The effects of textured insoles on quiet standing balance in four stance types with and without vision

A convenience sample of healthy young adults (n = 28; 13 females; mean age = 26.86 ± 6.6 yrs.; height = 171.82 ± 9.46 cm; weight = 73.22 ± 17.02 kg) volunteered, who were aged 18-51 years [1], right-footed, injury-free and with an adult shoe size (3–12, UK). Participants were excluded if they had neurological or musculoskeletal disorders, needed mobility aids, were injured or pregnant, had communicable foot diseases, or were unable to provide informed consent. Ethical approval was granted by the School of Health and Social Care Research Governance and Ethics Committee at Teesside University. Informed consent was obtained from all participants prior to testing.

The study involved a three-factorial repeated measures design comprising of insole type (textured vs. smooth), stance type (bipedal vs. standard Romberg vs. tandem Romberg vs. unipedal), and vision type (EO vs. EC), administered in a fully-randomised order. The dependent variable was movement of the centre of pressure (COP) during 30s of quiet standing, comprising five parameters: mean sway range (mm) and standard deviation (SD) in both anterior-posterior (AP range, APSD), and medial-lateral (ML range, MLSD) directions, and overall mean sway velocity (mm/s) of the AP and ML data.

Two types of prefabricated insole were used, which have been investigated in previous studies [7, 12]. The upper surface of the TIs comprised small pyramidal peaks with centre-to-centre distances of approximately 2.5 mm (Evalite Pyramid EVA, 3 mm thickness, shore value A50; Algeos UK ltd.), cut to a range of men’s and women’s UK shoe sizes. The smooth insole (SI) acted as the control, having a completely flat upper surface (medium density EVA, 3-mm thickness, shore value A50; Algeos UK Ltd., Liverpool, UK). During the testing procedures the insoles were worn within a standardised shoe (rubber sole with upper canvas). Participants wore thin standardised socks for hygiene purposes.

Quiet standing balance was assessed using two Kistler force plates (Model: 9286AA, Kistler Instruments Ltd., Hampshire, UK), placed side-by-side and combined to make one plate (approximate 5 mm space between plates) via an integrated charge amplifier (Model: DAQ 5691, 16ch, Kistler Instruments Ltd.), sampling at 50 Hz. Two force plates were needed due to the tandem stance length (heel-to-toe) [17].

Leg dominance was assessed by watching participants kick a football [18]. Each condition (n = 16) consisted 5 × 30s balance trials, producing 80 trials per participant. Stance position was standardised on the force plates for each participant. During bipedal standing, participants were instructed to “stand in a position comfortable for you”. In standard Romberg participants stood with the medial arches of the feet together. The tandem Romberg stance was completed heel-to-toe, right foot forward [17]. Unipedal standing was performed on the dominant leg, since no differences exist between legs in this population [19]. During all stance types participants were instructed to stand as still as possible. During EO, participants looked at a target on a wall 3 m ahead of the force plates [7]. Trials commenced once participants adopted the standardised position (trunk erect, arms by sides, lower limbs extended). For EC trials participants closed their eyes beforehand. There was a 2 min seated break between stance conditions. All data was collected within a single testing session to assess acute effects of insole usage. Acute effects are important and commonly used because they provide the opportunity to assess the body’s initial response to changes in sensory input during balance (c.f., [3‐5, 7, 9]). They also afford the groundwork for future studies to assess longitudinal effects. They also afford the groundwork for future studies to assess longitudinal effects. In addition, safety of participants must be considered. By assessing acute effects in a laboratory setting we can monitor participants responses to an unfamiliar device, prior to exploring any longitudinal effects, whereby participants are not within the constraints of a supervised laboratory setting.

All COP variables were extracted from the force platform using Bioware software (Bioware V5.3, Kistler). All data processing was performed off-line using a commercial software package (MATLAB 2016a, The MathWorks Inc., Natick, MA). Since literature suggests 20–30s of data collection is sufficient to assess quiet standing [20], the minimum trial duration included for analysis was 20s. In bipedal and standard Romberg all participants successfully completed 30s quiet standing balance. In tandem Romberg and unipedal stances all participants completed 20–30s, with the majority of trials lasting 30s (n = 96.1 and 71.4%, respectively). Removed trials involved one foot stepping out of the required stance.

A three-factorial repeated measures ANOVA was used to investigate the factors of insole type (smooth vs. textured), stance type (bipedal vs. standard Romberg vs. tandem Romberg vs. unipedal) and vision type (eyes open vs. eyes closed) in each of the five COP variables. Analyses were conducted using SPSS Statistics 23 (IBM, Chicago, IL, USA). We adjusted for any violation of the homogeneity of variance assumption using the Greenhouse–Geisser correction. Alpha levels were set to 0.05, and effect sizes were calculated as partial eta squared values (\( {\eta}_p^2 \)). Simple main effect analyses with Fishers LSD were used as post-hoc tests.

The main effect of insole was significant in the APSD variable, F(1, 27) = 4.38, p = 0.045, \( {\eta}_p^2 \) = 0.14. Post-hoc tests revealed APSD was significantly reduced in the textured, compared to the SI condition (6.92 ± 2.16 mm vs. 7.14 ± 2.51 mm, respectively; mean difference of 0.22 mm, 95% CI: 0.004 to 0.44). Simple main effect analyses investigating each of the four stance types in the APSD data showed postural sway was only significantly reduced in the textured compared to the smooth condition in the bipedal stance, F(1, 27) = 5.84, p = 0.02, \( {\eta}_p^2 \) = 0.18 (4.29 ± 1.32 mm vs. 4.8 ± 2.29 mm; mean difference of 0.51 mm, 95% CI: 0.076 to 0.935). While this pattern was replicated in both the standard and tandem Romberg stances, the associated alpha levels were not significant, see Table 1. More focused simple main effect analyses on the bipedal data revealed postural sway was significantly less for the textured compared to smooth insole in the EC condition, F(1, 27) = 4.68, p = 0.04, \( {\eta}_p^2 \) = 0.15 (4.88 ± 6.62 mm vs. 5.46 ± 6.6 mm; mean difference of 0.58 mm, 95% CI: 0.03 to 1.14), and was close to significant in the EO condition, F(1, 27) = 3.41, p = .076, \( {\eta}_p^2 \) = 0.11 (3.71 ± 4.12 mm vs. 4.14 ± 1.79 mm; mean difference of 0.43 mm, 95% CI: -0.047 to 0.901). The main effect of insole type was not significant in the other four COP measures: AP range, F(1, 27) = 0.78, p = 0.39, \( {\eta}_p^2 \) = 0.03, MLSD, F(1, 27) = 0.03, p = 0.87, \( {\eta}_p^2 \) < 0.01, ML range, F(1, 27) = 0.80, p = 0.38, \( {\eta}_p^2 \) = 0.03, and overall mean velocity, F(1, 27) = 1.71, p = 0.20, \( {\eta}_p^2 \) = 0.06.

The main effect of stance type was significant in each of the five measures: APSD, F(1.57, 42.29) = 54.42, p < .001, \( {\eta}_p^2 \) = 0.67; AP range, F(1.61, 43.43) = 63.56, p < 0.001, \( {\eta}_p^2 \) = 0.70; MLSD, F(2.45, 66.22) = p < 0.001, \( {\eta}_p^2 \) = 0.93; ML range, F(2.41, 65.18) = 263.8, p < 0.001, \( {\eta}_p^2 \) = 0.91; overall mean velocity, F(3, 81) = 116.91, p < 0.001, \( {\eta}_p^2 \) = 0.81 (see Fig. 1). In almost all cases postural sway significantly increased between each stance type in the following order: bipedal < standard Romberg < tandem Romberg < unipedal. This trend repeated but did not reach significance in overall mean velocity for: bipedal vs. standard Romberg, tandem Romberg vs. unipedal, and in the MLSD data for tandem Romberg vs. unipedal. By exception, postural sway in ML range was greater in tandem Romberg vs. unipedal.

The main effect of vision type was significant in each of the five measures: APSD, F(1, 27) = 87.8, p < 0.001, \( {\eta}_p^2 \) = 0.77; AP range, F(1, 27) = 116.17, p < 0.001, \( {\eta}_p^2 \) = 0.81; MLSD, F(1, 27) = 782.6, p < 0.001, \( {\eta}_p^2 \) = 0.97, ML range; F(1, 27) = 430.76, p < 0.001, \( {\eta}_p^2 \) = 0.94; overall mean velocity, F(1, 27) = 160.73, p < 0.001, \( {\eta}_p^2 \) = 0.86. In each case postural s way was greater during EC compared to EO with the majority of the associated alpha levels < p = 0.001, those results that were above this threshold were non-significant (see Table 2).

This two-way interaction was significant within each COP measure: APSD, F(2.01, 54.27) = 143, p < 0.001, \( {\eta}_p^2 \) = 0.46; AP range, F(1.57, 42.45) = 23.37, p < 0.001; \( {\eta}_p^2 \) = 0.46, MLSD, F(3, 81) = 211.91, p < 0.001, \( {\eta}_p^2 \) = 0.89; ML range, F(3, 81) = 66.48, p < 0.001, \( {\eta}_p^2 \) = 0.71); overall mean velocity, F(2.12, 57.27) = 85.23, p < 0.001, \( {\eta}_p^2 \) = 0.76. Post-hoc results reflected the general finding that while removing vision produced postural disturbances in all four stance types, this effect was stronger in those stances with inherently larger postural sway (i.e., unipedal > tandem Romberg > standard Romberg > bipedal). See Table 2. All other two-way and three-way interactions within each measure were not significant.