Effect of using knee extension aid for gait training in subacute hemiplegia: a randomized controlled trial

Abstract

[Purpose] In normal gait, the knee flexes approximately 60° during the swing phase, and the lower leg’s inertial force contributes to knee extension during terminal swing, enabling heel-first initial contact and energy-efficient gait. However, slow-walking individuals with hemiplegia generate insufficient inertial force, resulting in inadequate knee extension during terminal swing and difficulty achieving heel strike. When initial forefoot contact occurs, ankle plantar flexors may become excessively active during loading response, reinforcing abnormal movement patterns. This randomized controlled trial investigated the therapeutic effect of a knee extension aid during terminal swing. [Participants and Methods] We randomized 30 hemiplegic inpatients who could perform gait training using an ankle-foot orthosis in convalescent rehabilitation wards at five facilities in Japan into two groups. The intervention group used the ankle-foot orthosis and knee extension aid during gait training, while the control group used the ankle-foot orthosis alone. We measured knee joint angles during comfortable-speed walking before and after 2 weeks of gait training. [Results] In the intervention group, knee joint angles during terminal swing significantly decreased by approximately 4° after 2 weeks. [Conclusion] The knee extension aid may facilitate heel-first initial contact and reduce abnormal ankle muscle activation by decreasing knee flexion during terminal swing.

INTRODUCTION

In normal gait, the knee flexes to approximately 60° during the swing phase, and the inertial force of the lower leg contributes to near-full knee extension during terminal swing. This movement enables heel-first initial contact, promoting energy-efficient gait¹⁾. In contrast, excessive knee flexion during walking has been noted as a typical gait pattern in individuals with hemiplegia who walk at slower speeds²⁾. Because slow-walking hemiplegic patients generate insufficient inertial force to advance the lower limb as a double pendulum during the swing phase, they often do not achieve adequate knee extension during terminal swing, making heel strike difficult. When heel strike does not occur, the ankle plantar flexors tend to become excessively active during loading response, reinforcing abnormal movement patterns. Therefore, achieving sufficient knee extension during terminal swing is clinically important.

Murayama and Yamamoto reported that, in patients with subacute hemiplegia, using an ankle-foot orthosis (AFO) that facilitates heel-first initial contact without restricting ankle plantar flexion during loading response for 2 months increased the tibialis anterior activity ratio and decreased the forward tilt angle during loading response³⁾. This finding indicates that heel-first initial contact plays a crucial role in promoting more physiological muscle activation patterns. During gait training, physical therapists often support patients from behind to ensure safety and guide weight shifting. However, when assisting from behind, therapists are unable to support the lower leg, making it difficult to facilitate knee extension. Thijssen et al. developed a device consisting of elastic straps running from the trunk to the foot that assist hip and knee flexion and ankle dorsiflexion during the swing phase. They reported that using this device reduced the energy cost of walking and improved gait speed and stride length compared with walking without it⁴⁾. However, because the elastic straps exert their maximum tension during pre-swing to initiate limb advancement, they do not necessarily facilitate knee extension during terminal swing.

Other methods to facilitate knee extension include using knee-ankle-foot orthoses (KAFO) equipped with spring-loaded knee joints; however, the springs also produce extension forces during initial swing, which may impede knee flexion. Electrically controlled orthoses that regulate knee motion have been developed. However these devices are costly and not typically fabricated for individual patients during the several-month-long convalescent rehabilitation period. Consequently, no simple assistive device is currently available that allows physical therapists assisting from behind to promote knee extension during terminal swing in gait training for patients with subacute hemiplegia.

To address this issue, the authors developed a knee extension aid that enables therapists to facilitate knee extension during terminal swing during assisted gait (utility model registration number 3232914). In a prior study, using this device in combination with a KAFO set to allow free knee flexion resulted in reduced knee flexion angle at initial contact compared with a standard assistive loop⁵⁾. However, gait using a KAFO with free knee flexion differs biomechanically from gait with an AFO. Moreover, because that study did not compare the device with standard assistance methods, its therapeutic effect remains unverified.

Randomized controlled trials with 2-week intervention periods have been recommended to evaluate the therapeutic effects of orthoses in patients with subacute hemiplegia⁶⁾. Furthermore, to appropriately assess the therapeutic effect of the knee extension aid, gait should be evaluated before and after a 2-week intervention under conditions without use of the knee extension aid in both the intervention and control groups, so that observed changes reflect therapeutic effects of training rather than the immediate mechanical effects of the device.

Therefore, the present study conducted a randomized controlled trial to investigate the therapeutic effect of a knee extension aid that supports knee extension during terminal swing by comparing gait parameters before and after a 2-week gait training intervention.

PARTICIPANTS AND METHODS

This study was conducted at five convalescent rehabilitation hospitals in Japan. Eligible participants were patients with hemiplegia admitted to convalescent rehabilitation wards who were able to undergo gait training using an AFO. Exclusion criteria included severe communication difficulties and marked limitations in the range of motion or contractures of the knee or ankle. Participants were recruited using convenience sampling. A total of 30 patients were assessed for eligibility, and the trial was reported in accordance with the consolidated standards of reporting trials (CONSORT) 2010 Statement⁷⁾. Using a random number table, participants were randomly assigned to either (1) the intervention group, which performed gait training with the knee extension aid or (2) the control group, which performed gait training without the aid. A total of 15 participants were assigned to the intervention group and 15 to the control group between October 2023 and September 2025 (Fig. 1).

Fig. 1.

Consolidated standards of reporting trials (CONSORT) flow diagram of participant inclusion.

This study was approved by the Research Ethics Committee of Niigata University of Health and Welfare (approval number 18720-210916). Written informed consent was obtained from all participants following both verbal and written explanations of the study procedures. This clinical trial was registered in the UMIN Clinical Trial Registry on May 23, 2023 (UMIN000044875).

The knee extension aid was originally developed for use with a KAFO but was later modified so that it could also be used with an AFO to broaden its applicability. The device consists of a thigh cuff secured by a pelvic belt, a cylindrical grip mounted horizontally across the anterior thigh cuff with a pulley at each end, and an elastic strap that runs from the pulley anterior to the knee and attaches to the shank of the AFO. The therapist stands behind the patient and grasps the pulley assembly. When the therapist flexes the wrist, the cylindrical grip rotates, winding approximately 5 cm of the elastic strap around the pulley and transmitting an extension-assist force to the knee via the AFO shank (Figs. 2, 3). During therapist-assisted gait from behind, the therapist operates the aid by holding the pulley assembly and applying volar wrist flexion during terminal swing. This action generates tension in the elastic strap, thereby assisting knee extension during terminal swing.

Fig. 2.

Schematic of device components.

Fig. 3.

Photo of the device being worn.

In both groups, knee joint angles were measured during comfortable-speed walking over a distance of approximately 10 m, with participants wearing only an AFO. In the intervention group, measurements were obtained before and after the 2-week training period, during which participants trained with both the AFO and the knee extension aid. In the control group, measurements were obtained both before and after 2 weeks of training with the AFO alone. Knee joint angles were measured at 1,000 Hz, using inertial sensors (ATR-Promotions; AMWS020B) attached to the anterior surfaces of the thigh and shank. The relative angle between the thigh and shank was calculated to determine the knee joint angle during walking.

For both the intervention and control groups, continuous data from three gait cycles were collected at baseline and after the 2-week training period. Initial contact of the paretic limb was identified from characteristic spikes in the vertical acceleration signal recorded by the inertial sensors. Each sequence was segmented into individual gait cycles and averaged. The knee joint angle was expressed as the angular displacement relative to the position at initial contact, which was defined as 0°. The gait cycle was divided according to standard gait phases—initial contact (0–2%), loading response (2–12%), mid-stance (12–31%), terminal stance (31–50%), pre-swing (50–62%), initial swing (62–75%), mid-swing (75–87%), and terminal swing (87–100%)¹⁾—and mean knee joint angles were calculated for each phase. Categorical variables were compared using the χ² test. Knee joint angles at baseline and after 2 weeks were compared between groups, using the Friedman test; if significant differences were found, a multiple comparison using the Wilcoxon signed rank sum test with Bonferroni correction was performed. Statistical analysis was performed using JSTAT software (ver. 16.1) with a significance level of 5%.

RESULTS

The clinical characteristics of the intervention group (n=15) and the control group (n=15) are shown in Table 1. No significant differences were observed between the groups in age, sex, height, weight, paretic side, time from the stroke onset to the measurement date, Brunnstrom recovery stage of the lower limb, Modified Ashworth Scale for the plantar flexors, Functional Independence Measure score, Stroke Impairment Assessment Set score, or Berg Balance Scale score. The median knee joint angles for each gait phase at baseline and after 2 weeks in each group are presented in Table 2. In the intervention group, the knee joint angle during terminal swing significantly decreased from 3.5° at baseline to −0.5° after 2 weeks, indicating an improvement of approximately 4° toward greater knee extension. No significant changes were observed in the other gait phases. In the control group, no significant changes were observed in any gait phase, including terminal swing.

Table 1. Clinical characteristics of the 30 stroke patients.

Item	Intervention group (15 stroke patients)	Control group (15 stroke patients)
Age (years)	66.5 ± 13.5	65.1 ± 19.3
Sex (male/female)	10/5	8/7
Height (cm)	162.3 ± 8.5	162.1 ± 7.6
Weight (kg)	59.0 ± 11.0	61.6 ± 12.6
Paralyzed side (right/left)	7/8	6/9
Stroke onset to measurement date (days)	102.8 ± 35.1	127.2 ± 39.3
Lower limb Brunnstrom stage (II/III/IV/V)	0/4/8/3	2/3/9/1
Plantar flexor Modified Ashworth Scale (0/1/1+/2/3)	4/4/3/3/1	4/6/1/2/2
Functional Independence Measure full score 126 (IQR)	96 (16.5)	97 (26)
Stroke Impairment Assessment Set full score 76 (IQR)	52 (12.5)	52 (15.5)
Berg Balance Scale full score 56 (IQR)	43 (9)	45 (10)

Statistical significance was assessed using the Mann–Whitney U test or χ² test as appropriate, with p<0.05 set as the level of significance. No significant differences were found. SD: standard deviation.

Table 2. Initial measurements and results after 2 weeks for knee joint angle in the intervention and control groups.

Item	Intervention group (15 stroke patients)					Control group (15 stroke patients)
	Baseline		2 weeks later		p-value	Baseline		2 weeks later		p-value
	Median	IQR	Median	IQR		Median	IQR	Median	IQR
Initial contact	−0.3	0.6	−0.2	0.5	0.482	0.0	0.5	0.0	0.5	0.917
Loading response	0.4	4.6	1.4	6.2	1.000	4.4	4.7	3.9	2.4	0.917
Mid-stance	−2.7	8.9	−8.6	13.0	1.000	−0.7	11.1	1.5	8.4	1.000
Terminal stance	−9.7	9.2	−10.2	13.1	1.000	−7.7	9.7	−4.2	9.0	0.378
Pre-swing	−2.5	18.9	4.3	16.3	1.000	−0.4	18.2	4.3	12.4	1.0
Initial swing	16.8	26.5	13.2	16.1	1.000	8.4	24.4	13.1	20.8	0.482
Mid-swing	10.2	7.8	16.0	13.8	0.086	6.7	12.0	8.9	7.5	0.831
Terminal swing	3.5	3.0	−0.5	2.5	0.008*	1.3	3.2	1.3	3.7	1.000

Flexion direction is +. *Significant difference according to the Wilcoxon signed rank sum test with Bonferroni correction (p<0.05). IQR: interquartile range.

DISCUSSION

The intervention group underwent 2 weeks of gait training, using an AFO with the knee extension aid. Over this period, the knee joint angle during terminal swing significantly decreased compared with baseline. In our previous study, when a KAFO was adjusted to allow knee flexion and used together with our knee extension aid, the knee flexion angle at initial contact decreased compared with that observed when using a conventional assist loop⁵⁾. However, because the previous study compared gait with and without the aid, its findings were limited to demonstrating the immediate effects of wearing the device. In contrast, the present study compared gait with only an AFO before and after a 2-week intervention period using the knee extension aid. Therefore, the observed reduction in knee joint angle during terminal swing reflects the therapeutic effect of the intervention.

Concerns have been raised that immobilization of the ankle joint in an AFO may lead to disuse atrophy of the tibialis anterior^{8, 9)}. However, Murayama and Yamamoto reported that compensating for heel-first initial contact increases both the relative activation of the tibialis anterior and the forward tilt angle of the shank during loading response³⁾. Additionally, according to Swayne et al., cortical excitability increases at approximately 3 months post-stroke, and intercortical networks are reorganized to maximize the efficiency of the residual corticospinal tract as part of the motor recovery process¹⁰⁾. Furthermore, Kamibayashi et al. demonstrated that, during passive gait using a robotic device for the paretic lower limb, corticospinal excitability to the tibialis anterior increases to a greater extent than that to the rectus femoris, biceps femoris, or soleus¹¹⁾. By promoting heel-first initial contact, the knee extension aid facilitates ankle plantar flexion during loading response. This movement passively stretches the tibialis anterior during each gait cycle, and the repeated afferent sensory input may enhance the excitability of the corticospinal tract and intercortical networks. Taken together, these findings suggest that the use of a knee extension aid can reduce knee flexion during terminal swing and promote heel-first initial contact, thereby potentially preventing the disuse atrophy of the tibialis anterior that has been a concern in previous studies.

This study has several limitations. Although participants were randomly assigned to the intervention and control groups, neither the participants nor the physical therapists who conducted the gait training were blinded. Therefore, psychological influences on both the participants and therapists cannot be ruled out. In addition, although no statistically significant difference was found between the groups in time from stroke onset to the first measurement, the intervention group underwent initial assessment approximately 4 weeks earlier than did the control group, which may have influenced the outcomes. Future research should ideally standardize the participants’ recovery stages and increase the sample size. In this study, the knee joint angle extended by approximately 4° during the terminal swing, but it is necessary to consider whether this angle is clinically meaningful. It is said that the ankle joint plantarflexes by 5° during the loading response phase. In cases where initial contact from the heel is difficult and full-foot contact is observed, it is thought that a 4° extension of the knee joint will ensure initial contact from the heel as well as the plantarflexion angle during the loading response. However, to use the knee extension aid effectively and with the correct timing, it is thought that physical therapists need to practice beforehand.

In conclusion, the use of a knee extension aid during 2 weeks of gait training with an AFO in subacute hemiplegic patients reduced knee flexion during terminal swing. These results suggest that the device promotes heel-first initial contact and may help reduce abnormal movement patterns of the ankle dorsiflexor and plantar flexor muscles.

Funding

This study was supported by JSPS KAKENHI (grant number 23K16657).

Conflict of interest

The authors declare no conflicts of interest related to this study.

Funding Statement

This study was supported by JSPS KAKENHI (grant number 23K16657).