Functional electrical stimulation during walking in children with unilateral spastic cerebral palsy: A randomized cross‐over trial

To study if functional electrical stimulation (FES) of the peroneal nerve, which activates dorsiflexion, can improve body functions, activities, and participation and could be an effective alternative treatment in individuals with unilateral spastic cerebral palsy (CP).

Functional electrical stimulation (FES) has been suggested as an alternative treatment because it might function as a dynamic functional orthosis. 14,15FES is defined as 'the electrical stimulation of muscles (that might have impaired motor control) to produce a contraction to obtain a functionally useful movement'. 16,17FES during walking can be applied to several muscles, including the quadriceps and gluteus maximus; however, FES of the ankle dorsiflexors via stimulation of the peroneal nerve is the most common non-invasive application. 14Stimulation of the ankle dorsiflexors (mainly the anterior tibial muscle) is useful in individuals walking with insufficient ankle dorsiflexion, such as foot drop gait (gait type I according to Rodda and Graham 3 ).The effects of FES on the ankle dorsiflexors have been reported in the International Classification of Functioning, Disability and Health (ICF) 18 domains of body functions and structures as an increase in ankle dorsiflexion during walking (2-12° compared to without FES), step length, and ankle dorsiflexion strength (see the systematic review by Moll et al. 14 for details on the heterogenous outcome measures).Little is known about the effect of FES on the domain of activity and participation: only one study reported that FES decreases the frequency of toe drag and improves self-perceived performance and satisfaction. 14More evidence is needed on patient-centred outcome measures and long-term effects in general (such as ankle mobility).Furthermore, practical guidelines for patient selection and FES settings are needed.
We hypothesized that FES of the peroneal nerve (activating the ankle dorsiflexors) can improve body functions, activities, and participation and might be an effective alternative to AFOs in daily life for individuals with CP.The trial objectives included all domains of the ICF. 18The primary objective was to show whether FES of the ankle dorsiflexors improves the activities and participation of individuals with unilateral spastic CP in daily life, while assessing individual goals using the Goal Attainment Scale (GAS). 19The secondary objective was to determine whether FES has an effect on body functions and activities, as assessed using three-dimensional gait analysis, questionnaires, and physical examination.The tertiary objective was to collect relevant information on the acceptability of FES and patient selection that could guide the implementation of FES in clinical practice.

M ET HOD
The full, detailed study protocol of this trial was described previously. 20This randomized cross-over intervention was performed on 25 children with unilateral spastic CP.Patients aged between 4 and 18 years, classified in Gross Motor Function Classification System (GMFCS) level I or II, with unilateral foot drop of central origin, and currently undergoing treatment with an AFO or adapted shoes were included.All participants underwent 12 weeks of FES treatment and 12 weeks of conventional treatment (AFO or adapted shoes), separated by a 6-week washout period.The treatment sequence was randomized.FES treatment consisted of wearing the WalkAide device (Innovative Neurotronics, Reno, NV, USA) daily, with surface electrodes stimulating the peroneal nerve during the swing phase of gait (from just after toeoff to initial contact).Stimulus intensity and device wearing time were gradually increased in the first 4 weeks of FES therapy.
For the primary objective, two individual goals were defined for each patient according to the GAS: GAS 1 was aimed at anything that can be linked to FES treatment; GAS 2 was aimed at walking distance.The Cerebral Palsy Quality of Life questionnaire was used to assess participation.
The secondary objective was to determine whether FES is effective in the domains of body functions and activities, by means of a physical examination, the 6-Minute Walk Test distance, questionnaire-based frequency of falling, and ankle kinematics and kinetics measured during three-dimensional treadmill-based gait analysis (Computer Assisted Rehabilitation Environment [CAREN], MOTEK Medical, Amsterdam, the Netherlands]). 20The definitions and calculations used for the kinematic results can be found in Appendix S1.All measurements were performed twice per treatment phase (measurement 1 at the beginning and measurement 2 at 12 weeks of the treatment phase).Physical activity monitoring, measured as the average number of steps per day, was performed three times.
The tertiary objective was to collect relevant information for clinical implementation, including acceptability using the device log file and the registration of side effects.It also included patient characteristics, such as treatment history (e.g.botulinum neurotoxin A injections or surgery) and the type of brain abnormality assessed using magnetic resonance imaging (MRI).
The sample size calculation was based on the primary outcome measure, that is, the GAS dichotomized into 'goal achieved' (score 0 to +2) and 'goal not achieved' (score − 3 to −1).A difference of 30% in the proportion of participants achieving their goals per treatment phase was assumed to be clinically relevant.To be able to detect this difference with a power of 80%, using an alpha of 5%, 22 participants were needed based on the formula to calculate a sample size for a cross-over design.Twenty-five participants were included to compensate for possible dropout (10%).The study was approved by the medical ethical board of the azM/UM (study no.172033/NL63250.068.17) and was registered as a clinical

What this paper adds
• Plantarflexion increased while using functional electrical stimulation (FES).• FES was preferred by 50% of patients with unilateral cerebral palsy classified in Gross Motor Function Classification System levels I and II.• FES helped some patients to improve participation by achieving individual goals.
trial (Clini calTr ials.govregistration: NCT03440632).All participants (aged from 12 years onwards) or caregivers (if the participant was younger than 16 years) signed an informed assent and consent form before participating in the study.SPSS v27 (IBM Corp., Armonk, NY, USA) was used for all the analyses; p < 0.05 was considered statistically significant.Baseline data were described using summary statistics.Two-tailed tests were applied.The GAS was analysed using the non-parametric McNemar test.Secondary outcomes were analysed using the Wilcoxon signed-rank test on the change scores.d Surgery after completion of the FES on participation (FESPa) trial (surgery had already been arranged before the FESPa trial).
e Reasons for dropout: (1) FES therapy was not feasible because of a very thin leg (due to growth restriction) and the need for precise placement of the electrodes; (2) the ankle range of motion deteriorated during the conventional treatment phase, thus conflicting with the exclusion criteria; (3) the patient could not walk well without an AFO or with FES because of too much spasticity of the tibialis posterior muscle; (4) the patient was no longer interested in the trial because of delays related to the COVID-19 pandemic and a new walking aid instead of an AFO (foot-up); (5) the patient was unable to start the trial because of clinical complications; (6) the patient could not get used to FES and was afraid of the sensation of the current, making it impossible to increase the intensity to any effective level; (7) after 1 week of FES therapy, the use of any walking aid was stopped because of social and emotional problems.The problems were not specific to the FES device but were pre-existing when using an AFO.f This patient underwent computerized tomography instead of an MRI.The MRI score can not be applied.
foot clearance [MFC] in centimetres, and peak foot progression [in degrees]), linear mixed-effects models were evaluated, with 'condition' (AFO, no AFO, FES, no FES) and 'time' (measurement 1 or 2) as fixed factors, an interaction term 'condition*time' if applicable, a random intercept for individuals, and Bonferroni correction for multiple comparisons.Using linear mixed-effects models, both changes between conditions and over time can be analysed.

R E SU LTS
From August 2018 to February 2021, 25 patients were enrolled in the trial.Median age at inclusion was 9 years 8 months (interquartile range [IQR] = 7 years-13 years 8 months), 15 patients were male, 23 patients were classified in GMFCS level I and two in GMFCS level II.One patient did not start the trial and six patients stopped before completing the trial, leading to seven dropouts in total.In three cases, the reason for dropout was related to FES therapy (dropout nos. 1, 3, and 6).Table 1 shows the patient characteristics, including the reasons for dropout, and the GAS results.The CONSORT flow diagram can be found in Appendix S1.

ICF: activities and participation
Seventeen of 25 patients achieved their first goal regarding quality of walking.There was no statistically significant difference in the proportion of goals achieved in the FES versus conventional treatment phase for GAS goal 1 (48% vs 20% achieved, p = 0.065), nor for the GAS goal 2 (36% vs 32% achieved, p = 1.00; see Figure 1 and Table S1 for the raw GAS scores).All dropouts were coded as 'goal not achieved' (worst case scenario).If only complete cases were analysed, the proportions of goal achievement were 67% (FES) versus 28% (conventional) for GAS goal 1 (p = 0.065), and 50% (FES) versus 44% (conventional) for GAS goal 2 (p = 1.00).No significant differences were found regarding changes in the participation domain scores of the Cerebral Palsy Quality of Life questionnaire during the FES phase versus the conventional treatment phase (z = −1.334,p = 0.206; Table S2).The median of the participation domain scores ranged from 75 (FES 2, IQR = 71-87) to 76 (FES 1, IQR = 65-84) for the FES phase and from 68 to 76 (out of 100, both IQR = 61-84) for the conventional treatment phase.
For the treadmill-based gait analysis, the median individual comfortable walking speed was 1.09 m/s (IQR = 0.83-1.27).The spatiotemporal and kinematic gait parameters are shown in Tables 2 to 5. The effect of the condition (with or without FES or AFO) on the affected leg was analysed.The two participants who underwent treatment with adapted shoes and no AFO in the conventional treatment phase (patient nos. 5 and 7) only performed gait analysis with adapted shoes in the conventional treatment phase; these measurements were labelled as 'without AFO' for the group analysis.(Gait analysis without shoes is not safe on the CAREN treadmill.)Note that 'orthosis' and 'orthoses' is used as a general term encompassing both the FES device (WalkAide) and an AFO.

Physical examination
The median passive ROM (PROM) of the affected ankle was 5° to 10° with the knee in 90° flexion, and 0° to 5° with the knee extended.No significant differences were found during the conventional versus FES treatment phase for PROM with the knee in 90° flexion (z = −1.35,p = 0.19) or with the knee extended (z = −0.53,p = 0.63).No significant differences were found regarding changes in spasticity, measured as the difference between PROM and angle of catch, during the conventional versus FES treatment phase with the knee in 90° flexion (z = −1.52,p = 0.14) or with the knee extended (z = −0.51,p = 0.65) (Figure S5 and Tables S5 and S6).
The dynamometry median values of the moment (Nm/ kg) for the affected leg were 14.9 Nm/kg to 17.3 Nm/kg for the gastrocnemius, 12.1 Nm/kg to 14.6 Nm/kg for the soleus, and 6.0 Nm/kg to 7.3 Nm/kg for the dorsiflexor.No statistically significant differences were found during the conventional versus FES treatment phase for the gastrocnemius (z = −0.94,p = 0.37), soleus (z = −1.50,p = 0.14), or dorsiflexor (z = −0.33,p = 0.77) (Figure S6 and Table S7).

ICF: personal and environmental factors
According to the log files of the WalkAide device, median use of the device was 4.4 hours a day (IQR = 2.6-6.6).However, errors or discrepancies were present in some cases, as shown in Table S8.After removing these values, median usage was 4.1 hours a day (IQR = 2.6-6.3).The median step count recorded by the device was 2165 steps per day (IQR = 1530-4083, no exclusions).The median final stimulus intensity was setting 3.0 (IQR = 2.5-4.0) on the WalkAide device.
No statistically significant differences were found in the changes during the conventional versus FES treatment phase regarding patient satisfaction with the walking aid (z = −0.632,p = 0.563), nor regarding the feelings about dressing based on the Cerebral Palsy Quality of Life questionnaire (z = −1.736,p = 0.105; Table S2).Figure 2 shows the individual patient satisfaction values (see also Table S9).
Adverse events are reported in Table 6.While using FES, skin irritation at the place of the FES electrode occurred in one patient, plantar fasciitis in another patient, and an increase in the frequency of convulsions in the arm (possibly epileptic) in another patient.Furthermore, some adverse events occurred during the conventional treatment phase or without the use of FES.
We encountered four cases where FES was not acceptable or feasible (Table 1).Reasons for dropout included: (1) spasticity of the tibialis posterior muscle; (2) fear of the electrical stimulation; (3) social and emotional problems related to orthoses; and (4) a very thin leg preventing accurate electrode placement.
We explored whether patient characteristics were significantly related to successful FES treatment.Univariate

F I G U R E 1
Percentage distribution of goal achievement per goal and per study phase for all participants (a) and for complete cases (b).Abbreviations: FES, functional electrical stimulation; GAS, Goal Attainment Scale.
logistic regression variable: achievement of GAS goal 1) only showed a significant association for 'comfortable walking speed' (p = 0.04, odds ratio = 175.70,95% CI = 1.41-21 897.95;Table S10); multivariable testing showed no significant results.This result should be interpreted with caution because of the wide confidence interval.

DISCUS SION
In this study, we aimed to show that FES of the ankle dorsiflexors is an effective alternative to AFOs in daily life for some individuals with CP.We took several outcome measures into account, that is, activities, participation, body functions, and body structures.

Primary objective
The percentage of patients who achieved their first GAS goal was not significantly higher in the FES treatment phase (48%) versus the conventional treatment phase (20%), even after the exclusion of dropouts (67% vs 28%).Thus, individual goals reflecting the domain of participation were achieved in both treatment phases but more so in the FES treatment phase; however, the difference was not statistically significant.This shows that FES treatment is at least not significantly worse than AFOs.The non-significant result may have been caused by the higher-than-expected number of dropouts.This suggests that FES is not effective; however, based on the individual GAS results, we observed that FES had really positive effects for some patients.The relatively low total percentage of goal achievement (34% on average for all goals combined) might raise the question of how realistic the goals were.Previous literature reported around 60% to 100% of patients achieving 1 or more goals for different treatments. 21,22The cross-over design of our study should be kept in mind: patients formulated the goals at the start of the study, while still receiving conventional treatment.In line with expectations, few goals would subsequently be achieved in the conventional treatment phase unless other changes also occur.The most common topic of the first GAS goal was quality of walking.This led us to surmise that most patients and parents are conscious of their child's impaired gait pattern and wish their child to have a 'normal' gait pattern.

Secondary objective
In the domain of activities, we did not find significant differences between changes in the FES treatment phase versus the conventional treatment phase.In the domain of body functions and structures, we found no significant effects of FES on the ankle dorsiflexion angle (both during mid-swing and at initial contact) or ROM swing at the group level.As shown by the significant time effect for the ankle dorsiflexion angle during mid-swing, ankle dorsiflexion decreased over time, especially in the FES treatment phase.On the other hand, ROM swing was significantly larger with (and without) FES than with AFO and did not decrease over time.
We hypothesize that these findings are caused by the effect of not wearing AFOs during the FES period after a long period of wearing them.Patients who are used to wearing an AFO walk with less plantarflexion and it takes time for this effect to fade.During the FES period (12 weeks during which patients did not wear their AFO), plantarflexion at push-off could gradually increase, increasing the ROM during swing.
The increase in plantarflexion due to not wearing an AFO was higher than the increase in dorsiflexion due to FES.The fact that the ROM swing did not decrease over time, nor did the PROM at the physical examination, shows that the patients did not develop contractures.However, 12 weeks is a relatively short period for contractures to develop and a longer follow-up is necessary.
The MFC was significantly affected by both FES and AFO, but no significant differences were found in the pairwise comparisons, although the mean values when wearing an AFO were approximately 0.2 cm higher than when not wearing one (with FES 1.95 cm, without FES 1.76 cm, with AFO 1.93 cm, without AFO 1.67 cm).][25] A higher walking speed can increase MFC, but this factor was constant for each patient over the different treatment conditions and measurements.Mean MFC values were higher than those reported by Romkes et al. 26 for a comparable group: 0.86 cm (SD = 0.15) and 1.37 cm (SD = 0.15) on even and uneven ground for the affected side while barefoot.To our knowledge, until now the effect of treadmill versus overground walking on the MFC has not been reported in the literature, but it is possible that treadmill walking would increase the MFC because patients want to be more careful on a treadmill.
Furthermore, neither FES nor AFOs significantly affected the peak foot progression angle during swing, although the mean value for FES was lower (indicating more out-toeing) than the other treatment conditions.More out-toeing with FES could be explained by stimulation of the superficial peroneal nerve, as well as the deep peroneal nerve.We noticed that the effect of FES on the peroneal nerve on the foot progression angle was variable per patient and depended on the exact placement of the electrodes and stimulus intensity.In some patients, it was easier to obtain an eversion reaction by FES than a dorsiflexion reaction; the exact location and pressure on the electrodes make a substantial difference in motor response.The motor response is mostly a mix of dorsiflexion and eversion.An eversion reaction is helpful for some patients because some have the tendency towards in-toeing (e.g. because of tibialis posterior spasticity), increasing the risk of tripping over their legs.For one patient, the foot progression values of the FES measurements were excluded because they showed a peculiar swabbing pattern in the foot progression angle, with both a negative and positive peak value.Therefore, analysis of one peak value was not suitable (see Figure S7 for the details).
Analysis of the physical examination parameters did not show significant differences in changes during the conventional versus FES treatment phase.For the PROM, it is a positive finding that no clear deterioration occurred during the FES treatment phase.The median PROM with the knee extended was higher at the FES measurement 2 than at the conventional measurement 2. For spasticity with the knee flexed, fewer patients presented without an angle of catch at the FES measurement 2 (Figure S7).However, a longer follow-up is needed to gather more information on the effect of FES instead of AFO on PROM.Regarding muscle strength, we hypothesized that FES

Event Frequency
Skin irritation at the place of the FES electrodes 1

Plantar fasciitis 1
Increase in frequency of convulsions in the arm, possibly epileptic 1 Deterioration of the ankle ROM during the conventional treatment phase 1 Skin irritation because of the plasters of the activity monitors 3 Injured toe because the patient hit a peg on a camping site while walking on flip-flops without a walking aid 1 Sudden stop of the treadmill during walking for the gait analysis 1
would increase dorsiflexor strength because a previous study reported an increase in muscle volume of the tibialis anterior and gastrocnemius. 27However, the median values for dorsiflexor strength were slightly higher in the conventional treatment phase; neither the strength of the gastrocnemius nor soleus was higher in the FES treatment phase at the group level.Possible explanations could be sought in the duration of FES treatment and selectivity: some patients had difficulties performing a specific movement without contracting other muscles.Also, as reported in our protocol, the smallest detectable difference was 9% to 30%. 28

Tertiary objective
The acceptability of FES was reasonable: 12 of 22 patients who tried FES chose to continue FES therapy instead of using an AFO; three patients dropped out and thus never tried FES.However, FES might not be feasible for all patients.FES was unsuccessful for some patients because of their small leg, fear of electrical stimulation, social and emotional problems regarding orthoses (this also applied to AFOs), and spasticity of the tibialis posterior muscle.Also, early heel rise during stance is mostly not inhibited by FES.Furthermore, side effects can complicate the use of FES: three adverse events possibly related to FES treatment were reported.Skin irritation at the place of the FES electrodes could be attributed to the electrode material (including the gel used for conduction) or to the electrical current.The patient reporting this side effect used a high stimulus intensity, which could have contributed to the skin irritation; this subdued after a rest period.The same patient developed pain under the foot during FES treatment, which was diagnosed as plantar fasciitis.Factors that could have contributed to plantar fasciitis were stretch of the Achilles tendon because of ankle dorsiflexion caused by FES stimulation (FES caused 15° more dorsiflexion than without FES in this patient), and a more intense heel-floor contact at initial contact than this patient was used to with an AFO.The condition improved with gel insoles and less intense FES (fewer hours of use by alternating with an AFO and using a lower stimulus intensity).Another patient showed an increase in the frequency of (possibly) epileptic convulsions of the arm while using FES (no electroencephalogram during an attack was available).The exact cause is unknown; however, attack frequency decreased after the medication dose was increased.Although patient satisfaction did not show any significant difference between the different orthoses, it was more variable for FES than for AFOs: some patients were really happy and satisfied with FES, while some were absolutely not.
Identifying patients for whom FES treatment is likely to be successful (based on goal achievement) is important.Besides the statistically significant relationship with walking speed, age is also important.We advise selecting patients aged 6 years and older for FES therapy because of device size, stimulus intensity, and check-up frequency.In this study, FES treatment was successful in one of the two patients classified in GMFCS level II.We think that the chance of successful FES treatment is higher in patients with less severe gait deviations and who are more physically active because deterioration of the ankle ROM is probably less likely.It is important to realize that discrepancies sometimes exist between the effects on the ICF domains.For example, some patients achieved their GAS goal even if their ankle dorsiflexion did not improve, and vice versa.
Cost analysis, another important aspect for clinical implementation, will be reported in a separate publication.

Limitations
This study has a few limitations.First, although we managed to include the number of patients needed based on our power calculation, we had a higher dropout rate than anticipated.This potentially decreases the power of our study.Therefore, we also present the percentages of individuals achieving GAS goals without including the dropouts.Because of practical reasons, including the COVID-19 pandemic, we were not able to extend the trial further.Second, also because of practical reasons, the physical examination could not always be performed by the same physiotherapist.This could have led to higher variability between measurements of the ROM and muscle strength.Lastly, measurement of the MFC should be mentioned.This measurement is based on the second metatarsal head, while the lowest point of the foot changes depending on the ankle angle.By measuring the second metatarsal head marker, this change in the lowest point is not taken into account.However, our method of using a toe marker is a good approximation because toe drag is a common cause of tripping in this population with CP.

Clinical recommendations
A point of interest is the set-up of the FES device.In our opinion, it is essential that both the placement of the electrodes on the leg and the timing of the stimulus are well controlled because these factors directly influence the effect of FES.As described in our protocol, 20 we used electromyography measurements to optimize the timing of the FES stimulus.In all cases, we had to change the timing of the FES stimulus to occur earlier than the standard device settings.Otherwise, the stimulation would not have started before mid-swing, which is the most critical time point for ankle dorsiflexion and foot clearance.The stimulus also lasted too long after the initial contact.To make these adaptations, we needed the support of a specialist retailer to have more options to adapt the timing of the stimulus.Electromyography measurements, on which the FES stimulus is visible as an artefact, are more precise than using auditory feedback for stimulus timing.We strongly recommend setting up the FES device in a gait laboratory.Furthermore, we strongly recommend an FES testing period before purchasing an FES device for the patient, so that device suitability and the personal preference of the patient can be evaluated.This approach will increase the percentage of patients using the device daily and increase the likelihood of achieving individual treatment goals.

Future directions
We recommend performing a long-term follow-up study, first to see if deterioration of the ankle ROM occurs when using FES for more than 3 months.Additionally, outcome measures across all the ICF domains should be included, especially the participation domain.Selection criteria should be a point of attention because our study cannot fully answer this question.

Conclusion
Although no statistically significant effect of FES was found on the participation domain or ankle dorsiflexion at the group level, FES had positive effects on individual goals for some patients with CP.Thirteen patients chose to continue FES treatment.Based on our results that FES is at least not significantly worse than AFOs, FES may be a suitable alternative to conventional treatments (AFO and adapted shoes) in a selected group of patients.We therefore recommend an FES testing period to evaluate the suitability of a patient and, in case of continuation of FES therapy, a long-term followup, including gait analysis and regular physical examination to monitor, among others, the ankle ROM.

AC K NO W L E D GE M E N T S
We thank the participating children and their parents for their time and effort.We thank Orthomedico, the company that made two of the six FES devices available for the trial as an unrestricted gift.This study is funded by HandicapNL (formerly Revalidatiefonds, project no.R201605614), in collaboration with the Cornelia Foundation.Furthermore, we thank Y. J. M. Janssen-Potten, E. Claassen, R. Hovenier, D. Heijnen, and E. Stoop for their help with the measurements.The Phelps Foundation supported the (3-year) follow-up of the FES study and provided the devices for the participating patients, who continued using FES.

F I G U R E 2
Patient satisfaction with the walking aid (individual data; median [wider red bars], interquartile range [narrower red bars]).Abbreviations: FES, functional electrical stimulation; VAS, visual analogue scale.T A B L E 6 Adverse events recorded during the trial.

Patient and treatment sequence Age, years Sex GMFCS level MRI pattern a Age at MRI, years:months MRI score b BoNT-A
Patient characteristics and Goal Attainment Scale results.
T A B L E 1 Kinematics and spatiotemporal gait parameters: comfortable walking speed.
Kinematics and spatiotemporal gait parameters: measurement no. 1 of spatiotemporal gait parameters for each phase (weeks 0 and 18) based on comfortable walking speed.
T A B L E 4Abbreviations: AFO, ankle foot orthosis; FES, functional electrical stimulation.T A B L E 5 Kinematics and spatiotemporal gait parameters: measurement no. 2 of the spatiotemporal gait parameters for each phase (weeks 12 and 30) based on comfortable walking speed.Abbreviations: AFO, ankle foot orthosis; FES, functional electrical stimulation.