices: complete gait cycle approach (in the form of trend symmetry, geometric properties of cyclograms region of deviations or symbol-based methods); statistical approaches; and nonlinear approaches (in the form of multiresolution entropy or cross-fuzzy entropy). Discrete types of indices are calculated on a single value, chosen from a cycle, whereas other methods provide a more complex view of the analysed phenomena, but require more values/measurements to create valuable results. Walsh and Taylor [19] analysed forward and backward gait symmetry using a harmonic ratio calculated from accelerations in the lumbar region. Apart from this publication, no other studies related to forward and backward gait symmetry analysis were found.

This study analysed the symmetry of pressure parameters of backward gait in comparison to forward gait using different symmetry indices. This study was performed to determine if there are significant asymmetries in forward or backward gait that would differentiate these two types of gaits. Such differentiation can be helpful in further studies of normal and pathological gaits, e.g., in the clinical assessment or identification of the early stages of gait abnormalities that are not discernible in forward gait due to its automaticity.

2. Materials and Methods

2.1. Participants

Eighty-one healthy people aged between 19 and 84 years (mean 39.8 ± 16.9 years) took part in the study. The mean bodyweight of the subjects was 73.4 ± 14.5 kg, and the mean height was 171.5 ± 9.7 cm. The mean BMI score of the subjects was 20.93 ± 2.32 kg/m².

A person was not qualified to the study if at least one of the following exclusion criteria were found: post-traumatic injuries, falls, or the presence of coexisting diseases or surgical procedures that may significantly affect the locomotor system in the last twelve months; current complaints of pain on the part of the locomotor system or other complaints which could influence the result of the examination; poor state of health in the participant’s own opinion, which would make the examination impossible; and lack of consent of the participant for the examination.

2.2. Measurement Procedures

For this study, shoe insoles F-Scan Wireless measurement system (Tekscan Inc., South Boston, MA, USA) was used. Data were collected using F-Scan VersaTek insoles with a resolution of 3.9 sensels per cm² with a frequency of 100 Hz (Figure 1). A calibration procedure was performed using step procedure separately for each participant according to the protocol of Hsiao et al. [20] and the manufacturer’s guidelines. The test consisted of walking a distance of 10 m forward and backward three times along a predefined path. The tested person had to walk a designated distance with a comfortable speed while looking forward. In addition, subjects were asked not to turn their head while walking backward. After two trial runs to familiarise the test person with this type of gait and the measuring equipment, measurements were recorded during the third run. The average result was calculated from the 8–10 gait cycles, thus reducing any measurement error due to the different gait speeds. The first and the last steps were excluded from the analysis.

2.3. Calculation Methods

To calculate the analysed results from the F-Scan device, a custom script was developed in the interactive Matlab environment (version R2016b; MathWorks, Natick, MA, USA). This software allowed for the efficient calculation of selected parameters for the entire database of raw data collected during the investigation. The algorithm used in this script can be divided into a few steps. The first step in the data analysis was to segment the recording into individual steps. For this purpose, instantaneous pressure was calculated for each frame of the video from the F-scan by summing the pressures across the foot matrix. Based on the time course of the instantaneous pressure, segmentation into individual cycles was performed. The gait cycle was then divided into stance and swing phases. The results were calculated from the stance phase. The foot area was then divided into more detailed segments. We decided to divide the foot into 4- and 10-segment templates (Figure 2) [21,22,23]. A segmentation template was created for each foot. It was created on a view of the foot composed of the sum of all frames of the recording. Three main lines divided the foot area into four regions: toes (T1–5), metatarsal bones (M1–5), midfoot (MF), and heel (H), according to the level of pressure intensity in the foot image. The heel area (H) was then divided by a line passing through the geometric centre of the heel in the sagittal axis into medial (MH) and lateral (LH) parts. The metatarsals/forefoot area (M1–5) was divided into five parts (the 1st metatarsal bone—M1, the 2nd metatarsal bone—M2, the 3rd metatarsal bone—M3, the 4th metatarsal bone—M4, the 5th metatarsal bone—M5). The toes area was divided into the big toe area (T1), designated as 1⅓ the width of the first metatarsal bone, and toes 2–5 (T2–5). The procedure was controlled by the data analyst, who manually corrected the automatic division when it was incorrect due to different foot types, thus reducing the error in the results.

Two types of parameters were calculated for this study: mean and maximum pressure values from each area of the template mentioned above, for forward and backward gait direction. These calculations were made for each foot and gait direction separately.

The following symmetry indices were calculated:

• Delta (Δ):

$$\mathsf{\Delta }=X_{\alpha }-X_{\beta }$$

(1)

where α is the value for the non-dominant leg, whereas β is the value for the dominant leg.

This parameter gives a general overview of the symmetry/asymmetry of the observed values. In the interpretation, Δ = 0 means ideal symmetry, |Δ| > 0, asymmetry. This index is the most basic one. Its main disadvantage is the lack of an upper limit.

The next four indices described below were calculated from the respective area of the left and right foot is denoted as X_α and X_β, where, ∀α,β X_α > X_β. It means that in this study we analysed symmetry/asymmetry, without taking into consideration the side (left/right or dominant/non-dominant).

2.

Modified Ratio Index (RI):

$$RI=\left(1-\frac{X_{\beta }}{X_{\alpha }}\right)100\%$$

(2)

The original version of the index is: $RI=\frac{X_{\alpha }}{X_{\beta }}$, thus presented modification gives results in percentage. The value RI = 0 informs about perfect symmetry, |RI| > 0 about the level of asymmetry. Originally, this index was used for amputees’ gait asymmetry [24,25]. According to Viteckova et al. [18], the disadvantage of this index is, similarly to the delta index, the lack of limitations of the results and a difference in symmetry results can be observed depending on the proportion of nominator and denominator.

3.

Symmetry Index (SI, also called Robinson Index):

$$SI=\frac{2\left|X_{\alpha }-X_{\beta }\right|}{X_{\alpha }+X_{\beta }}100$$

(3)

The Robinson Index can be treated as a measure of asymmetry, giving information in positive values. The value SI = 0 is considered as an indicator of symmetrical gait when SI > 0 means asymmetry. According to Viteckova et al. [18] this index is currently the most popular index used and cited in publications devoted to gait symmetry. It is variable specific, which is its main disadvantage [18]. Thus, it should not be used as one criterion value to assess gait asymmetry when working with more than one variable.

4.

Gait Asymmetry (GA):

$$GA=ln\left(\frac{X_{\beta }}{X_{\alpha }}\right)100\%$$

(4)

This formula is a logarithmic transform of the original RI factor presented in percentage values. According to Viteckova et al. [18], this index is counted as a modified RI index. In the study of Plotnik et al. [26] this version of the RI index was used to calculate asymmetry in kinematic parameters of swing time duration. In the interpretation, GA = 0 means symmetry of examined parameters, whereas GA > 0 its asymmetry.

5.

Symmetry Angle (SA):

$$SA=\frac{45°-\arctan \left(\frac{X_{\beta }}{X_{\alpha }}\right)}{90°} ·100\%$$

(5)

The Symmetry Angle index is a measure for discrete values obtained for examined parameters. It is related to the angle formed when a one-side value is plotted against a second-side value and compared to the line of symmetry [6]. It can be interpreted as follows: SA = 0 indicates perfect symmetry, SA = 100% indicates that the two examined values are equal but opposite in magnitude. This index was described by Zifchock et al. [6] and it was used to bypass limitations of classical SI index, especially reference value. Symmetry Angle index has a standard scale for the interpretation of the asymmetry. Nevertheless, according to Viteckova et al. [18] a high level of correlation between SA and SI is observed.

The above-described symmetry indices were chosen because of the wide data spectrum to be analysed. We tried to find different localisation of the asymmetries that could be detected by different symmetry indices. Additionally, they are quite often used for gait analysis and each of them gives slightly different information about gait asymmetry [27]. Symmetry Index and Symmetry Angle were previously used to analyse symmetry of pressure parameters during forward gait [22].

2.4. Statistical Analysis

The statistical analysis was conducted using PQ Stat v. 1.8.2 (PQStat Software, Poznań, Poland). The significance level was set at 0.05. The normality of data was tested using the Shapiro–Wilk test, which shows that most of the calculated parameters do not meet the requirements for normality. For this reason, the Wilcoxon matched–pairs test was applied to detect differences between forward and backward gait. To identify the areas with the highest and the lowest asymmetries Friedman’s ANOVA test with post hoc Dunn test with Bonferroni correction were applied. In the ANOVA test, each foot segmentation template was tested separately. Afterwards, we tried to find a dependence between asymmetry and age or BMI. For this reason, mean indices were calculated for each person, template, and gait direction. The mean index was correlated with age or BMI using Spearman rank coefficient and k-means cluster analysis.

2.5. Ethical Approval

The study protocol was approved by the Bioethics Committee of the Medical University of Warsaw (No. KB/227/2015). The study was conducted according to the ethical guidelines and principles of the Declaration of Helsinki.

3. Results

3.1. Delta

In most of the analysed areas, the direction of symmetry was consistent in the direction of forward and backward gait. The exceptions were mean values in the following areas: toes (negative values during forward gait which means higher values in the dominant leg; positive values during backward gait which means lower values in dominant than non-dominant leg); and forefoot and heel (minor negative values in forward gait, positive values in backward gait), although these differences were not statistically significant (Figure 3a). Statistically significant differences were found in maximum values of pressure within the forefoot (p = 0.0349) and midfoot (p = 0.0188) (Figure 3b).

The highest asymmetries were found for mean pressure values in metatarsal bone V, forefoot, midfoot, lateral, and medial heel for both forward and backward gait, as well as in the heel area in backward gait only. Delta of maximum pressures in heel area was also significantly higher than in other areas during backward gait. The lowest asymmetries were found in toes I–V, toes II–V and metatarsal bone IV areas for mean pressure values in both walking directions as well as in the heel area for forward gait only. Asymmetries in maximum pressure values were the lowest in toes II–V in forward and backward gait. In forward gait only, the lowest asymmetries were found in forefoot and heel areas, whereas in backward gait only, the highest symmetry was observed in metatarsal bone IV and toes I–V areas.

3.2. Modified Ratio Index

Generally, we found lower values of RI during forward than during backward gait. The exceptions were mean and maximum values of forefoot V and midfoot areas. Significant differences between forward and backward gait were found in mean values of pressure within following areas: toes II–V (p = 0.0195), metatarsal bone III (p = 0.0008), metatarsal bone IV (p = 0.0325), forefoot (p = 0.0047), lateral heel (p = 0.0119), medial heel (p = 0.0424) and heel (p = 0.0001) (Figure 4a). In maximum pressure values there were significant differences in RI in metatarsal bone I (p = 0.0303), II (p = 0.0349), III (p < 0.0001) and forefoot (p < 0.0001) (Figure 4b).

The highest asymmetries were found in metatarsal bone I and V in both mean an maximum pressure values in both types of gait and additionally in toe I and toes II–V for backward gait. The lowest asymmetries were found in metatarsal bone III, forefoot and heel, also in both types of gait for mean and maximum pressure values.

3.3. Symmetry Index (Robinson Index)

Lower values of SI were found in forward than in backward gait except for mean and maximum values from metatarsal bone V and midfoot areas. Significant differences between forward and backward gait were found for mean pressure values in the following areas: toes II–V (p = 0.0176), metatarsal bones III (p = 0.0015) and IV (p = 0.0210), forefoot (p = 0.0037), lateral heel (p = 0.0073), medial heel (p = 0.0299) and heel (p = 0.001) (Figure 5a). SI of maximum pressure values were different for backward vs. forward gait in following areas: metatarsal bones I (p = 0.0337), II (p = 0.0444) and III (p < 0.0001), as well as forefoot (p < 0.0001) (Figure 5b).

The highest asymmetries in mean and maximum pressure values were observed in metatarsal bone I and V in both forward and backward gait and additionally in toes areas (divided into I and II–V) only for backward gait. The highest symmetry was observed in the forefoot area as well as the heel area for both forward and backward gaits.

3.4. Gait Asymmetry

Gait asymmetry was lower in forward than in backward gait except for mean and maximum values in tarsus. We found significant differences between forward and backward gait in GA from mean pressure values for following areas: toes II–V (p = 0.0174), metatarsal bones III (p = 0.0018) and IV (p = 0.0185), forefoot (p = 0.0036), lateral heel (p = 0.0066), medial heel (p = 0.0272) and heel (p = 0.0001) (Figure 6a). Gait asymmetry from maximum values was significantly higher in backward than in forward gait in the areas: metatarsal bones I (p = 0.0353), II (p = 0.0469), III (p < 0.0001) and the entire forefoot (p < 0.0001) (Figure 6b).

The lowest asymmetries were observed in metatarsal bones III and IV as well as in forefoot and heel areas for both mean and maximum pressure values in both forward and backward gait. The highest asymmetries were found in toes area divided into I and II–V areas, metatarsal bones I and V as well as in midfoot in both forward and backward gait for mean and maximum values of pressure.

3.5. Symmetry Angle

We found significant differences between forward and backward gait in the symmetry angle calculated from mean pressure values for following areas: toes II–V (p = 0.0183), metatarsal bones III (p = 0.0012) and IV (p = 0.0241), the entire forefoot (p = 0.0039), medial heel (p = 0.0325), lateral heel (p = 0.0082) and the whole heel area (p = 0.0001) (Figure 7a). Symmetry Angle from maximum pressure values was significantly different in backward vs. forward gait in following areas: metatarsal bones I (p = 0.0341), II (p = 0.0419) and III (p = 0.0001), as well as the entire forefoot (p < 0.0001) (Figure 7b).

The symmetry angle from both mean and maximum pressure values was significantly higher in the toes area divided into I and II–V, metatarsal bones I and V, and the midfoot in both forward and backward gait. The lowest asymmetries were observed in the forefoot area and heel area in both types of walking.

3.6. Correlation Analysis

We found no significant correlation between symmetry indices and age or BMI, neither in Spearman’s rank correlation nor in cluster analysis.

4. Discussion

Analysis of the backward gait and, especially in comparison to forward gait, showed many points of convergence in these two activities, although there are also some major differences between them [8,11,28]. The purpose of this work was to analyse symmetry in pressure parameters of backward gait in comparison to forward gait using different symmetry indices.

The present study shows higher asymmetries during backward than during forward walking. Błażkiewicz et al. [27] confirm that slight asymmetry may reflect functional differences between limbs in propulsion and control during gait. The symmetry of a movement-like gait is an indicator of functional recovery and shows that the movement task is controlled by subcortical brain regions [8,29,30]. Zych et al. [31] provided evidence that forward and backward walking is governed by the same control and adaptation mechanisms. Because the backward gait is for most people a new movement task, it is not automated [32] and requires more control from frontoparietal cortical regions similarly to gait with dual-task [8,15,30], greater activity of the muscles [11], and is not repeatable [19,28]. Wafai et al. [23] suggest that plantar pressures asymmetry can be caused by the fact that consecutive steps taken even during forward gait are not perfectly repetitive and may cause local asymmetries. Lack of symmetry in individual regions may be part of a compensatory mechanism that allows for the maintenance of global symmetry during gait. This can be confirmed partly in the present study, where some asymmetries were observable only in small regions (e.g., delta in lateral heel and medial heel), whereas in bigger areas (the entire heel) the asymmetries were no longer noticeable. Walsh and Taylor [29] showed no effect of walking direction on the harmonic ratio calculated from accelerometer data for all three axes, but they analysed only global symmetry at the level of L5. For a new movement task, like walking backward, the number of corrective movements that allow for the maintenance of global symmetry and dynamical stability is relatively high and causes higher ratios of asymmetry in small regions than during forward gait. The poorer stability of the backward gait was confirmed by Hoogkamer et al. [33]. Bollens et al. [8] showed that backward gait induces significantly higher mean cadences and a greater magnitude of stride duration fluctuations. Additionally, Kastavelis et al. [28] reported higher gait variability regarding the joint range of movement and stride intervals in backward gait. These factors can lead to greater asymmetries during gait. Greater asymmetries in backward walking can also be caused by the lower functional performance of muscles. The musculoskeletal system structure is optimised for forward gait and walking in the opposite direction forces both ligamental and muscular structures. Guadagnin et al. [34] suggest that older adults with impairment in functional performance, like gait, may also have bilateral asymmetries in muscular performance. Furthermore, there is a lack of investigations addressing the relationship between muscular and functional parameters and gait asymmetries in adults and therefore, this hypothesis can neither be confirmed nor rejected.

Asymmetries found in the present study had different localisations depending on the symmetry index that was used, although many of them were similar. In a comparison of backward to forward gait, higher asymmetries were found within the following areas: toes II–V (RI, SI, GA, SA), the entire forefoot (all indices), metatarsals I, II, and III (RI, SI, GA, SA) as well as in medial and lateral heel (RI, SI, GA, SA) areas. Greater asymmetries in these areas can be caused by higher COP excursions appearing during backward gait. Chisholm et al. [35] have shown that patients with greater sensorimotor impairments showed more asymmetric COP excursions, which are also related to changes in plantar dynamics. Mandell et al. [13] pointed out that the cumulative asymmetric pressure loading on the foot could be a risk factor for stress injuries in the foot.

Higher asymmetries found in the heel (medial and lateral), metatarsals (especially I, II, and III) as well as in toes II–V areas can suggest that the COP trajectory during backward gait moves back and forth between these areas and therefore is not smooth nor repeatable. Backward gait requires the involvement of the forefoot, instead of the heel, at the beginning of the stance phase. Even slight differences in foot angle between the right and left foot can indicate loading asymmetries in these areas. The mechanism of asymmetries shown in the heel area can be analogous, as this area is loaded mostly at the end of the stance phase while walking backward. This can be indirectly confirmed by Fritz et al. [17] who reported a larger base of support while walking backward. On the other hand, the anatomy of the foot and the entire lower limb is highly asymmetrical along the anteroposterior axis, and therefore imposes different biomechanical constraints on backward and forward gait [11]. Also muscle activation patterns are different [31]. These can be reasons for the absence of typical transition between terminal stance and pre-swing phases during backward gait, which can lead to high asymmetries observed in the above-mentioned areas. It is worth emphasizing that this transition is very important in generating propulsion force, which influences gait efficiency [36,37]. The different pattern of gait phases during backward gait can therefore lead to the occurrence of minor injuries of some ligamental or muscular feet structures, pain, or even major pathologies [38]. Detailed biomechanical analysis on the relation between the biomechanics of anatomical structures and mechanics of the backward gait is not possible due to the lack of studies.

In the present study, we found no dependence between age and pressure symmetry indices, although the age range in the study group was relatively large. The symmetry index in the study of Kobayashi et al. [39] was affected by age. Fritz et al. [17] also reported interactions between age and variability of gait. It must be pointed out that the above-mentioned studies analysed spatial and temporal parameters, whereas we measured plantar pressure distribution. The symmetry of plantar pressure distribution can be more dependent on an individual’s foot structure [40] than on age. McKay et al. [41] confirmed that gait speed is similar for adolescents and adults (from 10 to 59 years). This can suggest that gait pattern remains quite steady in this range of ages, if the person stays healthy, and, therefore, that no relationship between symmetry indices and age can be found.

We also found no correlation between the symmetry of mean and maximum pressure values and BMI. Telfer et al. [42] stated, that peak plantar pressures at forefoot and midfoot regions can depend on BMI, although Said et al. [40] found no relation between physiological characteristics, including body weight or height and types of foot posture. We can conclude that even if peak pressures are higher in people with higher BMI, the symmetry of pressures can depend more on foot structure than on BMI.

There are several indices described in the literature which calculate symmetry. However, it seems that none of them is precise enough. The assessment of the usefulness of the symmetry indicators was not the aim of this study, but we noticed that results obtained using Ratio Index, Symmetry Index, Gait Asymmetry and Symmetry Angle were similar. Wafai et al. [23] confirmed the high correlation between SI and SA in plantar pressure symmetry analysis. Błażkiewicz et al. [27] have shown that usefulness of the SI, RI, and GA indices is similar in the assessment of symmetry of gait in healthy subjects. They also pointed out the clear advantage of RI and SI over SA in assessing spatio-temporal gait parameters. On the other hand, Zifchock et al. [6], state, that SA does not require a reference value to be selected, as is the case in SI, and can be a good substitute for SI. In the present study, it is not possible to point out any particular advantages or disadvantages of these four symmetry indices. Regardless of the preference of choosing one of the indicators, it seems unreasonable to calculate them all in plantar pressure symmetry analysis.

Some limitations of this study should be acknowledged. Firstly, the magnitude of the asymmetry can be influenced by gait speed. In the present study, participants walked forward and backward with self-selected comfort speeds, which could be different with respect to gait direction. Chisholm et al. [35] reported higher asymmetries correlated with higher walking speeds in patients after strokes. Also, Riskowski et al. [43] observed in higher speeds larger asymmetries inground reactions forces considering different foot regions. We did not analyse walking speed in the present study, but we can expect that the speed of backward gait without visual feedback was lower than when walking forward. Still, asymmetries found in backward gait were higher than in forward gait. On the other hand, forcing a slower speed while walking forward or a faster speed while walking backward could also change asymmetry due to strained walking speeds. Our goal was to analyse natural movement and therefore, we decided to confer the choice of gait speed to each participant. Secondly, we analysed five different symmetry indices, but the analysis showed that results obtained in Symmetry Index, Robinson Index, Gait Asymmetry, and Symmetry Angle were similar. Therefore, it would be justified to analyse only one of them. Due to the constraints of this study, we analysed only mean and maximum pressure values. It would be worthful to include other parameters, such as pressure-time integral or force-time integral as well as spatio-temporal parameters in further studies. A complex analysis of the symmetry of the backward gait should also include kinematics of the lower limbs and trunk as well as an assessment of kinetic parameters.

5. Conclusions

The present study showed lower symmetry ratios of backward compared to forward gaits, which can be caused by the higher number of corrective movements that allow for the maintenance of body balance and global symmetry of gait. Greater asymmetries found in toes II–V, metatarsals, and heel areas can be caused by the higher centre of pressure excursions that appear during backward gait. Both mechanisms could be caused by increased cortical control of the backward gait, which was a new movement task for all participants. Symmetry analysis showed that the results obtained in Symmetry Index, Robinson Index, Gait Asymmetry, and Symmetry Angle are similar. Therefore, it is suggested that only one of the symmetry indices be selected in further studies. Subsequent research should also include an analysis of the symmetry of the kinematic and kinetic parameters, as well as pressure-time indicators.