Intermittent vibrational stimulation can affect the lower limb kinematics three months after anterior cruciate ligament reconstruction: a randomised controlled trial

Abstract

Anterior cruciate ligament (ACL) injuries often lead to significant gait alterations, with traditional rehabilitation sometimes failing to address these issues. Intermittent vibration stimulation (IVS) can enhance muscle function and reduce pain in knee pathologies. This randomized controlled trial investigated whether IVS affects lower limb kinematics during gait and stair ambulation in individuals three months post-ACL-reconstruction (ACLR). Twenty-seven participants, aged 18–45, were randomly assigned to receive IVS or a sham device. Additionally, 24 healthy participants were recruited to provide normative data. The primary aim was to measure knee sagittal angles during gait at various speeds and during stair navigation. A secondary aim was to assess the sagittal angles of the hip and ankle. Statistical analyses involved a two-way mixed ANOVA. After Benjamini–Hochberg correction, significant group×time interactions were found for minimum knee flexion during normal walking (F(1,25) = 8.50, p = .007, partial η² = 0.254) and slow walking (F(1,25) = 6.32, p = .019, partial η² = 0.208). Post-hoc comparisons at the intervention phase showed greater minimum knee flexion in the IVS group than in the sham group during normal walking (Δ = 4.60°, SE = 1.37°, p = .003, partial η² = 0.309) and slow walking (Δ = 4.67°, SE = 1.37°, p = .003, partial η² = 0.316). First-peak knee flexion showed main effects of time at normal (F(1,25) = 6.47, p = .018, partial η² = 0.206) and slow speeds (F(1,25) = 14.59, p < .001, partial η² = 0.378), driven by within-IVS increases (normal: +1.77°, SE = 0.59°, p = .006, partial η² = 0.264; slow: +2.36°, SE = 0.56°, p < .001, partial η² = 0.423). For fast walking, minimum knee flexion also showed a main effect of time (F(1,25) = 8.17, p = .008, partial η² = 0.246), with a within-IVS increase (+ 2.25°, SE = 0.63°, p = .001, partial η² = 0.340). During stair ascent, minimum knee flexion increased over time (F(1,25) = 6.02, p = .022, partial η² = 0.208), with a within-IVS change of + 2.06° (SE = 0.73°, p = .010). These findings suggest that IVS affects lower limb kinematics during rehabilitation after ACLR. Further research is needed to explore its long-term effects.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-23042-7.

Introduction

Anterior cruciate ligament (ACL) injuries are a prevalent concern, especially among active young people¹. Such injuries lead to extended changes in everyday life and sports activities. They are followed by a lengthy rehabilitation process, usually lasting at least nine months, before returning to their pre-injury activity levels². Despite rehabilitation efforts, many individuals still experience persistent ambulatory changes in everyday activities such as walking and ascending or descending stairs^3–5.

The ambulatory changes reported after ACL injury have important clinical implications, indicating a need to explore innovative approaches to address these issues. Quadriceps weakness, often observed post-ACL injury, was first associated with ‘quadriceps avoidance gait’ by Berchuck et al.⁶, which can impede effective rehabilitation. This adapted gait can lead to asymmetry, slower walking speed, altered joint movement patterns, and changes in joint mechanics and loading, all of which may contribute to the development of osteoarthritis (OA)^7,8. These gait changes and the increased risk of developing OA highlight the need for early interventions to improve abnormal gait patterns, potentially reducing the risk of long-term OA.

These changes are pronounced at three months post-surgery, a critical mid-stage rehabilitation window when patients transition from early range-of-motion restoration to more demanding functional tasks⁹. Implementing interventions during this phase that can safely enhance quadriceps activation and promote more physiological knee motion may help restore gait mechanics and reduce future functional limitations.

Intermittent vibration stimulation (IVS) targets pain-related sensory nerves and engages pathways that respond to mechanical stimuli like vibration and pressure. This method uses the somatosensory system’s “gate control” mechanism to regulate pain¹⁰. According to this theory, pain signals transmitted by smaller nerve fibers are moderated by a “gate” in the spinal cord, which can be influenced by larger nerve fibers that activate cutaneous mechanoreceptors that respond to non-painful sensations, such as vibration and pressure. When these larger fibers are activated, they can inhibit pain signal transmission by “closing” the gate, thus reducing pain perception and supporting improved function. Beyond the ‘gate control’ mechanism, IVS could influence the neuromuscular function by stimulating mechanoreceptors through the tonic vibration reflex, wherein vibration activates muscle spindles sensitive to stretching^11,12, possibly enhancing the proprioceptive feedback from the quadriceps and increasing muscle activation.

Previous studies have shown that IVS can increase knee flexion moments and angles during gait and stair navigation, improve quadriceps activation, and reduce pain in individuals with mixed knee pathologies^13–15. However, the specific effects of IVS on the kinematics of the lower limb following ACL reconstruction (ACLR), particularly during the early phase of rehabilitation, remain underexplored.

While previous studies from our group have demonstrated the potential of IVS in mixed knee pathology populations, the present study focuses on its application during the critical early rehabilitation phase (three months post-ACLR) in a homogeneous cohort. Moreover, by assessing multi-joint kinematics during overground walking and stair ambulation using wearable IMUs, this study extends earlier findings and provides task-specific insights into the potential rehabilitative benefits of IVS in post-ACLR recovery.

Persistent alterations in knee mechanics following ACLR are well-documented, often becoming more pronounced during complex tasks such as gait ambulation and walking at different speeds. Increased task difficulty, such as stair ambulation, led to increased alterations, while varying walking speeds exacerbated gait asymmetries in individuals with ACLR^16,17. Additionally, joint loading and knee flexion are significantly altered during stair ambulation post-ACLR, highlighting the importance of early interventions to address these deficits¹⁸.

The main aim of this study was to evaluate the effects of IVS on knee kinematics among individuals three months after ACLR while walking overground, ascending stairs, and descending stairs. A secondary aim was to assess the hip and ankle kinematics during these movements. A healthy reference cohort was recruited to provide normative IMU-based kinematic values. We hypothesized that individuals receiving a device that applies IVS to the knee during ambulation would demonstrate increased peak knee flexion angles during weight-bearing compared to controls.

Methods

Study design

The study was a randomized controlled trial with a parallel 2-group before and after design and followed the Consolidated Standards of Reporting Trials (CONSORT) Statement¹⁹.

The institutional review board of Rambam Health Care Campus approved this study protocol (0089-21-RMB). The study was registered at clinicaltrials.gov (NCT05001594) on August 12th, 2021. The data collection started in January 2022. The procedures were conducted in accordance with the ethical standards outlined in the Helsinki Declaration. All participants provided written informed consent to participate in the study.

All measurements and intervention setup were conducted by a single researcher, a physiotherapist with eight years of clinical experience and three years of experience in a motion analysis lab. The researcher was not blinded to group allocation. While the use of a sham device helped mitigate participant expectation bias, assessor blinding was not feasible due to the practical aspects of the device setup and operation.

No changes were made to the trial design, eligibility criteria, and outcomes after the trial commenced.

Participants

The inclusion criteria comprised males and females aged 18–45 who had undergone ACLR surgery at the Rambam Health Care Campus three months prior. Exclusion criteria included failure to provide informed consent, a history of previous ACL injury, prior lower limb injuries, and any active cardiovascular, neurological, or respiratory conditions.

Additionally, a convenience cohort comprising healthy participants was recruited through the university message boards and social media; This cohort was included to establish normative gait and stair-ambulation kinematics for IMU measurements.

The inclusion criteria for the healthy cohort were healthy males and females aged 18–45 without any lower limb pain. The exclusion criteria were previous lower limb surgery or injury, active cardiovascular, neurological, or respiratory conditions, and failure to provide informed consent.

No patient or public involvement was undertaken in the design and conduct of this study.

Research protocol

The participants were randomly assigned to two groups using an online random number sequence generator. Due to practical constraints, the researcher who enrolled participants also assigned them to interventions and was not blinded to group allocation. One group received an IVS device (KneeMo®, SomaTX Design, Inc.), while the other group received a 3D-printed sham device without vibration. Each device consisted of two elastic bands, one above and one below the knee. Both devices were positioned on the participant’s injured leg, providing pressure. However, only the IVS device delivered external vibration stimulation. The vibration was synchronized to the gait cycle, activated just before the heel strike, and turned off at mid-stance¹⁵. The participants were informed that they would receive one of two interventions: either local pressure or pressure and vibration. The sham device was made to look and feel like the IVS device. The sham was built from the same soft material as the IVS device and weighed approximately the same to provide comparable skin pressure, produced no sound or vibration, and differed only in color, which was not disclosed to participants (Fig. 1).

Fig. 1.

The vibrational device and the sham device. The vibrational device is on the left (A), and the sham is on the right part of the figure (B).

The participants continued standard postoperative rehabilitation without restrictions or additional interventions. The study protocol followed a similar structure to what was used in previous studies^14,15.

Employing a sham device as a control helps reduce potential placebo effects, where participants may anticipate benefits solely from receiving an intervention. By comparing the effects between the groups, we can more confidently attribute any observed improvements to the IVS.

Data were recorded at 120 Hz using seven Inertial measurement units (IMUs) (MTw Awinda, Movella) attached to the participants’ lower limbs using Velcro straps: Upper leg (x2), lower leg (x2), feet (x2), and pelvis (x1)²⁰. The Participants acclimated to the sensors before data collection, initially walking without any device (baseline phase) and then with either the IVS or sham device, depending on randomization (intervention phase).

The participants first walked, ascended, and descended stairs without any device (baseline) and then walked, ascended, and descended stairs with a device (IVS or sham). Each participant completed three repetitions of walking along a 20-meter corridor at three speeds: self-selected normal, slow, and fast. The instructions for each condition were as follows and were similar for all the participants. For the self-selected speed: “Walk across the corridor at your normal speed”; for the slow speed: “Walk across the corridor at a slow speed,” and for the fast speed: “Walk as fast as possible across the corridor”²¹. Participants then ascended and descended a 20-step staircase (rise = 17 cm, run = 30 cm) at their comfortable speed for three repetitions.

Following data collection, each participant’s gait cycles were segmented based on the recorded IMU data. We excluded each trial’s first and last cycles to avoid acceleration or deceleration effects. The remaining cycles were averaged for each condition (normal, slow, and fast walking). From these average cycles, key kinematic variables were extracted for analysis.

For overground walking, the gait cycle was divided into loading response (0–10%), mid-stance (10–30%), terminal stance (30–50%), pre-swing (50–60%), and swing (60–100%) phases. For stair ascent, the cycle was subdivided into weight acceptance (0–10%), pull-up (10–32%), forward continuance (32–60%), foot clearance (60–80%), and foot placement (80–100%). For stair descent, the cycle was subdivided into weight acceptance (0–12%), forward continuance (12–30%), controlled lowering (30–62%), leg pull-through (62–85%), and foot placement (85–100%). The main outcomes included the minimum and maximum knee sagittal angles, while the secondary outcomes focused on the hip and ankle minimum and maximum sagittal angles (Fig. 2).

Fig. 2.

Points of interest during the gait cycle in the knee, hip and ankle joints.

Gait cycle graphs for knee, hip, and ankle angles during walking, stair ascent, and stair descent. Points of interest (minimum, first peak, second peak, maximum) are marked for each joint. Light vertical lines indicate subphase boundaries: LR (loading response), MS (mid-stance), TS (terminal stance), PS (pre-swing), and SW (swing) for walking; WA (weight acceptance), PU (pull-up), FC (forward continuance), CL (clearance), and FP (foot placement) for stair ascent; and WA (weight acceptance), FC (forward continuance), CL (controlled lowering), LP (leg pull-through), and FP (foot placement) for stair descent.

All participants were recruited and assessed at the Rambam Health Care Campus, Haifa, Israel. Throughout the intervention phase, participants were monitored for adverse events, including discomfort and skin irritation, with no harms reported during the trial.

Data and statistical analysis

The kinematic data were exported using Xsens’ proprietary software (MVN Analyze 2023.2). A custom Python algorithm was used to extract the maximum and minimum joint angles of each movement cycle. The outcome measure was the average of these cycles.

The normality of the data distribution was assessed using the Shapiro-Wilk test. Demographic differences between groups and between the legs of the healthy cohort were compared using the Student’s t-test for continuous variables and Fisher’s exact test for nominal data.

Separate two-way mixed ANOVA tests evaluated between-group (group = IVS vs. sham device) and within-group (time = baseline vs. intervention) differences of the gait and stair ambulation data, with significant interactions further evaluated using the ANOVA pairwise comparison. Effect sizes were reported using partial Eta squared (partial η²).

To control for Type I errors across the 99 statistical comparisons, we applied the Benjamini-Hochberg procedure. This correction was applied separately to the primary (knee, 39 tests) and secondary (hip/ankle, 60 tests) aims to preserve statistical power. This approach maintains statistical power within each family, preventing the results of one aim from affecting the other. We set the FDR level at 0.1, balancing the need to detect true effects.

An additional comparison was made between the ACLR cohort’s injured leg and the healthy participants’ leg using an independent sample t-test. The statistical significance was set at p < .05, and analyses were performed using IBM SPSS Statistics (Version 29).

No interim analyses or stopping guidelines were pre-specified due to the short-term nature of the intervention.

A priori power analysis was conducted to determine the minimum sample size required to detect a medium effect based on an F-test and a within- and between-group interaction statistical test using G*Power (Version 3.1.9.7). The results indicated that the required sample size to achieve 80% power for detecting a medium effect (0.25), at a significance level of α = 0.05, was n = 34.

Results

Participant characteristics

Recruitment took place between February 2022 and October 2023. Due to the COVID-19 pandemic, data collection was slower and stopped earlier than planned. Therefore, the final sample size is smaller than the planned number. With 27 participants instead of 34, our planned type 2 error rate increased from 20% (80% power) to 30% (70% power).

The mean age of the ACLR participants was 23.5 ± 5.7 years, with a mean weight of 75.8 ± 11.6 kg and a mean height of 1.77 ± 0.1 m, respectively. Data for stair ambulation from one ACLR participant was corrupted, resulting in 26 participants being included in the stair ambulation analysis. An additional 24 healthy participants were recruited. Demographic characteristics of all participants are presented in Table 1. The overall study flow is illustrated in Fig. 3.

Table 1.

Demographics of the participants.

	Participants with ACLR (n = 27)
	Sham Group (n = 12)	IVS Device Group (n = 15)	p- value	Healthy Participants (n = 24)
Age (years)	25.8 ± 7.1	21.7 ± 3.7	0.066	28.6 ± 5.8
Sex (%) Males Females	10 (83) 2 (17)	11 (73) 4 (27)	0.662	13 (54.2) 11 (45.8)
Height (m)	1.81 ± 0.1	1.75 ± 0.1	0.060	1.68 ± 0.1
Weight (kg)	79.3 ± 9.8	73.3 ± 12.2	0.175	65.0 ± 15.0
The time between injury and reconstruction (days)	255 ± 194	214 ± 152	0.555	N/A
Graft (%) Hamstrings BTB Quadriceps Allograft	9 (75) 2 (16) 0 1 (8)	8 (53) 3 (20) 4 (26) 0	0.168	N/A

Two-sided paired sample t-test for continuous variables. Fisher’s exact test for nominal data Bold text = Significant at P < .05.

Fig. 3.

CONSORT flow diagram.

Table 2 summarizes the baseline sagittal-plane kinematics during walking without a device for both ACLR and healthy participants. Detailed hip and ankle kinematics for both groups are provided in Appendix 1 and Appendix 2, respectively.

Table 2.

Baseline knee sagittal angle of the participants while walking by group.

Walking Pace	Angle°	ACLR (n = 27)		p- value	Healthy Participants (n = 24)		p- value
		Injured	Contralateral		Left	Right
Slow	Minimum	1.1 ± 3.1	-3.3 ± 2.3	< 0.001	-3.5 ± 3.3	-2.7 ± 3.3	0.155
	First Peak	10.2 ± 5	9.4 ± 7.6	0.486	6.9 ± 8.2	8 ± 7.6	0.103
	Second Peak	55.6 ± 6.1	60.7 ± 4.5	< 0.001	59 ± 4.3	60.3 ± 3.9	0.098
Normal	Minimum	1 ± 3.1	-3.8 ± 2.6	< 0.001	-3.4 ± 3.8	-2.7 ± 3.5	0.058
	First Peak	12.8 ± 4.9	14.1 ± 6.5	0.287	14.8 ± 6.5	15.3 ± 7.1	0.347
	Second Peak	60.1 ± 7	63.3 ± 5.1	0.003	62.2 ± 3.8	63.3 ± 4.3	0.111
Fast	Minimum	2.1 ± 4	-2.5 ± 2.8	< 0.001	-2 ± 4.6	-1.6 ± 3.7	0.529
	First Peak	16.2 ± 5.1	21.4 ± 4.7	< 0.001	20.2 ± 6	21.4 ± 6.7	0.100
	Second Peak	61.1 ± 5.4	62.9 ± 4.6	0.047	61.6 ± 4.1	62.2 ± 4.3	0.390

Two-sided paired sample t-test for both legs of each cohort. Bold text = Significant at P < .05. ACLR = Anterior cruciate ligament reconstruction.

IVS effects on knee kinematics

As detailed in Table 3, several effects on knee kinematics remained significant after Benjamini-Hochberg correction. Figures 4 and 5 describe the interaction effects of walking and stair ambulation, respectively.

Table 3.

Summary of knee kinematic statistically significantfindings after Benjamini-Hochberg correction

Activity (Angle°)	Group	Time		Group by Time Interaction (p-value)	Main Effects (p-value)
		Baseline (without device)	Intervention (with device)
					Time	Group
Slow Walking (Minimum)	Sham	-0.3 ± 2.7	0.5 ± 2.8	.019*	<.001*	.007*
	IVS	2.1 ± 3.0	5.1 ± 4.1
Slow Walking (First Peak)	Sham	9.1 ± 4.1	9.9 ± 4.1	.07	<.001*	.25
	IVS	10.5 ± 5.3	12.9 ± 5.6
Normal Walking (Minimum)	Sham	-0.1 ± 2.4	0.7 ± 3.1	.007*	<.001*	.011*
	IVS	1.9 ± 3.3	5.3 ± 3.9
Normal Walking (First Peak)	Sham	11.6 ± 6.2	12.1 ± 4.9	.16	.018*	.15
	IVS	13.8 ± 4.9	15.6 ± 5.4
Fast Walking (Minimum)	Sham	0.9 ± 2.8	1.3 ± 3.2	.06	.008*	.045
	IVS	3.1 ± 4.6	5.4 ± 4.6
Stair Ascent (Minimum)	Sham	14.5 ± 4.8	15.1 ± 4.6	.227	.022*	.85
	IVS	12.9 ± 4.5	14.9 ± 5.1

Fig. 4.

Knee joint interaction plots across walking conditions and groups. Mean values of the minimum, first-peak, and second-peak knee angles during slow, normal, and fast walking for the sham and IVS groups at baseline (time 1) and intervention (time 2). Error bars represent 95% confidence intervals. Boxes inside each subplot display the change (Δ) from baseline to intervention for each group

Fig. 5.

Knee Joint Interaction Across Stair Ambulation Conditions and Groups. Mean values of the minimum and maximum knee angles during stair ascent and descent for the sham and IVS groups at baseline (time 1) and intervention (time 2). Error bars represent 95% confidence intervals. Boxes inside each subplot display the change (Δ) from baseline to intervention for each group.

A significant group-by-time interaction was found for the minimum knee flexion angle during normal walking (F(1,25) = 8.50, p = .007, partial η2 = 0.254) and slow walking (F(1,25) = 6.32, p = .019, partial η2 = 0.208). Post-hoc analysis for normal walking revealed that the minimum knee flexion angle was significantly higher in the IVS group compared to the sham group during the intervention phase (mean difference = 4.60°, SE. = 1.37°, p = .003, partial η² = 0.309). Similarly, at a slow walking speed, the IVS group’s angle was higher than that of the sham group (mean difference = 4.67°, SE. = 1.37°, p = .003, partial η² = 0.316).

Furthermore, significant main effects of time that survived correction were found. For the first peak knee flexion during walking, a main effect of time was present at normal (F(1,25) = 6.47, p = .018, partial η² = 0.206) and slow speeds (F(1,25) = 14.59, p < .001, partial η² = 0.378). For fast walking, a main effect of time was also found for the minimum knee flexion angle (F(1,25) = 8.17, p = .008, partial η² = 0.246).

Post-hoc analysis for normal speed indicated a significant increase within the IVS group (mean difference = 1.77°, SE = 0.59°, p = .006, partial η² = 0.264), while the sham group showed no significant change (mean difference = 0.49°, SE = 0.66°, p = .470, partial η² = 0.021). At slow speed, post-hoc tests showed a significant increase in flexion from baseline to intervention in the IVS group (mean difference = 2.36°, SE = 0.56°, p < .001, partial η² = 0.423), whereas the sham group exhibited a smaller, non-significant increase (mean difference = 0.80°, SE = 0.61°, p = .198, partial η² = 0.068). Similarly, for fast speed, post-hoc analysis revealed a significant increase from baseline to intervention within the IVS group (mean difference = 2.25°, SE = 0.63°, p = .001, partial η² = 0.340), while the sham group showed no significant change (mean difference = 0.44°, SE = 0.70°, p = .538, partial η² = 0.015).

A main effect of time was also identified for the minimum knee flexion angle during stair ascent (F(1,25) = 6.02, p = .022, partial η² = 0.208), with post-hoc analysis showing a significant increase within the IVS group (mean difference = 2.06°, SE. = 0.73°, p = .010) whereas the sham group exhibited a smaller, non-significant increase (mean difference = 0.67°, SE = 0.83°, p = .426, partial η² = 0.028).

Knee kinematic interaction data are available in Appendix 3 for walking, Appendix 6 for stair ascent, and Appendix 7 for stair descent.

IVS effects on hip and ankle kinematics

No statistically significant interactions or main effects were found for any hip or ankle kinematic variables after correcting for multiple comparisons. Detailed statistics are available in Appendices 4–7.

Discussion

We investigated the effects of IVS on lower-limb kinematics three months after ACLR. An intervention×time interaction suggests larger kinematic changes with the IVS than with the sham. Because the sham controlled for wearing a device, these differences suggest that the effects are IVS-specific rather than due to the placebo or device acclimation. Both devices affected knee flexion, but the IVS group showed higher and more consistent minimum and first peak knee flexion angles across all walking speeds and during stair ascent.

The gait cycle typically lasts 0.98–1.07 s,²² with the minimum knee flexion angle occurring at or just before heel-strike, and the first peak flexion angle occurring during the subsequent loading response phase, approximately 100 milliseconds apart. The IVS intervention was activated from heel-strike to mid-stance, directly overlapping with these load-acceptance events. This synchronization suggests that the observed increase in minimum knee flexion might be a response to the IVS.

Currently, IVS appears to have its most pronounced effect near the minimum flexion angle, which results in gait patterns that deviate from those of healthy participants. However, enhancing IVS effects around the first peak flexion angle could promote more normalized movement patterns, aligning the ACLR cohort’s gait more closely with that of the healthy cohort. This effect may be due to IVS inducing a stronger quadriceps contraction and reducing pain during the stance phase, when the participant bears weight on the injured leg, from heel strike to weight acceptance¹².

Contrary to the “stiff-knee” gait sometimes observed after ACLR, our cohort’s injured knee was more flexed at baseline than that of healthy participants. This may reflect a learned compensatory strategy to guard the joint and enhance stability through muscle co-contraction. While a measurement artifact is a consideration, the fact that both cohorts were measured with the identical system and the group-by-time interaction effects were robust suggests this is a true baseline pattern. Nonetheless, because some effects approach the IMU RMSE, we interpret cautiously.

We hypothesize that this effect is an acute, phase-specific modulation of load acceptance rather than an immediate normalization to the healthy gait. In the early stance, the ground-reaction force passes posterior to the knee, creating an external knee-flexion moment. A small increase in knee flexion at heel-strike lengthens the GRF lever arm and slightly increases this moment, which the quadriceps counter with an internal extension moment. Within physiological ranges, this can improve load acceptance by facilitating quadriceps engagement, therefore, inducing smoother shock absorption. Previous studies found that local vibration has shown potential in enhancing early quadriceps strength recovery and improving muscle torque in patients undergoing ACL rehabilitation, supporting its role as an adjunctive tool for optimizing recovery^23,24.

Previous studies generally support the use of IVS/local vibration in increasing knee flexion or quadriceps engagement across various tasks and populations. In a mixed knee-pathology cohort among older adults, IVS showed a small trend toward higher loading-response flexion (16.9° vs. 16.4°). However, minimum flexion was not assessed¹³. The bigger differences in the current study may be due to the more acute nature of ACLR. Pamukoff et al. (2016) reported enhanced quadriceps activation after 30 Hz local vibration in individuals four years post-ACLR. However, this study didn’t report on the kinematic changes²⁵. Blackburn et al. (2020) demonstrated improvements to gait biomechanics linked to post-traumatic OA, likely via increased quadriceps excitability²⁶.

In line with our findings, Mohamadi et al. (2014) reported that local vibration applied to the quadriceps improved active knee flexion range of motion in women with knee OA.²⁷ Although their study focused on an older OA population and did not directly assess dynamic gait kinematics. Similarly, Rippetoe et al. (2020) applied focal muscle vibration around the knee to people with diabetic peripheral neuropathy, reporting trends toward increased peak knee flexion. Lowe et al. (2023) used localized vibration before a drop-landing task with individuals after ACLR, resulting in increased peak knee flexion^28,29.

Next, we observed kinematic changes during stair ascent when using the IVS device, indicating a significantly higher minimum knee flexion angle compared to the sham group. These findings partially align with those of Fischer et al. (2021), who reported no changes in knee flexion during stair ascent and a reduction in knee flexion during stair descent among patients with knee pain¹⁵. Similarly, we observed a reduction in knee flexion during stair descent, but it did not reach statistical significance. This discrepancy may be attributed to differences in population characteristics, such as age, and the more heterogeneous nature of Fischer’s cohort. Furthermore, the more acute nature of ACLR in our study population, compared to chronic knee conditions, may lead to different responses to IVS.

While IVS led to significant effects during overground walking, stair ascent and descent showed lesser effects in knee kinematics. This may be attributed to the increased task demands during stair navigation, where the knee must support greater loads and navigate larger ranges of motion at the joint. It is also possible that the vibration timing should be different for stair ambulation compared with walking on flat terrain.

Limitations and future work

This single-session, cross-sectional study could not establish the long-term effects of IVS. The findings are based on a relatively small sample size, which may limit the generalizability to a broader population. The final statistical power was 70%, increasing the risk of Type II errors (failing to detect true effects), particularly for our secondary outcomes. Additionally, subgroup analyses by graft type and sex were not feasible. While the statistically significant findings for our primary knee outcomes, despite a smaller cohort, suggest a robust effect, the conclusions should be interpreted with caution.

Baseline differences were minor but present. The IVS group was younger than the sham group (p = .066), which may have influenced the results.

Assessor blinding was not implemented, introducing potential bias. While the use of objective kinematic data from IMUs mitigate subjective measurement bias, unconscious influence during participant instruction, sensor placement, or the manual stages of data processing cannot be entirely excluded.

While IMUs offer ecological validity, their precision (RMSE of 2.5°-5° for knee sagittal plane measurements) is lower than that of gold-standard optical motion capture systems,³⁰ some significant findings, such as the 4.6° change in minimum knee flexion, are close to this error range. However, the effect was consistent across walking speeds, suggesting a systematic change rather than random error. Further, our before-and-after design assesses the change within each participant, which is less sensitive to absolute offset errors than cross-sectional comparisons, and the significant group-by-time interaction effect confirms that the IVS group changed differently compared with the sham group, an outcome that is robust to measurement offset.

Finally, we did not collect joint kinetics, proprioception, metabolic cost, or patient-reported outcomes (PROMs); therefore, the clinical meaning of the kinematic shift remains inferential and should be confirmed.

Future studies should recruit larger, more diverse, assessor-blinded cohorts (ideally multi-centre) to increase power and enable subgroup analyses, validating the IMU angles against optical motion capture, and pair kinematics with kinetics, metabolic and proprioception measures, functional tests, and PROMs to establish clinical significance. Given the phase-specific response, trials should optimize IVS timing/dose by gait phase and task and follow participants longitudinally to evaluate durability and potential impacts on re-injury risk or post-traumatic osteoarthritis.

Conclusions

Our results indicate that IVS can acutely increase in minimum knee flexion during gait and stair ascent in individuals three months post-ACLR. These findings suggest that IVS may be a helpful adjunct in early rehabilitation. However, because the study was underpowered relative to the a priori calculation, these findings should be interpreted with caution, and future studies are warranted to confirm the multi-joint effects and explore their clinical significance.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

T.Y was responsible for the conceptualization and design of the study, formal analysis, investigation, and methodology. He also contributed equally to the original draft writing and the review and editing of the manuscript. B.P participated in the conceptualization and investigation, project administration, supervision, draft writing and review process. A.G.F, was responsible for project administration and resources, conceptualization, formal analysis, investigation, methodology, supervision, the original draft writing and review and editing of the manuscript.All authors contributed to the interpretation of the data and critically revised the manuscript for important intellectual content. All the authors finally approved the manuscript. A.G.F was responsible for obtaining project funding and takes responsibility for the integrity of the work as a whole. All authors have read and agreed to the published version of the manuscript.

Funding

Arielle Fischer was supported by the Zuckerman STEM Leadership Program. In addition, this research was supported by the Israel Science Foundation (grant No. 2070658). The funders had no role in the study design, data collection, analysis, interpretation, or decision to submit the manuscript for publication.

Data availability

The datasets generated and analysed during the current study are available in the Zenodo repository at: [https://zenodo.org/records/16022194](https:/zenodo.org/records/16022194).

Declarations

Competing interests

The authors declare no competing interests.

Declarations of interest

Arielle Fischer is on the SDI advisory team. She did not receive compensation as a member of the scientific advisory team. Tomer Yona and Bezalel Peskin declare no potential conflict of interest.

Ethics approval

The Rambam Health Care Campus Helsinki Committee approved this study (0089-21-RMB).

Clinical trial registration number

NCT05001594 (12.08.2021).