An articulated ankle foot orthosis with plantarflexion resistance decreased soleus and tibialis anterior muscle activity, and plantarflexion moment of the paretic lower limb of persons with stroke

Abstract

Ankle foot orthoses (AFOs) are important devices for assisting persons with stroke during walking, especially in mediating dorsiflexor weakness. The plantarflexion (PF) resistances of AFOs have been shown to augment dorsiflexor muscle function during the stance and swing phases of gait. However, it is not clear how this translates to the muscle activities of their paretic lower limbs. To further explore this question in stroke rehabilitation, 3D technology was used in designing and fabricating an individual-specific articulated AFO, which we assembled with a mechanical joint that enabled the application of PF resistances. We investigated the impact of regulating two PF resistances (PF1, PF2) on their paretic lower limb muscle activity. At PF2, there was significantly lower mean muscle activity for the Soleus in LR and the Tibialis anterior in the second rocker, pre-swing, and swing phases. The device only impacted the activity of the medial gastrocnemius muscle at the second rocker. The findings corroborated evidence regarding how AFOs with PF resistance impact the muscle activity of stroke survivors following a single session use; however, longitudinal studies are necessary to establish the impacts with long-term use, as most stroke survivors use AFOs long-term.

1. Introduction

Dorsiflexor muscle weakness is a common occurrence after a stroke (Kobayashi et al., 2010). The primary ankle dorsiflexor muscle, the tibialis anterior (TA), plays an important role in the early phase of stance during initial contact (IC) when it eccentrically contracts while the foot goes into plantarflexion (PF) to allow the body weight to be accepted at loading response (LR). It also maintains the foot in a neutral position through swing, contributing to swing limb clearance (Murayama and Yamamoto, 2020; Ohata et al., 2011). Conversely, weakness of the ankle dorsiflexors following a stroke could result in a ‘slap of the foot’ at IC / LR and a toe dragging through the swing phase of gait (Alam et al., 2014; Cullell et al., 2009). Foot slapping at IC / LR typically disrupts the first rocker phase of gait, which consequently sets up for a poor 2nd rocker of gait, resulting in a lack of stability in the stance phase of gait (Chen et al., 2018). Coupled with the toe drag through the swing phase of gait, these functional limitations result in poor mobility and balance in persons with stroke, thereby requiring the prescription of ankle foot orthoses (AFOs) with PF resistances (Yamamoto et al., 2011).

Articulated AFOs that enable independent adjustment of AFO characteristics have made it possible to adjust AFO PF resistance without impacting dorsiflexion (DF) resistance (Singer et al., 2014; Yamamoto et al., 2011), and 3-dimensional (3D) technology has enabled the ease of fabricating such devices. Compared to the manual method that entails plaster of Paris (POP) casting and post-trimming, the 3D technique aids in easily customizing AFO devices using 3D scanners and printers (Silva et al., 2022). The 3D methodology for making articulated AFOs has simplified the investigation of individual AFO characteristics, such as PF resistance, independently of others. AFOs with PF resistance can enhance foot clearance during the swing phase of gait while also improving the first rocker in individuals with stroke (Kobayashi et al., 2015, 2013; Singer et al., 2014; Yamamoto et al., 2011). The PF resistance instigates a resistive moment to rapid PF at the 1st rocker of gait, augmenting eccentric contraction of dorsiflexors to enable a clean heel strike to weight acceptance. An effective 1st rocker supports a smooth 2nd rocker function. However, excessive plantarflexor resistance can restrict their eccentric contraction, limiting controlled shank rotation over the ankle. (Kobayashi et al., 2015, 2013; Singer et al., 2014; Yamamoto et al., 2011). Evidence shows that an experimental articulated AFO with oil damper joints significantly changed the ankle joint kinematics of persons with stroke when its PF resistance was regulated through different settings (Kobayashi et al., 2013). Also, increasing the resistive PF moment of an articulated AFO decreases ankle plantarflexion through the phases of the gait cycle, and this also mediates the amount of knee hyperextension (genu recurvatum) by reducing the knee internal flexion moment (Kobayashi et al., 2015). While the impact of the PF resistance of some AFOs on the kinematic gait characteristics of stroke survivors has been previously investigated, it is not entirely clear how the PF resistance of an AFO impacts the muscle activity of persons with stroke during the observed changes in kinematics.

Current evidence shows that ankle foot-orthoses with PF resistance decreased the muscle force of the TA muscle and increased the gastrocnemius (GM) muscle force at the stance phase of gait in persons with stroke (Yamamoto et al., 2019), while another study showed an increased mean activity ratio of the TA muscle during loading response compared with other phases of the gait cycle when persons with stroke used an AFO with PF resistance for 2 months (Murayama and Yamamoto, 2020). The evidence explaining how AFO plantarflexion resistances impact muscle activity is variable. An understanding of how PF resistance impacts the muscle activity of stroke survivors can aid in making decisions on the prescription of articulated AFOs that enable independent adjustment of PF resistance (Yamamoto et al., 2011).

The present study used a custom individual-specific AFO design with 3D technology to investigate the impact of PF resistance of an articulated AFO on persons with stroke. 3D technology enabled the ease of designing custom AFOs with orifices to insert muscle activity sensors. We hypothesized that regulating the PF resistance – from the low to the high settings – would decrease the mean activity of the plantarflexor muscles and that of the TA muscles. This is because of the anticipated resistive effect on plantarflexor muscle function and evidence that showed decreased TA activity with increased PF resistance (Lairamore et al., 2011; Nikamp et al., 2019). The AFO effects on peak paretic ankle angle, moment, power, knee angle, and knee moment were also assessed along with the changes in the muscle activity level of the Rectus femoris (RF) and Biceps femoris (BF) muscles. We anticipated more typical joint kinematics, increased RF activity in swing, and decreased BF activity due to the augmentation effect of the PF resistance on the dorsiflexor muscle function.

2. Methods

2.1. Participant demographics

For this study, five participants with stroke (S01 – S05) were recruited (3 M, 2F). The participants ranged in age from 54 to 70 (years); (62.40 ± 5.71), had a stroke for more than 6 months (5.40 ± 5.39 years from a stroke), could walk independently with or without a walking aid, and two of the participants had a clinically prescribed AFO. Three of the participants had left-sided hemiparesis. Other inclusion criteria included a blood pressure between 90/60 and 170/90 mmHg and a resting heart rate between 40––100 beats per minute (bpm). Individuals who had pain in their legs or spine while walking, unexplained dizziness in the previous six months, ona-botulinum treatment (neurotoxin injection) in the lower limbs, and more than one stroke or a cerebellar stroke were excluded. The research was conducted over two visits to the University of Nebraska at Omaha’s Biomechanics Research Building. Before collecting written consent from participants at the first visit, the study was explained to them both orally and in writing. At the first visit, clinical tests such as measuring their passive PF and DF range of motion (ROM), manual muscle testing (MMT) of the paretic lower limb muscles using the Oxford muscle grading scale, 10-meter walk test, and the Timed Up and Go (TUG) test were performed for each participant. Also, 3D scanning of the participants’ shank-to-foot was performed to prepare for the design and printing of each participant’s AFO (Fig. 1). The University of Nebraska Medical Center’s Institutional Review Board approved the study.

Fig. 1.

The image above shows the Ankle Foot-Orthoses (AFO) fabrication process. (a) The 3D scanning process with the Creaform scanner. (b) Shows the image produced from the 3D scanner. (c) The ‘.stl’ image of the AFO in which the scanned image of each participant is fit. (d) The 3D-printed AFO calf and foot section which we assembled with the mechanical joint, is also shown. (e) Shows the assembled AFO with the orifices for fitting the EMGs. (f) A pilot participant wears the fabricated AFO.

2.2. Individual specific articulated ankle foot orthosis design and resistance Regulation

The articulated ankle foot orthosis used in this study was personalized for each participant using 3D scanning, 3D printing, and a commercially available AFO mechanical joint (Triple Action. ^® 2.0 by Becker Orthopedic Alliance CO., USA). The first visit included clinical assessment as well as scanning of each participant’s paretic leg from the shank to the foot (from the inferior border of the patellar cap to the toe and underfoot) with a Creaform scanner (Ametek, Ultra Precision Technologies, USA) (Fig. 1a). The participants’ paretic legs were scanned in a neutral position of the foot at 90° to the shank on a transparent scanning platform. A certified orthotist fitted the stereolithography (STL) image of each scan into a predetermined design for each participant’s footplate and calf section (Fig. 1b & c). The footplate design extends anteriorly to terminate just behind the proximal phalanx to enable the rolling movement of the foot at the terminal stance. The AFO foot plate was designed so that its anterior aspect encompassed the medial and lateral borders of the foot to ensure that the joints’ support forces were central to the sagittal plane. This was done to minimize frontal and transverse plane movements since hemiparesis is a tri-planar issue, and it is common clinical practice to control for supination/pronation and internal/external rotation. The final design for each participant was used in 3D printing of the calf section and foot plate with a Polylactic Acid (PLA) printing material (Fig. 1d). The calf section was printed in a BCN3D Epsilon W50 printer (BCN3D Technologies, Inc. Barcelona, Spain) due to its size while we printed the smaller foot section using a Prusa MINI + printer (Prusa Research a.s., Czech Republic). The EMG sensors for the medial gastrocnemius (MG) and Soleus (SO) muscles were placed on the respective muscles during the 3D scanning of each participant’s leg to account for their location (EMG sensors gap) in the calf section of the printed AFO device (Fig. 1e & f). Two Velcro straps (one at the calf section and the other at the foot section across the ankle) were used to secure the AFO to the limb. To create the final device, we assembled both 3D printed sections with the mechanical joint on the lateral side and a pivot joint on the medial side (Fig. 1e & f).

The study’s use of Becker’s Triple Action joint allows for independent adjustment of ankle alignment, PF / DF resistance, and AFO range of motion. For this study, we independently tuned the PF resistance settings by rotating (clockwise or counterclockwise) the posterior booster compartments housing the PF resistance spring, and the DF booster was unlocked throughout the entire collection. All participants had the same resistance settings (low-PF1 and high-PF2) for the walking trials. A clockwise rotation of the PF compartment increases the joint resistance (increases preload) by the number of turns in that direction, and a counterclockwise rotation decreases the resistance property (decreases preload). The moment – angle characteristics of the AFO were quantified within a range of ± 30Nm, equivalent to an angular range of ± 15° (Kobayashi et al., 2018, Kobayashi et al., 2017). We assessed the impact of regulating the 2 PF resistance settings as low- PF1(4 turns of the booster away from the locked position) and high- PF2 (locked position of the booster) referencing the graded resistances as quantified in the Kobayashi et al. (2018) study in which the same mechanical joints were used (Kobayashi et al., 2018). The stiffnesses were experimentally quantified using a custom-made device to apply loads similar to what would be applied during gait, quantifying PF1 as 0.074Nm/deg and PF2 as 0.573 Nm/deg. Each participant was fitted with their AFO device at the second visit. The range of motion and shank-vertical angles of the AFOs were not limited to a particular setting in order to isolate the effects of joint resistance for the participants.

2.3. Data collection

The participants were fitted with their custom ankle foot orthosis at the second visit. The fit, comfort, and perceived safety of the device were confirmed verbally with the participants. We randomized the PF resistance settings (PF1 and PF2). Forty-seven (47) retroreflective markers were placed on anatomical landmarks of the participants’ shoulder, sternum, torso, and lower extremities using a modified Cleveland marker placement protocol (Sutherland, 2002). For the foot markers, the markers were placed on the participant’s shoes on anatomical landmarks. The paretic leg lateral malleolus marker (ANL) was placed on the AFO triple-action joint. The offset distance from the anatomical lateral ankle malleolus to the ANL marker on the AFO joint was measured and referenced while calculating the affected ankle and foot kinematic variables. Five electromyography (EMG) sensors (Trigno Avanti, Delsys, Natick MA, USSA) were used in collecting the muscle activity of the rectus femoris (RF), biceps femoris (BF), medial gastrocnemius (MG), tibialis anterior (TA), and soleus (SO) for the affected lower leg. The EMG sensors were placed using guidelines from the text “Anatomical Guide for the Electromyographer” (Beattie, 1995).

The EMG system with a sampling rate of 1000 Hz was synchronized with a 20-camera motion capture system at 100 Hz (Motion Analysis Corporation, Rohnert Park, CA, USA) and 8 (4 big square-shaped and 4 small square-shaped) inground force plates at 1000 Hz (AMTI, Watertown, MA, USA), used in collecting muscle activity data and walking data respectively. The participants walked for 3 different walking trials along the same direction for both AFO resistance settings totaling 6 walking trials for each participant. The participants were allowed a 2-minute walking adaptation period to get accustomed to each resistance setting before data was collected. The participants were instructed to walk at their comfortable speed for the trials in each resistance setting of PF1 and PF2. They had a 5-minute rest period between each walking trial, and those who requested more rest time were allowed. A safety harness was worn by each participant for their safety throughout the walking period.

2.4. Data analysis

2.4.1. EMG analysis

The EMG data were collected with the motion capture system and were processed in Visual3D (C-Motion, Inc., Germantown, MD, USA). The EMG raw signal was stored as analog data exported into Visual3D. A bandpass filter between 50 and 500 Hz was applied to raw data to remove movement artifacts, and a linear envelope was computed from the average Root Mean Square (RMS) value obtained every 50 ms in each phase/window interval. The peak value from all walking trials of the PF resistance conditions was derived from the gait cycles of each muscle. All frames through the gait cycle were set to the maximum value, and the linear envelope signal was divided by the derived peak to normalize to the maximum of 1.0. This normalization method was used because of the difficulty in isolating a maximum voluntary contraction for stroke participants under isometric conditions (Murayama and Yamamoto, 2020). The EMG waveforms were normalized to a 100 % gait cycle to account for the different phases of the gait cycle, and the mean activation of all processed trials for each resistance condition was derived and utilized for the participants. The stance (loading response (LR), 2nd rocker, and third rocker or pre-swing) and swing phases of the muscle activity were exported, so the mean values for the muscle activity were derived at phases where we expected to detect the AFO impact; the RF in the stance (LR, 2nd rocker, and pre-swing), and swing phases, BF in the stance (LR, 2ndrocker, and pre-swing) and swing phases, TA in the stance (LR, 2nd rocker, and pre-swing) and swing phases, MG in the stance (LR, 2nd rocker, and pre-swing) phase, and the SO muscle in the stance (LR, 2nd rocker, and pre-swing) phases, as the AFO PF resistance aids weak TA function through stance and swing phases of gait.

2.4.2. Kinematic/Kinetic analysis

The walking data was exported to Visual 3D software (C-Motion, Inc., Germantown, MD, USA), where the different kinematic and kinetic variables were calculated. The data were filtered using a 4th-order low-pass Butterworth filter for the kinetic (cutoff frequency = 60 Hz) and kinematic data (cutoff frequency = 6 Hz). The peak joint angles were derived using the segment coordinate system as the maximum angles in the orientation of the main segment to a reference segment (usually the most proximal segment to the main segment) in the Cardan sequence of the capture environment. The kinetic variables were calculated using inverse dynamics analysis in Visual 3D while using a resolution coordinate system, which is the coordinate system of the nearest proximal segment to the joint. The mean of 9 clean steps was analyzed for each participant’s paretic limb, from the 3 trials for each condition, to derive the peaks for all gait cycles through the time series for each walking trial. The mean of all trials for each resistance condition was found for statistical analysis of the paretic ankle angle, moment, and power (stance and swing). This was repeated for the knee angle and moment (stance and swing) for each participant.

2.5. Statistical analysis

The mean differences in the muscle activity level of the derived muscles were statistically analyzed with a non-parametric test to accommodate for small sample size. The Wilcoxon signed-rank test was used to compare the differences in mean activities of the selected muscles at the PF1 and PF2 resistances across the participants at the highlighted phases of the gait cycle; the RF muscle in the stance (LR, 2nd rocker, and pre-swing) and swing phases, BF muscle in the stance (LR, 2nd rocker, and pre-swing) and swing phases, TA muscle in the stance (LR, 2nd rocker, and pre-swing) and swing phases, MG muscle in the stance phase (LR, 2nd rocker, and pre-swing), and the SO muscle in the stance phases (LR, 2nd rocker, and pre-swing) of the gait cycle. The mean of peaks of paretic ankle angle, moment, and power, and that of the knee angle and moment at the stance and swing phases were also analyzed with the Wilcoxon signed-rank test. Also, we assessed the differences in overground comfortable walking speed (through the anteroposterior sacral velocity). Friedman’s non-parametric test compared the 3 trials within the PF1 and PF2 resistance conditions.

Statistical analysis was performed in SPSS V. 26.0 (IBM Corp. Armonk, USA) with a p-value of ≤ 0.05 for determining statistical significance.

3. Results

3.1. Participants walking speed

The comfortable walking speeds for each participant within the walking trials of each resistance condition were not significantly different for each of the PF1 (x²= 1.20, p = 0.55) and PF2 (x²= 0.50, p = 0.78) conditions.

3.2. Participants muscle activity

Table 2 reports the means and standard deviations (including median and median absolute deviations) of the participants’ muscle activities (% MVC). The participants experienced significantly lower muscle activity for the TA muscles (Fig. 2) at the second rocker (PF1: 0.39 ± 0.12, PF2: 0.23 ± 0.12; Z = −2.02, p = 0.04), pre-swing (PF1: 0.33 ± 0.22, PF2: 0.23 ± 0.20; Z = −2.02, ρ = 0.04), and at swing (PF1: 0.42 ± 0.11; PF2: 0.30 ± 0.14; Z = −2.02, ρ = 0.04) for PF2 compared to PF1, while there was no significant difference between the resistances at the LR phase of gait (ρ = 0.23). The muscle activity for the SO muscle was significantly lower for PF2 than PF1 in the LR phase of gait (PF1: 0.34 ± 0.11; PF2: 0.23 ± 0.04; Z = −2.02, p = 0.04), and the PF resistances did not significantly impact the SO in the second rocker, pre-swing and at swing (p > 0.05). The participants’ MG muscle activity was not significantly affected by the PF resistance during the LR, pre-swing, or swing phases of gait (ρ > 0.05), but there was a significant decrease (PF1: 0.37 ± 0.13, PF2: 0.29 ± 0.17; Z = −2.02, p = 0.04) at PF1 compared to PF2 at the second rocker.

Table 2.

Mean, standard deviation, median, and median absolute deviations of participants’ muscle activities at resistances PF1 and PF2.

Muscles	Gait Phases	Muscle activities
		PF1		PF2		p
		Mean ± SD	Median, MAD	Mean ± SD	Median, MAD
Tibialis anterior	LR	0.54 ± 0.08	0.51, 0.06	0.44 ± 0.12	0.46, 0.08	0.23
	SR	0.39 ± 0.12	0.40, 0.09	0.27 ± 0.14	0.23, 0.12	0.04
	PS	0.33 ± 0.22	0.30, 0.17	0.23 ± 0.20	0.15, 0.14	0.04
	Swing	0.42 ± 0.11	0.37, 0.09	0.30 ± 0.14	0.31, 0.11	0.04
Medial gastrocnemius	LR	0.29 ± 0.11	0.26, 0.08	0.23 ± 0.15	0.19, 0.12	0.14
	SR	0.37 ± 0.13	0.38, 0.09	0.29 ± 0.17	0.31, 0.12	0.04
	PS	0.35 ± 0.25	0.29, 0.18	0.39 ± 0.13	0.36, 0.10	0.35
	Swing	0.22 ± 0.11	0.17, 0.09	0.21 ± 0.09	0.21, 0.07	0.69
Soleus	LR	0.34 ± 0.11	0.32, 0.08	0.23 ± 0.04	0.24, 0.03	0.04
	SR	0.28 ± 0.18	0.31, 0.15	0.28 ± 0.18	0.38, 0.15	0.69
	PS	0.32 ± 0.25	0.34, 0.20	0.34 ± 0.27	0.35, 0.22	0.50
	Swing	0.18 ± 0.10	0.18, 0.07	0.14 ± 0.06	0.14, 0.04	0.50
Rectus Femoris	LR	0.54 ± 0.13	0.53, 0.10	0.49 ± 0.14	0.46, 0.10	0.04
	SR	0.42 ± 0.11	0.38, 0.08	0.36 ± 0.11	0.30 ± 0.11	0.08
	PS	0.19 ± 0.03	0.18, 0.02	0.17 ± 0.05	0.15, 0.04	0.23
	Swing	0.29 ± 0.15	0.29, 0.12	0.23 ± 0.10	0.24, 0.08	0.08
Biceps femoris	LR	0.48 ± 0.10	0.45, 0.08	0.42 ± 0.10	0.47, 0.08	0.08
	SR	0.27 ± 0.13	0.22, 0.10	0.22 ± 0.13	0.21, 0.09	0.04
	PS	0.12 ± 0.10	0.10, 0.08	0.11 ± 0.09	0.07, 0.07	0.23
	Swing	0.29 ± 0.04	0.29, 0.04	0.24 ± 0.05	0.25, 0.03	0.04

Note: LR – loading response; PS –pre-swing; SR – Second Rocker; MAD – Median Absolute Deviation.

Fig. 2.

Graph of the mean muscle activity at low resistance (PF1) and high resistance (PF2) conditions for the assessed muscles at the loading response (LR), second rocker (SR), pre-swing (PSW), and swing (SW) phases of gait (normalized to 1.0). Significant differences in comparisons are represented with ‘*’. The X-axis labels are muscles attached to gait phases for each muscle activity mean (BF – biceps femoris, MG – medial gastrocnemius, RF – rectus femoris, SO – Soleus, TA – tibialis anterior).

The RF muscle in the LR phase of the gait cycle was significantly lower in PF2 than in PF1 (PF1: 0.54 ± 0.13; PF2: 0.49 ± 0.14; Z = −2.02, p = 0.04) (Fig. 2), and that of the BF in the second rocker (PF1: 0.27 ± 0.13, PF2: 0.22 ± 0.13; Z = −2.02, p = 0.04) and at swing phase of gait significantly decreased at PF2 compared to PF1 (PF1: 0.29 ± 0.04; PF2: 0.24 ± 0.05; Z = −2.02, p = 0.04). There were no significant differences in RF muscle activity at the second rocker, pre-swing, and swing phases of the gait cycle nor in the BF muscle activity at the LR and pre-swing phases of gait (p > 0.05) for comparison of both resistance conditions.

3.3. Paretic lower extremity kinematics

The time series plots of the mean ankle angle, moment, and power and the knee angle and moment of the participants showed changes when the AFO PF resistance was regulated (Fig. 3). These changes were only significantly different for the peak ankle PF moment at the swing phase of gait, which was lower at PF2 than at PF1 (PF1: 0.27 ± 0.22 Nm/Kg; PF2: 0.21 ± 0.18 Nm/Kg; Z = −2.02, p = 0.04). Regulating the AFO PF resistance did not significantly impact the peak ankle angle in stance and swing, the peak ankle PF moment in stance, the peak ankle power in stance and swing, nor in the peak knee angles and peak moments in stance and swing (p > 0.05).

Fig. 3.

Time series graph of the mean kinematics and kinetics of the participants at PF1 and PF2. (a) plot of the Ankle angle. (b) plot of the Ankle moment. (c) plot of the Ankle power. (d) plot of the knee angle. (e) plot of the knee moment. All are normalized to % gait cycle.

4. Discussion

In this study, we fabricated a custom articulated AFO assembled with mechanical joints to evaluate how plantarflexor resistance impacted the muscle activity level of the lower extremity muscles of persons with stroke. Two plantarflexion resistances – (PF1) and (PF2) – were evaluated for all participants with the articulated AFO. Each articulated AFO was customized to the participants by using 3D technology to scan the anatomical features of each participant’s leg to their foot and having an orthotist make modifications to control for frontal and transverse plane movement through the AFO footplate design. The results showed that this articulated AFO with mechanical joints significantly changed the mean muscle activity for some of the evaluated muscles when the mean activations were compared at low and high PF resistances, and it only impacted the ankle moment in swing but not ankle joint angle, moment, or power in stance.

The PF resistance of an AFO opposes PF movement of the ankle during gait, augmenting the function of weak TA muscles to provide controlled loading in gait and prevent foot drop during swing to improve the gait characteristics. Here, higher PF resistance settings decreased the mean activation of the TA muscle when compared to the lower PF resistance settings at the second rocker, pre-swing, and swing phases of gait. This finding corroborates evidence that tuning the PF resistance of AFOs may impact the activation of TA muscles of persons with stroke (Murayama and Yamamoto, 2020; Yamamoto et al., 2019). Specifically, as observed in the mean muscle activations of participants (Table 2) in this study, increasing PF resistance of an AFO may decrease TA muscle demand of stroke survivors (Yamamoto et al., 2019). The observed decreased TA action in this study was significant in all gait phases except the LR, and this may be because the utilized resistances were not objectively prescribed for each participant; the resistance to PF movement of the custom AFO may have compensated for the dorsiflexor muscle function of some participants more distinctly when compared to others, resulting in the observed lower mean activity of the TA muscle in all gait phases but the LR, at PF2 resistance compared to PF1.

For the plantarflexor muscles in the stance phase of gait, a study with 10 healthy participants reported that increasing the PF resistance of an articulated AFO resulted in an increased MG muscle force between different PF resistance conditions (Yamamoto et al., 2019). Contrary to their findings in healthy individuals, this study found that for individuals’ post-stroke, changing the AFO PF resistance settings decreased the MG muscle function only at the second rocker while there was no impact in other phases of gait, and there was also decreased SO muscle activity at PF2 than PF1 only at the LR. AFOs with PF resistances are designed to resist ankle PF at initial contact to LR of gait, thereby improving the first rocker and enabling weight acceptance (Yamamoto et al., 2009). This action may decrease the activity of PF muscles, as observed in SO muscle activity during the LR in this study. The lack of impact on the MG muscle during LR, PS, and swing, and the SO muscle at pre-swing and swing, may be attributed to the low resistances PF1 and PF2 applied through the articulated AFO in this study. The utilized resistances were within the limits of the commercial articulating joints assembled with the fabricated AFO calf and foot plate section. Since most participants had plantarflexors muscle strength graded up to 4 on the MMT (Table 1), using higher PF resistances might have had a more significant impact on their gait and muscle activity.

Table 1.

This table shows subject demographics and clinical characteristics. (MMT – manual muscle testing using the Oxford muscle grading scale; 10 MWT – Ten-meter walk test).

Subject	Sex	Age (Years)	Years since Stroke	Paretic Limb	Assistive device	Clinically Prescribed AFO	10MWT speed (seconds)	PF MMT	DF MMT
S01	Male	64	16	L	Single point cane	dynamic carbon fiber	7.71	2	3 −
S02	Female	70	3	R	Single point cane	dynamic carbon fiber	5.22	4	3 +
S03	Male	66	4	L	None	none	6.11	5	4
S04	Male	54	1	L	None	none	3.65	4+	4 +
S05	Female	58	3	R	Single point cane	none	7.57	4	4

The PF2 resistance also decreased the mean activity of the RF muscle during LR and the BF muscle at the second rocker and swing phases of gait. Studies show that AFOs could indirectly impact the knee joint gait characteristics of persons with a stroke while walking (Bleyenheuft et al., 2010; Kobayashi et al., 2016, 2015). They showed that increasing the PF resistance decreased excessive hyperextension of the knee with a resultant reduced knee flexion moment at the stance phase of gait (Bleyenheuft et al., 2010; Kobayashi et al., 2016, 2015). While the results here showed no significant changes in knee angles and internal moments, the observed lower activity of the RF muscle at PF2 during LR likely resulted from the observed continued extension moment of the knee joint through the 1st rocker phase of gait as observed in the knee moment time series plot (Fig. 3e). This impact on the 1st rocker phase of gait may influence the 2nd through the 3rd rocker, where the time series plots showed increased PF moment and knee flexion moment during pre-swing (Fig. 3b and 3e). The observed increase may have been statistically significant with improved magnitude of PF resistances compared to the low resistances used in this study, while the lower mean activity of the BF muscle through the swing phase may have resulted from the increased knee flexion moment during pre-swing.

The possibility of decreased muscle activity and atrophy with long-term AFO use has been evaluated from different perspectives. Like this current study, other studies with different AFO designs have shown reduced muscle activity, especially the TA muscle while using AFOs in a single session (Geboers et al., 2002; Hesse et al., 1999; Lairamore et al., 2011; Mulroy et al., 2010; Yamamoto et al., 2019). The long-term use of some AFO types may result in decreased muscle activity that could cause muscle atrophy, yet studies like that of Nikamp et al. (2019) have shown that a reduction in muscle activity following a single session use of an AFO device may not result in decreased muscle activity after long-term use (Kim et al., 2016; Nikamp et al., 2019). Also, it is necessary to reiterate that the heterogeneity of the stroke population presumes there are persons with a wide range of clinical presentations with the level of involvement being mild to severe. This implies that some individuals may be non-functional without these AFOs, with the devices serving to improve other primary gait parameters and functional outcomes that could make consideration for its impact on muscle activity secondary. Therefore, it is certain that the magnitude of muscle force is a result of the activities of individuals (Bohannon, 2007), nevertheless, the high level of importance of AFOs as necessary walking aids for stroke survivors establishes the need for continual research. Investigations are essential to understand favorable conditions that could improve gait performance while reinforcing the physiological capabilities of their lower extremity muscles in the long term. While this study shows how the AFO device with PF resistance impacts the mean muscle activity and walking characteristics of the paretic lower extremities of stroke survivors in a single session, it is uncertain how this will translate with the long-term use of the device. This is a study limitation, considering that AFOs are mostly prescribed for long-term use. Future research should evaluate the impacts of PF resistance of the articulated AFO device in a longitudinal study design. Furthermore, the study did not specify the range of motion limitations or the shank vertical angle of the AFOs, nor were the dorsiflexion settings of the AFO joints individually tuned to accommodate each participant’s unique presentation; factors that could influence lower-limb kinematics and the ankle position in which the resistances are applied. This was done to project the impact of the AFO PF resistance; future work should integrate these AFO design and prescription components to be more individual-specific in the heterogeneous population of post-stroke individuals. An additional limitation of this study is the small sample size, among which some participants were previous AFO users, and others may not have been clinically indicated for an AFO. We also used the same general AFO design and resistance settings regardless of clinical presentation. This may have restricted the exploration of the broad dynamics of the stroke population, considering this population is highly heterogeneous. However, the approach to customizing this AFO enabled us to isolate the effects of the plantarflexion resistance to the impact of the mechanical joint on the sagittal plane only.

In conclusion, this study showed how the PF resistance of an articulated AFO designed and fabricated using 3D technology impacts the mean muscle activity and walking characteristics of the paretic lower extremity of persons with stroke. The stroke participants walked with the device in a single walking session. The higher PF resistance decreased the mean activity of their TA muscle at the pre-swing and swing phases and their SO muscle at the LR phase of gait compared to the lower PF resistance. At the same time, it did not significantly impact the mean activity of the MG muscle, but it decreased the ankle PF moment in swing. The device also reduced the mean activity of the RF muscle in LR and the BF muscle at the swing phase of gait. Despite the findings of this single-session study, there is a need for longitudinal studies to establish the impact of PF resistance of an AFO on the muscle activity of stroke survivors following the long-term use of the devices. This will give a clearer picture of how regulating PF resistance impacts muscle performance and physiology in the long term.