Medial heel wedge in balance training improves stability and function in chronic ankle instability

Abstract

Chronic ankle instability (CAI) impairs balance and increases reinjury risk. This randomized trial assessed whether adding medial heel wedges to a 4-week progressive balance program produced greater gains than balance training alone. Twenty-two participants with CAI were randomly assigned to train with or without medial heel wedges. Postural control was assessed during various standing tasks on different surfaces with eyes open and closed. Functional performance was measured using the modified star excursion balance test (mSEBT), the single-leg hop test (SLH) (with subsequent calculation of the limb symmetry index, LSI), and the single-leg squat. Both groups improved on most balance and functional measures. The intervention group showed greater gains in static postural control—especially single-leg stance on a soft surface with eyes open (p < 0.05)—and in mSEBT anterior reach (p = 0.004; η²=0.186). LSI, SLH, and single-leg squat improved within both groups with no between-group difference (p > 0.05). Progressive balance training is highly effective at improving balance and function in CAI, while the addition of medial heel wedges provides a selective, additional benefit for dynamic forward reach and static stability under demanding conditions. Because post-tests were performed without wedges, the between-group differences are consistent with neuromuscular adaptation, although the influence of mechanical carryover or learning effects cannot be entirely excluded. Heel wedges may be a simple adjunct in CAI rehabilitation and may potentially contribute to reduced reinjury risk; this requires prospective follow-up.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-32287-1.

Introduction

Chronic ankle instability (CAI) is a common condition that leads to impaired postural control, which is essential for maintaining balance and stability^1,2. Postural control relies on the central nervous system processing sensory information to generate motor responses that help maintain upright posture³. The ankle joint contains mechanoreceptors that send signals to the central nervous system to support balance, especially during movement. After an ankle sprain, these receptors may be damaged, disrupting sensory input and reducing the body’s ability to respond to balance challenges, contributing to ongoing instability⁴. Individuals with CAI often show delayed proprioceptive responses and weakened neuromuscular control, making it harder to recover from postural disturbances^1,4 and increasing the risk of further injury¹.

Both proprioception and postural control are commonly impaired in people with CAI². Xue et al. (2024) found that individuals with CAI have significant deficits in static postural control during single-leg stance, reflected by altered center of pressure (COP) trajectories⁵. Dynamic postural control is also compromised, as shown by reduced reach distances in the star excursion balance test⁶. Earlier studies by Freeman et al. (1965) demonstrated that balance exercises reduce proprioceptive deficits after ankle injury^7,8. Similarly, a meta-analysis by Wikstrom et al. (2009) confirmed that balance training improves postural control in individuals with CAI⁹. Various studies continue to report that balance training improves postural control in this population^10–12. These exercises, typically closed-chain, may enhance sensory-motor pathways, improve center of mass control, increase gamma motor neuron activity, boost muscle activation, and promote co-contraction of muscles—factors that collectively enhance balance^13,14. Traditional balance programs often include single-leg stance exercises on both stable and unstable surfaces. However, these programs may not sufficiently challenge the sensorimotor system to produce noticeable improvements in postural control. A more effective strategy may involve a balance training program that progressively increases task complexity, incorporating dynamic balance challenges such as sudden direction changes, landing from a hop, or reaching while standing. This approach, adapted from McKeon et al. (2008), combines dynamic balance components with progressive difficulty to enhance sensorimotor adaptation in individuals with CAI. McKeon et al. (2008) found that four weeks of progressive balance training significantly improved self-reported function and both static and dynamic balance¹¹. Mettler et al. (2015) reported a COP shift from anterolateral to posterolateral direction after progressive balance training, suggesting repair of sensorimotor pathways¹². Anguish et al. (2018) also reported improvements in self-reported function, dynamic control, and joint position sense following four weeks of progressive training¹⁵. Recent research by Park and colleagues (2025) demonstrated superior functional outcomes when implementing select components of the progressive balance training protocol compared to conventional balance exercises in CAI patients¹⁶.

Foot orthoses, including braces and insoles, represent another intervention for improving postural control in CAI patients¹⁷. Medial heel wedges have been proposed to alter rearfoot alignment and redistribute plantar pressures so as to shift COP trajectories and increase plantar contact area¹⁸. Mechanically, this can limit excessive inversion and promote a more neutral rearfoot alignment; sensorimotor, increased plantar contact and local pressure changes may enhance mechanoreceptor stimulation and proprioceptive feedback, facilitating central adaptation during balance training¹⁷. Based on these mechanisms, medial wedges may act synergistically with progressive balance exercises by providing both mechanical support and augmented somatosensory input, accelerating neuromuscular reorganization in CAI. In addition, Evidence from other lower-limb conditions supports the biomechanical relevance of heel wedging. For instance, Parkes et al. (2013) reported that lateral wedge insoles can meaningfully modify joint loading in individuals with medial knee osteoarthritis, demonstrating that small angular changes at the heel can alter lower-limb mechanics¹⁹. Similarly, Lu et al. (2022) found that heel lift insoles significantly affected lower-extremity muscle activation patterns and joint work during barbell squats²⁰. These findings indicate that heel wedges can influence both alignment and neuromuscular activation, providing a rationale for their potential application in improving postural control among individuals with CAI.

To situate our approach within the wider field of rehabilitation technologies, it is worth noting that advanced assistive devices such as lower-limb rehabilitation exoskeletons have been developed to augment sensorimotor recovery through active actuation and adaptive control strategies; however, these devices act at the level of gross limb assistance and control, and their mechanisms are mechanistically distinct from localized plantar and rearfoot interventions such as medial heel wedges. Recent work on exoskeleton control and adaptive admittance strategies illustrates advances in device-level rehabilitation but addresses different clinical problems and mechanisms than orthotic modulation of plantar mechanoreception and rearfoot alignment^21,22.

While balance training and foot orthoses independently improve postural control in CAI, their synergistic effects remain unexplored. Interestingly, post-intervention assessments in this study were conducted without wedges to evaluate lasting neuromuscular adaptations, which distinguishes this study from previous works. Recent studies have also introduced robotic and exoskeleton-based gait interventions to improve lower-limb function^21,22 but their high cost and technical complexity limit their practicality compared to simpler approaches such as medial heel wedges.

The primary outcomes of this study were postural control parameters, assessed under different conditions, and performance on the modified Star Excursion Balance Test (mSEBT). The secondary outcomes included the single-leg hop test, the limb symmetry index (LSI) calculated from the hop test, and single-leg squat performance.

Combining progressive balance training with medial heel wedges may amplify sensorimotor adaptation by simultaneously enhancing proprioceptive input and challenging dynamic stability. The present study aimed to evaluate the effect of progressive balance training combined with medial heel wedge insoles on postural control in individuals with CAI. We hypothesize that progressive balance training combined with medial heel wedges will result in greater improvements in static and dynamic postural control and functional performance than balance training alone.

Method

Participants

This single-blind randomized controlled trial recruited individuals with CAI through convenience sampling. Sample size was determined a priori using G*Power software (version 3.1), based on pilot data for COP velocity. Assuming an alpha of 0.05, power of 0.80, and an effect size of 1.10, a minimum of 11 participants per group was required. To account for potential dropouts, 27 individuals were initially screened. Three were excluded due to pregnancy (n = 1), ACL tear (n = 1), tibial fracture (n = 1), leaving 24 participants who were randomly assigned to two groups using permuted block randomization (block size = 4, 1:1 ratio) via Random Allocation Software. Allocation concealment was ensured using sequentially numbered, opaque sealed envelopes prepared by an independent researcher not involved in recruitment or assessment. Participants were assigned to one of two groups: (1) progressive balance training with a medial heel wedge (n = 12), or (2) balance training alone (n = 12). The 4-week intervention was completed by 22 participants (11 per group), with one withdrawal from each group due to personal reasons unrelated to the intervention.

Eligibility criteria were based on the International Ankle Consortium guidelines²³. Participants were required to have a history of at least one significant ankle sprain, occurring at least 12 months prior, that resulted in pain, swelling, or temporary loss of function. Additionally, they must have experienced at least two episodes of the ankle “giving way” within the six months preceding the study. Further inclusion criteria included a Cumberland Ankle Instability Tool (CAIT) score of 25 or less²⁴ and Foot and Ankle Ability Measure (FAAM) scores below 90% on the Activities of Daily Living (ADL) subscale and below 80% on the sports subscale²⁵. Exclusion criteria comprised mechanical ankle instability (confirmed via anterior drawer and talar tilt tests), visual, hearing, or somatosensory impairments that could affect balance, lower limb injury or surgery within the past year, or participation in any rehabilitation program during the six weeks prior to the study (Fig. 1).

Fig. 1.

Flow chart of the study.

Procedure

The study received ethical approval from the Ethical Review Committee of Tehran University of Medical Sciences, Iran (IR.TUMS.FNM.REC.1400.236) on March 14, 2022, and was conducted in accordance with the Declaration of Helsinki as well as relevant institutional guidelines and regulations. The trial was prospectively registered in the Iranian Registry of Clinical Trials (IRCT20220212054006N1) on March 26, 2022 (available at https://irct.behdasht.gov.ir). All participants provided written informed consent after receiving detailed study explanations, including consent for data and image publication. The trial maintained complete adherence to the protocol, with all 22 participants successfully completing the study period. Both groups took part in a 12-session progressive balance training program, conducted three times per week over four weeks. The control group followed only the progressive balance training program, while the intervention group did the same program but wore custom insoles—designed by an orthotist—with a 4-degree medial heel wedge. All assessments were conducted in two sessions: one the day before the first training session and the other the day after the final training session. Outcome assessor was blinded; participants were not fully blinded due to tactile differences, but they were not informed of the study hypotheses or which group constituted the intervention.

Progressive balance training program

The progressive balance training followed the protocol described by McKeon et al. (2008)¹¹.

(A) Single-limb hops to stabilization (10 repetitions per direction): participants performed 10 controlled hops in each of four directions—anterior/posterior, medial/lateral, anterolateral/posteromedial, and anteromedial/posterolateral—using distances of 18, 27, or 36 inches. After each hop, they had to stabilize on one leg before returning in the opposite direction. Advancement required completing 10 error-free repetitions. Error criteria included touching down with the non-stance limb, trunk lean > 30°, removing hands from hips during restricted arm use, bracing non-stance leg on stance leg, and missing the target (Fig. 2).

Fig. 2.

Directions and distances (in inches) for hop to stabilization activities.

(B) Hop to stabilization and reach (five repetitions): participants performed hop-to-stabilization sequences requiring single-limb stance stabilization before returning to the starting position, with each repetition involving hopping, stabilizing, reaching back to start, then hopping again toward the target. Advancement through seven progressive levels demanded five consecutive error-free repetitions per stage, with errors defined as either failed hop-to-stabilization or substantial reliance on the reaching leg for support. The progression included: 18-inch hops (arms free then hands-on-hips), 27-inch hops (same arm variations), 36-inch hops (with both arm conditions), terminating in a 36-inch hop from a 6-inch platform to maximize challenge. This method carefully increased the challenge by adjusting distance, arm position, and height, while requiring participants to meet strict accuracy standards before moving forward.

(C) Unanticipated hop to stabilization: participants stood in the middle of a nine-marker grid (Fig. 3). A sequence of numbers was read for the participants. Each number corresponded to a target position to which they would hop. As the progression of numbers changed, participants would hop to the new target position. The hop to stabilization rules were applied for this activity; however, in this case, participants were allowed to use any combination of hops (AP, ML, AM/PL, or AL/PM) they desired to accomplish the goal of getting through the sequence error-free. As a participant developed proficiency, the amount of time per move was reduced. In each session, participants performed three sequences of numbers.

Fig. 3.

Nine marker grids for unanticipated hop to stabilization.

Levels of unanticipated hop to stabilization: levels progress by reducing hold times from 5 s (Level 1) to 3 s (Level 2) and then 1 s (Level 3). At Level 4, if the subject completes all moves within 1 s without errors, a foam pad is added to one number while maintaining the same time constraint; failure results in stepping back a level. Level 5 adds a step to another number after error-free completion of Level 3 with one foam pad. If progression continues without errors, Level 6 introduces a second foam pad (two foam pads and one step), and Level 7 adds another step (two foam pads and two steps). Errors include touching down with the opposite limb, excessive trunk motion (> 30° lateral flexion), removing hands from hips during hands-on-hips tasks, bracing the non-stance limb against the stance limb, or missing the target. Each sequence of numbers was random such as 9, 7, 1, 6,4, 5, 3, 8, 2.

(D) Single-limb stance activities: participants performed three repetitions of single-limb stance activities in two categories—eyes open and eyes closed—each with seven progressively harder levels. For eyes open, tasks included standing with arms across the chest on a hard floor (60 s), then on a foam pad for increasing times (30, 60, 90 s), followed by medicine ball throws (20 throws over 30, 60, and 90 s). Eyes closed tasks started with arms out on a hard floor (30 s), progressing to arms across chest on hard floor (30 and 60 s), then on a foam pad with arms out (30 s) or across chest (30, 60, 90 s). Advancement required completing three consecutive error-free repetitions at each level. Errors were defined as touching down with the opposite limb, excessive trunk motion (over 30° lateral flexion), removing arms from across the chest when required, or bracing the non-stance limb against the stance limb.

Insole with medial heel wedge: the intervention group underwent progressive balance training with an insole featuring a 4-degree medial heel wedge placed at the rear foot. This wedged insole was used bilaterally and its application was restricted to the training sessions only. The insole was semi-rigid, consisting of a one-millimeter artificial leather top layer and a three-millimeter cow leather bottom layer. The medial heel wedge was made of polyethylene (Fig. 4. Top).

Fig. 4.

Top: Custom insole with a 4-degree medial heel wedge. Middle: Participant performing the star excursion balance test. Bottom: Static postural control assessments.

Assessment of outcome measures

All balance and functional outcomes were assessed under standardized conditions. Before formal data collection, participants completed practice trials for each task to become familiar with the procedures. Failed trials were defined as attempts in which the participant lost balance, touched the ground with the non-supporting limb, or failed to reach or land in a controlled manner. Such trials were immediately repeated after a short rest period, and only successful attempts were included in the final analysis.

Static postural control: static balance was measured using a Bartech 9090-15 force plate (Bartech Corp, Columbus, OH) under eight different conditions, combining double-leg and single-leg stances on both hard and soft surfaces, with eyes either open or closed (Fig. 4. Bottom). Participants stood barefoot, maintaining a stable position for 35 s during double-leg stance conditions and for 10 s during single-leg stance conditions. They were instructed to look at a fixed point four meters ahead, with arms resting at their sides. For single-leg stances, the non-supporting leg was held at about 30° of hip flexion and 45° of knee flexion. Each condition was repeated three times with a one-minute rest between trials, and the average of the three trials was used for analysis.

Dynamic postural control: dynamic postural control was evaluated using a modified version of the SEBT because it provides a dynamic measure of postural control and functional reach in multiple directions, offering a more comprehensive assessment of balance performance beyond static standing tasks²⁶. Participants stood on their involved leg at the center of the test platform, aligning the toes with the starting line. They reached with the opposite leg in three directions—anterior, posteromedial, and posterolateral while keeping their balance (Fig. 4. Middle). Each direction was tested three times, with a 30-second rest between attempts. The average of the three trials per direction was recorded for analysis.

Functional assessment: functional performance was assessed using the single-leg hop test and the single-leg squat test. Participants performed the single-leg hop test by completing three maximal forward hops on each leg, keeping their hands by their sides and without taking any steps beforehand. The distance was measured from the starting line to where the heel landed. After three practice hops, three recorded trials were done per leg. The LSI was calculated by dividing the average distance of the injured leg by the uninjured leg’s average distance, then multiplying by 100 ²⁷. In the squat test, participants stood on the injured leg, kept their hands on their hips, flexed the opposite leg to 90°, and performed a controlled single-leg squat while maintaining balance²⁸. The examiner recorded the time (in seconds) that each participant was able to maintain balance in this position. The trial was terminated when the non-supporting foot touched the ground or when the participant lost balance. Each test was performed twice, with 30 s rest between attempts.

Data analysis

COP data were recorded using a force plate at 100 Hz for 35 s during double-leg stance (with the last 30 s used for analysis) and for 10 s during single-leg stance. The data were processed in MATLAB (version 7.7.0471) using a zero-lag, second-order Butterworth low-pass filter with a 4 Hz cutoff. COP measures were calculated in both anteroposterior and mediolateral directions. These included: total sway displacement (DOT), range of movement forward/backward (Rfa) and side-to-side (Rsw), standard deviation of displacement, mean displacement velocity in each direction, total mean velocity (TMV), and COP displacement area. DOT represented the full path length of the COP. The range (Rfa and Rsw) measured the difference between the maximum and minimum displacement in each direction. Standard deviation reflected how much the displacement varied around the average position. Velocity was calculated by dividing displacement by time. TMV was obtained by dividing the total COP movement by the full trial duration. The COP displacement area was calculated as 95% of the total area covered by the COP, using an ellipse fit²⁹.

Statistical analysis

Data were analyzed using SPSS version 25.0 with significance set at p < 0.05. The Shapiro-Wilk test confirmed normal distribution for all outcome measures. Participant demographics were compared using independent samples t-tests. A 2 × 2 mixed-model ANOVA tested the effects of group (progressive balance training with medial heel wedge vs. progressive balance training alone), time (pre vs. post), and their interaction. The assumption of sphericity was assessed using Mauchly’s test; where sphericity was violated, the Greenhouse-Geisser correction was applied. For significant interactions, post-hoc with Bonferroni adjustment were conducted. Effect sizes for ANOVA results were reported as partial eta squared (η²) and interpreted as small (≥ 0.01), medium (≥ 0.06), or large (≥ 0.14)³⁰.

Result

There were 11 participants in each group. No adverse events or unintended effects were reported throughout the intervention or during assessments. There were no significant differences in age, height, weight, or BMI between the two groups (Table 1).

Table 1.

Descriptive statistics for the intervention and control groups.

Variables	Intervention (n = 11)	Control (n = 11)	Sig.
Age (years)	25.7 (4.68)	28.33 (5.82)	0.25
Height (m)	1.77 (0.08)	1.72 (0.09)	0.18
Weight (kg)	79.67 (15.03)	71.01 (13.28)	0.16
BMI (Kg/m2)	24.36 (2.74)	23.93 (3.76)	0.76
Female	4	4	N/A

Static postural control

Supplementary Tables 1 and 2 present descriptive statistics of postural control measures during double-leg and single-leg stance on the affected limb. Two-way mixed ANOVA tests were conducted to examine the effects of group (intervention vs. control), time (pre vs. post), and their interaction on COP parameters across stance and sensory conditions. The analysis showed that the medial heel wedge produced specific benefits, as indicated by significant group × time interactions. To control for multiple comparisons, Bonferroni adjustment was applied to post hoc tests (Table 2).

Table 2.

ANOVA results for postural control parameters during double-leg stance across conditions.

Condition	Effect	Statistic	COP parameter
			DOT	Rsw	Rfa	SD (ML)	SD (AP)	MV (ML)	MV(AP)	TMV	Area
DHO	Group	F-ratio	2.00	1.96	2.71	2.31	0.95	3.96	8.29	0.007	0.79
		Sig [η²]	0.16 [0.068]	0.16 [0.069]	0.10 [0.102]	0.13 [0.097]	0.33 [0.036]	0.053[0.103]	0.006 [0.317]	0.93 [0.001]	0.37 [0.060]
	Time	F-ratio	0.14	0.002	8.34	0.03	3.93	0.01	1.45	0.57	0.05
		Sig [η²]	0.70 [0.011]	0.96 [0.001]	0.006 [0.340]	0.86 [0.002]	0.054 [0.211]	0.89 [0.003]	0.23 [0.062]	0.45 [0.056]	0.81 [0.005]
	Group × Time	F-ratio	0.24	0.11	0.22	0.48	1.15	0.001	0.83	1.50	0.44
		Sig [η²]	0.62 [0.019]	0.73 [0.008]	0.63 [0.014]	0.49 [0.025]	0.28 [0.073]	0.97 [0.001]	0.36 [0.036]	0.09 [0.134]	0.51 [0.041]
DHC	Group	F-ratio	0.26	5.46	0.001	3.50	0.46	4.25	0.14	1.44	1.75
		Sig [η²]	0.60 [0.014]	0.02 [0.159]	0.97 [0.001]	0.06 [0.138]	0.50 [0.017]	0.04 [0.129]	0.70 [0.005]	0.23 [0.048]	0.19 [0.070]
	Time	F-ratio	0.21	0.04	0.005	0.56	0.21	4.95	0.01	0.69	1.26
		Sig [η²]	0.64 [0.039]	0.82 [0.004]	0.94 [0.001]	0.45 [0.030]	0.64 [0.016]	0.03 [0.305]	0.90 [0.001]	0.40 [0.058]	0.26 [0.710]
	Group × Time	F-ratio	0.60	0.60	0.31	0.90	0.38	0.42	3.82	2.78	2.60
		Sig [η²]	0.44 [0.010]	0.44 [0.052]	0.57 [0.022]	0.34 [0.048]	0.54 [0.029]	0.51 [0.037]	0.01 [0.248]	0.04 [0.197]	0.09 [0.136]
DSO	Group	F-ratio	2.29	0.01	0.29	1.04	1.77	3.74	5.85	0.24	1.92
		Sig [η²]	0.13 [0.081]	0.91 [0.001]	0.58 [0.010]	0.31 [0.053]	0.19 [0.060]	0.06 [0.109]	0.02 [0.176]	0.62 [0.008]	0.17 [0.072]
	Time	F-ratio	3.25	0.35	8.79	0.01	3.61	0.06	1.62	0.56	2.47
		Sig [η²]	0.04 [0.187]	0.55 [0.034]	0.005 [0.472]	0.90 [0.001]	0.02 [0.231]	0.79 [0.007]	0.20 [0.114]	0.45 [0.054]	0.08 [0.141]
	Group × Time	F-ratio	1.12	4.09	0.08	1.56	1.04	1.85	0.15	0.05	1.68
		Sig [η²]	0.29 [0.074]	0.05 [0.291]	0.77 [0.009]	0.21 [0.068]	0.31 [0.080]	0.06 [0.166]	0.69 [0.012]	0.81 [0.006]	0.20 [0.100]
DSC	Group	F-ratio	0.97	3.60	2.64	1.54	0.10	1.90	1.80	0.41	0.64
		Sig [η²]	0.32 [0.033]	0.06 [0.139]	0.11 [0.082]	0.22 [0.065]	0.75 [0.004]	0.17 [0.066]	0.18 [0.057]	0.52 [0.012]	0.42 [0.023]
	Time	F-ratio	8.60	15.58	3.48	12.04	8.55	6.90	1.11	5.57	0.21
		Sig [η²]	0.006 [0.422]	0.001 [0.470]	0.06 [0.248]	0.001 [0.407]	0.006 [0.410]	0.01 [0.343]	0.29 [0.097]	0.02 [0.446]	0.003 [0.450]
	Group × Time	F-ratio	0.54	1.44	0.45	0.54	0.001	1.48	1.24	0.47	0.77
		Sig [η²]	0.46  [0.044]	0.23 [0.076]	0.50 [0.041]	0.46 [0.030]	0.99 [0.001]	0.23 [0.101]	0.27 [0.107]	0.49 [0.064]	0.38 [0.059]

Center of pressure (COP); Double-leg stance on a hard surface with eyes open (DHO); Double-leg stance on a hard surface with eyes closed (DHC); Double-leg stance on a soft surface with eyes open (DSO); Double-leg stance on a soft surface with eyes closed (DSC); Total displacement of center of pressure (DOT); Range of sideways displacement (Rsw); Range of fore/aft displacement (Rfa); Standard deviation (SD); Mean velocity (MV); Total mean velocity (TMV); Anteroposterior (AP); Mediolateral (ML); Bold denotes statistically significant values (p < 0.05).

During double-leg stance, a significant interaction indicated that the intervention group showed a greater reduction in mean anteroposterior sway velocity on a hard surface with eyes closed (F(1, 20) = 6.589, p = 0.01, partial η² = 0.248), representing a large effect. This result remained significant after correction. The interaction for total mean velocity in the same condition (p = 0.04) did not survive adjustment and is therefore considered a non-significant trend (Table 3).

Table 3.

ANOVA results for postural control parameters during single-leg stance on the affected limb across conditions.

Condition	Effect	Statistic	COP parameter
			DOT	Rsw	Rfa	SD (ML)	SD (AP)	MV (ML)	MV(AP)	TMV	Area
SHO	Group	F-ratio	2.65	1.38	0.08	1.32	0.50	1.45	0.44	1.09	2.59
		Sig [η²]	0.11 [0.088]	0.24 [0.051]	0.76 [0.003]	0.25 [0.051]	0.48 [0.016]	0.23 [0.063]	0.50 [0.016]	0.30 [0.043]	0.18 [0.088]
	Time	F-ratio	13.73	15.85	9.48	15.82	8.78	16.15	23.09	21.07	12.44
		Sig [η²]	0.001 [0.521]	0.001 [0.524]	0.004 [0.454]	0.001 [0.504]	0.005 [0.487]	0.001 [0.466]	0.001 [0.636]	0.001 [0.577]	0.001 [0.491]
	Group × Time	F-ratio	0.13	0.05	0.41	0.08	1.02	0.03	1.07	0.77	0.89
		Sig [η²]	0.71 [0.011]	0.82 [0.004]	0.52 [0.003]	0.77 [0.005]	0.31 [0.100]	0.85 [0.002]	0.30 [0.075]	0.38 [0.048]	0.35 [0.064]
SHC	Group	F-ratio	19.95	2.80	7.43	3.87	10.44	7.37	9.19	12.34	7.58
		Sig [η²]	0.001 [0.511]	0.08 [0.138]	0.009 [0.228]	0.056 [0.130]	0.002 [0.307]	0.01 [0.240]	0.004 [0.260]	0.001 [0.347]	0.014 [0.267]
	Time	F-ratio	6.01	2.56	12.56	1.63	12.56	11.61	18.57	18.80	5.19
		Sig [η²]	0.01 [0.223]	0.11 [0.102]	0.001 [0.459]	0.20 [0.103]	0.001 [0.433]	0.002 [0.412]	0.001 [0.572]	0.001 [0.529]	0.03 [0.213]
	Group × Time	F-ratio	3.16	3.73	0.001	1.40	1.22	6.71	0.07	2.38	1.38
		Sig [η²]	0.08 [0.132]	0.06 [0.142]	0.71 [0.001]	0.24 [0.090]	0.27 [0.069]	0.01 [0.288]	0.76 [0.006]	0.13 [0.125]	0.24 [0.067]
SSO	Group	F-ratio	4.17	0.84	33.16	1.04	9.14	5.38	3.95	6.88	3.24
		Sig [η²]	0.04 [0.156]	0.36 [0.034]	0.001 [0.584]	0.31 [0.042]	0.004 [0.255]	0.07 [0.155]	0.084 [0.142]	0.01 [0.213]	0.11 [0.123]
	Time	F-ratio	19.44	38.21	79.91	28.65	17.83	40.19	30.38	52.17	30.88
		Sig [η²]	0.001 [0.528]	0.001 [0.001]	0.001 [0.830]	0.001 [0.643]	0.001 [0.572]	0.001 [0.789]	0.001 [0.654]	0.001 [0.782]	0.001 [0.646]
	Group × Time	F-ratio	5.61	0.64	17.61	1.80	6.65	0.49	1.34	1.52	4.72
		Sig [η²]	0.02 [0.244]	0.42 [0.039]	0.001 [0.518]	0.18 [0.102]	0.01 [0.333]	0.48 [0.044]	0.25 [0.077]	0.22 [0.095]	0.03 [0.218]
SSC	Group	F-ratio	3.08	1.52	3.95	0.23	4.20	0.32	2.41	2.47	2.09
		Sig [η²]	0.09 [0.132]	0.22 [0.079]	0.07 [0.149]	0.62 [0.038]	0.04 [0.164]	0.56 [0.013]	0.12 [0.113]	0.12 [0.102]	0.15 [0.097]
	Time	F-ratio	17.11	15.62	12.48	24.76	9.63	15.48	27.34	28.44	21.48
		Sig [η²]	0.001 [0.464]	0.001 [0.414]	0.001 [0.417]	0.001 [0.612]	0.003 [0.342]	0.001 [0.494]	0.001 [0.566]	0.001 [0.609]	0.001 [0.511]
	Group × Time	F-ratio	0.09	0.001	0.38	0.61	0.15	3.55	0.51	2.47	0.45
		Sig [η²]	0.75 [0.005]	0.99 [0.001]	0.53 [0.022]	0.43 [0.010]	0.69 [0.008]	0.04 [0.183]	0.47 [0.024]	0.12 [0.119]	0.50 [0.210]

Center of pressure (COP); Single-leg stance on a hard surface with eyes open (SHO); Single-leg stance on a hard surface with eyes closed (SHC); Single-leg stance on a soft surface with eyes open (SSO); Single-leg stance on a soft surface with eyes closed (SSC); Total displacement of center of pressure (DOT); Range of sideways displacement (Rsw); Range of fore/aft displacement (Rfa); Standard deviation (SD); Mean velocity (MV); Total mean velocity (TMV); Anteroposterior (AP); Mediolateral (ML); Bold denotes statistically significant values (p < 0.05).

Under the more demanding single-leg stance conditions, the intervention group’s changes were more distinct. The strongest effects occurred on a soft surface with eyes open, where the group showed significantly greater reductions in total sway displacement (F(1, 20) = 6.446, p = 0.02, partial η² = 0.244), forward–backward range (F(1, 20) = 21.464, p = 0.001, partial η² = 0.518), and anteroposterior standard deviation (F(1, 20) = 9.975, p = 0.005, partial η² = 0.333); all of these represent large effects and remained significant after Bonferroni adjustment. The interaction for sway area (p = 0.03) in this condition did not remain significant after correction. Furthermore, when vision was removed during single-leg stance on a hard surface, the intervention group showed a significantly greater reduction in mediolateral sway velocity (p = 0.01, η² = 0.288), also representing a large effect that persisted following adjustment (Table 3).

Dynamic postural control and functional performance

Analysis of dynamic postural control and functional performance showed a significant time effect for all outcomes (p < 0.001) with very large effect sizes, confirming that the 4-week progressive balance training program markedly improved both groups. A significant group × time interaction was observed only for the mSEBT in the anterior direction (F(1, 20) = 152.00, p = 0.004, partial η² = 0.186), which remained significant after Bonferroni correction, demonstrating that the intervention group achieved greater improvements in anterior dynamic stability. In contrast, no significant group × time interactions were found for other outcomes (p > 0.05), confirming that both groups improved similarly across these measures. Furthermore, there were no significant main effects of group for any outcome (p > 0.05). Analysis of the single-leg hop test showed that the significant improvement in the LSI was driven by large increases in hop distance on the involved side in both groups (Intervention: +22.73 cm; Control: +29.98 cm). A significant improvement in hop distance was also observed on the healthy, untrained side (Intervention: +6.62 cm; Control: +13.88 cm) (Table 4).

Table 4.

ANOVA results for star excursion balance test, single-leg hop test, limb symmetry index, and single leg squat.

Measure	Group	Before, mean(SD)	After, mean(SD)	Statistical effects; p-value [η²]
				Group	Time	Group × time
mSEBT-Anterior (% of limb length)	Intervention	73.90 (4.19)	92.45 (7.54)	0.51 [0.022]	0.001 [0.884]	0.004 [0.186]
	Control	75.14 (4.38)	88.21 (7.16)
mSEBT -Posteromedial (% of limb length)	Intervention	73.63 (9.73)	82.46 (11.41)	0.30 [0.634]	0.001 [0.634]	0.84 [0.002]
	Control	69.61 (7.73)	79.08 (6.17)
mSEBT -Posterolateral (% of limb length)	Intervention	78.55 (8.69)	89.19 (10.65)	0.23 [0.072]	0.001 [0.787]	0.49 [0.025]
	Control	73.73 (6.35)	86.29 (4.81)
mSEBT -Composite score (% of limb length)	Intervention	72.94 (7.50)	88.05 (8.93)	0.32 [0.049]	0.001 [0.834]	0.58 [0.016]
	Control	73.72 (6.59)	84.53 (5.13)
Single leg hop (healthy side) (cm)	Intervention	128.18 (46.12)	134.80 (48.22)	0.39 [0.037]	0.001 [0.455]	0.16 [0.095]
	Control	111.07 (20.39)	124.95 (22.55)
Single leg hop (involved side) (cm)	Intervention	109.93 (42.31)	132.66 (40.05)	0.23 [0.070]	0.001 [0.648]	0.41 [0.034]
	Control	89.41 (14.52)	119.39 (30.86)
Limb symmetry index (%)	Intervention	82.48 (10.46)	96.70 (3.13)	0.49 [0.024]	0.001 [0.546]	0.98 [0.001]
	Control	80.32 (14.21)	94.63 (7.97)
Single leg squat (s)	Intervention	21.86 (11.72)	38.36 (22.21)	0.39 [0.037]	0.001 [0.637]	0.23 [0.071]
	Control	20.28 (5.21)	31.07 (4.19)

SEBT, star excursion balance test. Bold denotes statistically significant values (p < 0.05).

Discussion

This study examined how a progressive balance training program, with or without a medial heel wedge insole, affected postural control and functional performance in people with CAI. Both groups improved in postural control after training. However, the intervention group that used the medial heel wedge demonstrated superior gains, achieving significantly greater improvements in static balance under the most challenging conditions and in forward dynamic balance.

Static postural control

This study showed that progressive balance training combined with medial heel wedge insoles led to meaningful improvements in static postural control among individuals with CAI, especially under more challenging conditions. The most notable effects occurred during single-leg stance on a soft surface with eyes open, where significant group at time interactions were found for multiple CoP measures (particularly DOT, Rfa, anteroposterior standard deviation, and area). These results suggest that the medial heel wedge may help improve balance by providing mechanical support and increasing sensory feedback during static standing tasks.

The intervention group showed trends toward greater reductions in sway and movement velocity, particularly under conditions involving unstable surfaces or reduced visual input. For example, during single-leg stance on a hard surface with eyes closed, all CoP parameters except standard deviation in the mediolateral direction significantly decreased over time in both groups, but were consistently lower in the intervention group. These findings are in line with previous studies, such as Chang et al. (2012), which found that medial heel wedges can reduce mediolateral sway and improve postural alignment in people with CAI^18,31. Likewise, Richie Jr. (2007) reported that orthoses might improve balance by stimulating plantar mechanoreceptors, which increases neurosensory input and postural control¹⁷.

Improvements were also seen in double-leg stance conditions, where the intervention group consistently outperformed the control group in several CoP parameters. A significant group × time interaction was observed for Rsw during double-leg stance on a soft surface with eyes open, suggesting greater gains in the intervention group. This supports the idea that medial heel wedges may act not only as a mechanical support but also as a proprioceptive aid, helping reorganize sensorimotor control in people with CAI.

Importantly, all post-training balance assessments were performed without the medial heel wedge insole in place, indicating that the improvements were not just short-term mechanical effects. Instead, they likely reflect longer-lasting neuromuscular changes. This supports the theory that repeated sensory input from orthotic use can enhance central nervous system plasticity and improve proprioceptive function over time^17,18,31. This aligns with findings from Mettler et al. (2015), who observed improved postural control and a backward CoP shift after balance training, pointing to better sensorimotor function¹².

Dynamic postural control and functional performance

This study found that a 4-week progressive balance training program significantly improved dynamic postural control and functional performance in individuals with CAI, as shown by large, meaningful improvements in mSEBT scores and single-leg squat performance in both groups. Both groups also demonstrated substantial improvements in limb symmetry, with the LSI derived from the single-leg hop test increasing from 82.48% to 96.70% in the intervention group and from 80.32% to 94.63% in the control group. Although no established LSI cutoff values exist for CAI, these findings suggest that progressive balance training may effectively reduce inter-limb asymmetry—an important consideration in functional rehabilitation³². The LSI cutoff is commonly used to guide return-to-play decisions in ACL rehabilitation³³, and its improvement here supports the relevance of this metric in CAI contexts as well. Similarly, significant improvements were observed for single-leg hop distance on both the healthy and involved sides, confirming improved functional performance across groups. These results are consistent with previous research indicating that targeted balance training promotes neuromuscular adaptations that improve dynamic stability^34–36 as evidenced by the large within-group effect sizes. Interestingly, while the 4-degree medial heel wedge did not enhance most functional outcomes, participants demonstrated a significantly greater improvement in anterior reach distance during the mSEBT compared to the control group. This selective benefit, coupled with the observed improvements in anteroposterior control during static tasks, suggests the wedge may provide specific advantages for sagittal-plane control. The limited between-group differences in other measures likely reflect the dominant effect of progressive balance training itself on neuromuscular adaptation³⁷, as evidenced by significant within-group improvements across all conditions. The wedge’s effect appears to be both task- and plane-specific, likely mediated by altered weight distribution and improved proprioceptive feedback that is most relevant to forward-reaching movements. This suggests that the mSEBT anterior reach may serve as a sensitive marker for tracking the effectiveness of the combined intervention. It should be noted that the current study used modified SEBT described by Gribble et al. (2012). Participants kept their arms at their sides rather than on their hips, which may slightly influence balance strategy; however, this modification was applied uniformly across groups and is unlikely to affect between-group comparisons²⁶.

Baseline data confirmed the expected impairment in single-leg hop performance typical of CAI. Augustsson and Sjöstedt (2023) reported a 9% between-limb deficit³⁸, whereas larger initial deficits were seen here (20% in the control and 14% in intervention groups). After 4 weeks of progressive balance training, hop distance on the involved side improved by 34% (+ 29.98 cm) in the control group and 21% (+ 22.73 cm) in the intervention group, indicating considerable recovery of this functional deficit. Similarly, the single-leg squat is recognized as an impaired task in CAI, with recent evidence showing altered biomechanics including reduced ankle dorsiflexion³⁹.

A secondary finding was the improvement in single-leg hop distance on the healthy, untrained limb, suggesting consistency with the cross-education effect. This observation aligns with the results of Elsotohy et al. (2021), who found that balance training applied only to the non-affected limb significantly improved postural control indices in the affected limb among individuals with CAI⁴⁰. The challenging and progressive balance training in the current study, requiring high proprioceptive demand and neuromuscular control, likely provided adequate stimulus for these central adaptations. These findings build on previous work by demonstrating that the cross-education effect extends to dynamic tasks such as the single-leg hop. This supports the value of unilateral balance training and suggests that its benefits may go beyond the targeted limb, with potential positive implications for athletic performance and injury prevention.

Limitations and future directions

While this study provides valuable insights into progressive balance training with medial heel wedges, some methodological considerations should be noted. The small sample of 22 participants limits statistical power and generalizability, necessitating future studies with larger cohorts. As both groups underwent identical balance training, the specific effect of the wedges cannot be isolated from the effect of the training itself. Future trials that include a no-intervention control group are needed to distinguish between these factors. All assessments were conducted within a single session, and although rest periods were provided, the possible influence of fatigue on performance cannot be fully excluded. Additionally, the post-intervention assessments were performed without the wedges to evaluate lasting neuromuscular adaptations, which did not fully match the training conditions and may have influenced the observed outcomes. Another limitation is the short-term nature of the study, which makes it difficult to draw conclusions about how lasting the observed neuromuscular changes might be. Absence of follow-up prevents conclusions about long-term retention or reinjury risk. Future studies with medium- and long-term follow-up are recommended. Moreover, the study did not include objective biomechanical assessments such as gait analysis or detailed kinematic/kinetic evaluations, which could clarify the mechanistic pathways through which medial heel wedges influence postural control. Future studies might also compare results with and without wedges to better understand both mechanical and neuromuscular effects. In addition, future research should identify which patients benefit most from the wedges by investigating factors like injury history, baseline severity, and activity level. Exploring different insole textures could help determine which design features most effectively support balance training and neuromuscular adaptation. While wedges appear low-cost and easy to implement, a formal economic analysis is beyond the scope of this study; future studies should evaluate cost-effectiveness and implementation considerations. Although the mean BMI differed numerically between groups, this difference was not statistically significant. Given the small sample size, small numerical differences in baseline BMI are expected and likely reflect random chance, not a flaw in the study design. It is also important that the groups were similar in demographic characteristics which could have affected the results. Therefore, it is unlikely that the small BMI difference influenced the improvements seen after the training; nevertheless, future studies with larger samples should examine whether BMI might play a role.

Conclusion

This randomized trial demonstrated that a 4-week progressive balance training program led to significant improvements in static and dynamic postural control as well as functional performance in individuals with CAI. Participants in both groups made large gains in mSEBT performance, limb symmetry, and single-leg squat ability, showing that balance training alone is highly effective. The addition of a 4° medial heel wedge produced a selective between-group advantage, not only for the SEBT anterior reach but also for static postural control, particularly during the most challenging single-leg stance conditions on an unstable surface. While LSI (from the single-leg hop test) and single-leg squat improvements did not differ between groups, the wedge group demonstrated superior adaptation in controlling sway, especially in the anteroposterior direction. Because the post-tests were done without wearing the wedges, this advantage likely reflects a true neuromuscular adaptation rather than a temporary mechanical effect and may be related to improved sagittal-plane weight transfer and increased plantar mechanoreceptor input. Overall, progressive balance training appears to be the main factor driving recovery, while medial heel wedges provide a measurable extra benefit for both dynamic forward reach and static stability under high demand. Further studies with longer follow-up and larger samples are needed to confirm these findings and to identify which individuals benefit most from wedge use.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: K.M., M.P.; Methodology: M.P., K.M., K.O., A.S.; Investigation: M.P.; Formal Analysis: M.P., K.M., K.O.; Writing – Original Draft: M.P.; Writing – Review & Editing: M.P., K.M., A.S.; Supervision: K.M., K.O., A.S.

Data availability

The data that support the findings of this study are available on request from the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.