Physical therapy approaches are often advocated as the first treatment choices for reducing spasticity and are deemed as the foundation upon which other therapies are built (Merritt 1981; Kirshblum 1999; Rosche 2002). Despite these contentions, there is a relative paucity of literature addressing the efficacy of either the passive techniques noted in the previous section or approaches involving active movement in individuals with SCI. In practice, active movement approaches may be conducted using a variety of exercise forms that may also provide benefits beyond spasticity reduction (e.g., strength, endurance, gait re-training). The studies meeting the criteria for the present review involve exercises performed in a therapeutic pool (i.e., hydrotherapy) (e.g., Kesiktas et al. 2004), those associated with FES-assisted cycling (e.g., Krause et al. 2008), locomotor training programs, whether assisted by FES (e.g., Granat et al. 1993; Kapadia et al. 2014; Mirbagheri et al. 2002), or a FES-powered orthosis (e.g., Mirbagheri et al. 2015; Thoumie et al. 1995).
Robot Assisted Exercise (involving voluntary or electrically assisted movement)
In Fang et al.’s (2015) crossover RCT (n=10) noted in the passive movement section previously, one of the conditions involved electrical stimulation-induced contractions for eight min, once/week over three weeks in addition to high speed cyclic passive exercise (50 cycles/min) or low speed cyclic passive exercise (20 cycles/min). Although the focus of that study was on the effects of robot passive-assisted exercise, a key finding related to ES-assisted muscle contractions was that H reflex amplitudes were reduced after passive exercise at both speeds but not after repeated ES-elicited contractions. This suggests reflex excitability is more affected by passive movement (i.e., stretching) than active muscle contraction. Also, electrical stimulation-induced contractions only had an effect in reducing total resistance during slow (p<0.05) but not fast cyclic stretching, whereas there was actually an ES-mediated increased resistance (p<0.05). Isometric torque decreased significantly after 8 minutes of electrical stimulation-induced contractions and the reduction persisted up to 20 minutes (p<0.05), thereby indicating fatigue.
Mirbagheri et al. (2002) previously reported on the use of a custom-made device producing sinusoidal ankle movements as well as mathematical modelling to assess variations in intrinsic (musculotendinous) and reflex contributions to spasticity calculated from measured ankle joint torque due to perturbations. This was employed as part of an RCT (n=46) reprted across two separate publications (Mirbagheri et al. 2015; 2013) examining the effects of robotic-assisted step training using Lokomat vs no treatment on ankle spasticity. The treatment group (n=23) received 1-hour sessions of robotic-assisted step training, three x/week for up to 45 minutes, for four weeks with measures completed prior to training and after 1, 2, and 4 weeks of robotic-assisted step training. The key finding was that reflex stiffness and intrinsic stiffness parameters of spasticity all showed significant decreases (p<0.05) whereas the control group did not show any change (p>0.05) for any of the parameters calaculated. Subjects were stratified into groups based on similar reflex or intrinsic parameter recovery patterns for both robotic-assisted step training and control groups. Three classes were identified based on greater baseline values for each of the reflex parameters and two classes were identified for the intrinsic parameters. Essentially, greater reductions in spasticity were seen in those with higher baseline levels of either reflex (r2=0.94, p<0.0001) or intrinsic (r2=0.84, p=0.01), although these changes were only apparent in the treatment group as there were no significant changes with the “no treatment” control condition.
In summary, there is some evidence that robot-assisted exercise (whether including voluntary or ES- assistance) appears to decrease all components of spasticity in the spinal cord injured individual: isometric torque, reflex and intrinsic stiffness. However, there is certainly more research required to identify more specific information associated with both the treatment and resulting spasticity parameters.
Exoskeleton Walking Device
Kressler et al. (2014) completed a case series study (n=3) of over-ground bionic ambulation training using an exoskeleton (Ekso). The study showed that there were no changes in Spinal Cord Assessment Tool for Spastic Reflex (SCATS), Spinal reflex excitability, recordings of electromyographic activity (EMG), or electroencephalography measures.
Del-Ama et al. (2014) conducted a case series study (n=3) to determine the effects of electrical-muscle stimulation -induced hybrid gait training (interventions with a hybrid bilateral exoskeleton for 4 days). Study subjects demonstrated improvements in spasticity as marked by differences in the AS (-0.2±0.4) and Penn Spasm Frequency Scale (PSFS) (-0.4±0.5) after the intervention.
With a larger cohort of individuals (N=45), Juszczak et al (2018) reported significantly reduced self-reported spasticity (p<0.001) after 8 weeks of graduated training over 3-4 sessions with the Indego Powered Exoskeleton. The importance of this finding is somewhat muted when it is noted that the self-reported spasticity assessment (i.e. 0-10 numerical rating) has not been psychometrically studied for SCI spasticity. Conversely, the MAS has been validated for use in assessment SCI spasticity and in this trial, the majority of participants (62.2%; n=28) did not register a change in spasticity with MAS. Of the remaining participants, reduced or increased spasticity, as measured by MAS, was detected in 26.7% (n=12) and 11.1% (n=5) after Indego training.
Functional Electrical Stimulation
Kapadia et al. (2014) completed an RCT (n=34) to assess FES walking and mobility, with spasticity (i.e., MAS, pendulum test) as primary outcomes. The intervention group (n=17) received FES on their quadriceps, hamstrings, dorsiflexors, and plantarflexors while performing ambulation exercises on a body weight supported treadmill. The control group (n=17) received conventional resistance and aerobic training. Both groups received 45-minute sessions, threex/week for 16 weeks with assessments at baseline, 16 weeks, 6 months, and 12 months post intervention. In general, there were no significant differences over time or between groups for the majority of the MAS and pendulum test. However, MAS scores significantly worsened over time (i.e., more spasticity) for both groups in the right quadriceps (p=0.015).
As part of a larger RCT, Manella et al. (2013) conducted an analysis (n=18) of high vs low clonus participants (i.e., high=at least four beats of clonus during drop test) in order to study the effect of various forms of locomotor training on ankle clonus and quadriceps muscle spasm. Study subjects were randomized to receive one of four body-weight supported locomotor training interventions, although sub-analyses were not conducted for each of these related to the spasticty measures. Participants underwent locomotor training for 1 hour per day, 5 days per week for 12 weeks. Only high clonus subjects showed significantly decreased extensor spasm duration (mean difference=-13.20 s, p=0.05), significantly increased overground walking speed (mean difference=0.06m/s, p<0.01) and non-significant decreases in clonus duration observed from the drop test (mean difference=-3.99s, p=0.06) and plantar flexion reflex threshold angle (mean difference=-5.82°, p=0.09). The low clonus group did not show any significant changes in any of theoutcome measures. The key finding was that walking speed improvements (no matter the modality of locomotor training) were strongly correlated with reductions in spasticity (i.e., reduced clonus and spasms).
Ralston et al. (2013) completed a crossover RCT (n=14) to study the effect of FES-assisted cycling. Study subjects were randomized to receive 2 weeks of either FES cycling for 30-45 minutes per day, 4 days per week combined with usual care vs only usual care, with a 1-week washout between the two conditions. FES was applied to the quadriceps, hamstrings and gluteals of each leg. Usual care consisted of standard inpatient physical and occupational therapies including treatment for poor strength, restricted joint mobility, limited fitness, reduced dexterity, pain and functional skills training. Urine output, as the primary measure, as well as lower limb swelling, spasticity (Ashworth) and individuals’ perception of treatment effect (patient reported impact of spasticity measure- and Global impression of change scale-GICS) were measured. No measures achieved statistical significance, perhaps related to the relative brevity of the intervention. However, all means were improved in favour of an effect of FES-cycling. Urine output increased by 82 mL with FES cycling compared to the control group. Lower limb swelling (-0.1 cm), Ashworth scores (-1.9 points on a 32-point scale obtained by summing four muscle groups bilaterally), and individual reported impact of spasticity measure reports (-5 points) were all reduced with FES. All 12 study subjects reported improvements with FES cycling on the GICS with a median improvement of three points. During the study, two individuals reported adverse events: an increase in spasticity and one reported a bowel accident.
Kuhn et al. (2014) completed a pre-post FES cycling study (n=30) involving training sessions of 20 minutes, two times per week for 4 weeks. MAS scores had significant improvements (p=0.002) in hip abduction (p=0.016), knee flexion (p=0.003), knee extension (p=0.001), and dorsal extension (p=0.001). In contrast, no significant changes were seen on the MAS and four-point spasms scale in a small pre-post trial (n=5) of FES cycling conducted by Mazzoleni et al. (2013). This study involved training for 15-30 minutes per day, 3 days per week for 20 weeks with the pedaling time increasing as individuals progressed over time.
Sadowsky et al. (2013) reported an increase in muscle strength but not spasticity levels associated with a retrospective, controlled cohort trial (n=45) of lower extremity FES cycling. Participants were non-randomly grouped into intervention (n=25) and control (n=20) groups.
Reichenfelser et al. (2012) completed a prospective controlled trial (n=36) to study the effects of FES cycling over a period of 2 months. The treatment group was divided into Group A (n=13; non-spastic group, MAS score<1) and Group B (n=13; spastic group, MAS score>=1). The study subjects performed training sessions three times per week over an average timespan of 2 months. The control group (n=13; Group C, able=bodied individuals) performed the training session twice. Active power output (30 and 60 rpm), and spasticity (decrease within the cycling session as measured by passive resistance over the cycle) were assessed as part of a customized, commercial tri-cycle ergometer system. Group A treatment subjects showed the highest decrease in passive resistance overall (i.e., within a session and greater as time progressed) with Groups B treatment subjects and Group C control subjects showing lesser and comparable decreased passive resistance. However, the difference between these two groups increased at higher rpm and passive resistance decreased more in Group C control subjects across the various rpm. In addition, there was a monthly mean increase in power output of 4.4 W at 30 rpm and 18.2 W at 60 rpm.
Krause et al. (2008) used a randomized, crossover study design in which five individuals with complete AIS A SCI underwent 1) FES cycling and 2) passive movement by a motor-assisted cycling ergometer. For both of the interventions, the legs were moved for the same period of time at the same velocity and frequency. The study demonstrated that FES (i.e., active muscle contractions) was significantly more effective than passive movements at reducing spastic muscle tone in individuals with complete SCI, although even passive movement resulted in spasticity reductions. This was indicated by a greater reduction in the MAS for FES versus passive movement (p<0.001 versus p<0.05, respectively), also with the Pendulum Test (p=0.01). Further research may be useful in determining precise stimulation patterns to use for FES-cycling as Mela et al. (2001) have noted that specific stimulation frequency parameters may influence spastic reactions variably which suggests careful selection of stimulation parameters so as to optimize the delivery of FES as a clinical tool to reduce spasticity.
Mirbagheri et al. (2002) calculated reflex and intrinsic stiffness of the ankle, as described earlier, as a means of assessing spasticity prior to and following a FES-assisted walking training program. This program involved four individuals with longstanding AIS C or D SCI who underwent locomotor training for a minimum of 16 months. Both reflex and intrinsic stiffness were reduced following FES-assisted walking. Conversely, an individual with SCI, who was not using FES-assisted walking demonstrated no reduction in spasticity. Although the MAS was noted as an outcome measure in the methods, the authors failed to report the final results associated with this clinical measure.
In a similar trial of FES-assisted walking in people with longstanding SCI (Frankel C or D), Granat et al. (1993) also found reductions in spasticity as assessed by a pendulum drop test but did not show any change pre-and post-training when considering AS scores. Granat et al. (1993) performed the final spasticity assessment 24 hours after the final FES-assisted walking session; thereby ensuring the final outcomes would not be unduly influenced by the short-term effects of muscle stimulation.
Thoumie et al. (1995) examined the effects of a FES-assisted Reciprocating Gait Orthosis II on spasticity following a long-term program (i.e., three-13 months) of gait training. No group results (n=21) were reported for spasticity although it appeared that no systematic effects were obtained on a customized self-report version of the AS. Some subjects (n=7) reported decreases in spasticity in the short-term, while others reported increased spasticity (n=4).
Effects of Standing
Sadeghi et al. (2016) completed a prospective controlled trial (n=10) where study subjects underwent two different standing training interventions (dynamic and static). These interventions were given for 20 minutes and separated by 1 week. Outcome measures including the MAS, VAS, andEMG were completed immediately before standing training, at 5 minutes after, and at 1 hour after training. The study showed non-significant decreases in spasticity for both dynamic and static standing trials for individuals with SCI as measured by MAS, VAS and EMG.
Body Weight Supported Treadmill Training (BWSTT) verses Tilt Table Standing (TTS)
Adams et al. (2011) completed a small crossover RCT (n=7) to determine the effects of 12 sessions of body-weight supported treadmill training and tilt-table standing on clinically assessed and self-reported spasticity, MAS, soleus H/M ratio, SCATS, SCI-SET, and the PSFS. The study showed that extensor spasms decreased following tilt-table standing but not after body-weight supported treadmill training (electrical stimulation=0.68) and there was a greater reduction in passive resistance to movement (Ashworth; ES=0.69) and flexor spasms (electrical stimulation=0.57) after body-weight supported treadmill training compared to tilt-table standing . Following body-weight supported treadmill training, there was a greater reduction in the H/M ratio compared to tilt-table standing (electrical stimulation=0.50). Extensor spasms were reduced after tilt-table standing (electrical stimulation=0.95) with a greater reduction after tilt-table standing than body-weight supported treadmill training (electrical stimulation=0.79); however, flexor spasms were observed to be reduced to a greater extent after body-weight supported treadmill training as compared to tilt-table standing (electrical stimulation=0.79). There were no observed changes in the H/M ratio, SCI-SET or PSFS scores following either body-weight supported treadmill training or tilt-table standing.
Boutilier et al. (2012) completed a pre-post study (n=8) on dynamic standing using the Segway device. Study subjects underwent a 4-week dynamic standing program using a Segway, three times a week, for 30-minute sessions. Outcomes were measured by the MAS and the SCI-SET. The study showed that in individual muscle MAS scores there was a decrease after the intervention in at least two out of three muscle groups for every subject. There was a statistically significant reduction in MAS immediately from pre-to post training (p<0.001) though there was no significant improvement over time. In spasticity evaluations using the SCI-SET, an improvement from -0.91±0.30 initial, to -0.63±0.24 midway, to -0.57±0.24 final scores were observed. These improved SCI-SET scores were not statistically significant. However, there was a statistically significant reduction in pain over time (p<0.05) for the three visits (initial 42.75±8.49, midway 40.88±10.10, ﬁnal 32.88±7.17). There was no significant difference between initial and final visits for fatique, though mean fatigue scores did improve (mean=4.2±0.42 to 3.7±0.54).
Cortes et al. (2013) completed a pre-post study (n=10) of wrist spasticity using a robot training intervention with a primary focus on enhancing motor performance and neurorecovery. Study subjects received a 6-week wrist-robot training intervention from the InMotion 3.0 Wrist robot, for 1 hour/day, 3 day/week, for a total of 18 training sessions. After 6 weeks of robotic training there was no significant changes in upper limb spasticity as assessed by the MAS in the right trained arm (p=0.43) or left untrained arm following training (p=0.34).
Kesiktas et al. (2004) employed an experimental non-RCT design to test the effectiveness of a water-based exercise (i.e., hydrotherapy) program in reducing spasticity in a group of individuals (n=10) with complete and incomplete paraplegia and tetraplegia. Subjects were matched within a treatment group (i.e., hydrotherapy + conventional rehabilitation) versus a control group (conventional rehabilitation only) on the basis of age, gender, time post-injury, injury level and severity, spasticity (Ashworth) and function (Functional Independence Measure FIM). This study produced consistent results across all spasticity-related measures with spasticity reductions evident following the 10-week hydrotherapy treatment program for both AS scores and the Penn Spasm Severity scores. The control group also showed significant spasticity reductions relative to baseline with these measures but not to the same degree. In addition to these measures, dosages of oral baclofen were significantly reduced for those receiving hydrotherapy versus conventional rehabilitation only (i.e., >50%) and the hydrotherapy treated group made much greater FIM gains than did the control group. These latter results may reflect the deleterious effect that high baclofen doses can have on motor and cognitive function and the benefits of reduced spasticity on motor function. Kesiktas et al. (2004) did not indicate how soon after the final intervention the measures were taken so there is no indication of how long the beneficial effect might have been maintained.
Although unblinded study participants perceived that strength training of partially paralyzed muscles improved their strength and function, the clinically measured effect on spasticity (and strength) was inconclusive with respect to clinically meaningful changes as measured by the Ashworth (and maximal voluntary isometric strength). Importantly though, strength training did not have a deleterious effect on spasticity (or strength).
Combination therapies are intended to leverage simultaneous targeting of more than a single type of neural circuitry during rehabilitation training. Martinez et al (2018; level 2 evidence), in a small level 2 crossover trial, were not able to show significant differences between multimodal training and robotic treadmill training for spasticity as assessed with the SCI-SET (and lower extremity strength and reflexes and ambulation and pain). The control intervention consisted of 48 sessions of robotic treadmill training at increasing speeds and decreasing body-weight support to reach a predefined gait pattern and was compared to the control intervention augmented with simultaneous balance and skilled upper extremity exercises (i.e. multimodal intervention). Participants were assigned to start with treadmill only or multimodal training and after a washout period of 6 weeks, continued with the comparison intervention. Two of 9 participants actually showed an increase in spasticity after multimodal training while 4 of 9 showed a decrease after the control intervention. Three and 5 showed no change, respectively. These less than clear results may have been the result of only 43% (9/21) or participants completing both arms of the trial.
Physical and electroceutic neuromodulatory methods are sought as an alternative to pharmacological control of spasticity due to the avoidance of deleterious side effects accompanying drug treatments. Estes et al 2017 (level 2 RCT) systematically tested 4 non-pharmacological approaches against a sham-control to achieve spasticity reduction in the lower extremeties as quantified by the pendulum test. Stretching (hip/knee/ankle flexors/extensors), cyclic passive movement ( through a treadmill-mounted robotic gait orthosis), transcutaneous spinal cord stimulation ( using biphasis stimulation at T11/12 and umbilicus), and transcranial direct current stimulation (with anode 1cm anterior to the vertex and cathode over inion) while the sham-control consisted of a brief ramp-up/-down of knee/ankle stimulation with the leg extended and participant in the reclined position. Each participant was randomized to receive a different order of the 4 interventions and sham (all single sessions only), after 48 hours washout between sessions. Continuous passive motion and transcutaneous spinal cord stimulation were shown to be viable non-pharmacological treatments for induction of prolonged periods (45 minutes) of spasticity reduction. For immediate and only short-term spasticity reduction, stretching was the most effective compared to sham. The sham-control condition confirmed that spasticity is problematic with increased immobility. Further investigations of dosing and timing would be necessary to maximize efficacy for these alternative anti-spasticity treatments.
Controlling for confounding variations in overall body condition is a factor to be considered when assessing efficacy of novel interventions such as pharmacological and cellular therapies. To understand the effects of various rehabilitation therapies without the addition of novel interventions, Gant et al (2018; level 4 evidence)) utilized a 12-week training program for participants with chronic thoracic level motor-complete participants, to deliver body-weight-supported treatmill training for locomotion, circuit resistance training for upper body conditioning, FES for activation of sublesional muscles and wheelchair skills training for overall mobility. Without the addition of any novel intervention, upper extremity strength improved for all 8 participants and some also experienced improved function that was likely a result of the improved strength. However, no improvements led to a change in neurological function and changes to pain and spasticity were highly variable between participants. Obviously, a larger study is required to validate these findings in a generalizable way but the current small trial is an indication of the importance of balancing, standardizing and documenting rehabilitation therapies whether additional novel interventions are included when attributing functional recovery to the intervention under study. In this study, spasticity (assessed with SCATS) was not consistently influenced by this 12-week multi-modal training program.
Mazzoleri et al (2017; level 4 evidence) investigated the effect of integrated gait training (20 sessions of functional electrical stimulation cycling followed by 20 sessions of overground robotic exoskeleton training) on spasticity (and patient-robot interaction). Spasticity was assessed with the MAS, numerical rating scale and PSFS. Spasticity was significantly reduced after the first phase of FES based training (MAS; p<0.05) with a further decrease after the second phase of robotic exoskeleton training (MAS and NRS; p<0.05 for both). The effects on spasticity also likely contributed to improved gait parameters after overground robot-assisted gait training (significant increase of standing time and number of steps). Although the anti-spasticity effect reported here conflicts with results of other combination therapy studies (Martinez et al 2019, Estes et al 2017, Gant et al 2018), the particular combination and intensity of integrated gait training may be key to the differing results.
There is level 1b evidence from two RCTs (Fang et al. 2015; Mirbagheri et al. 2015) robot-assisted exercise appears to decrease all components of spasticity (isometric torque, reflex and intrinsic stiffness).
Functional Electrical Stimulation
There is level 4 evidence that single bouts of FES-assisted cycling ergometry, with a single level 2 study also showing that similar passive cycling movements are effective in reducing spasticity over the short-term although FES is more effective than passive movement.
There is level 1 evidence from 1 study, with conflicting evidence across two level 4 evidence studies, that show FES cycling decreases spasticity over the long-term.
There is level 4 evidence from three studies that a program of FES-assisted walking acts to reduce ankle spasticity in the short-term (i.e., £ 24 hours), however, a level 2 study showed no reduction across several lower limb muscles when considering an overall sustained effect.
There is no evidence to describe the optimal length and time course of FES-assisted walking for reducing spasticity.
Effects of Standing
There is level 2 evidence (from one prospective controlled trial; Sadeghi et al. 2016) that dynamic and static standing training does not reduce spasticity.
Body Weight Support Treadmill Training versus Tilt Table Standing
There is level 2 evidence (from one RCT; Manella & Field-Fote 2013) that electrical stimulation treadmill training and LOKOMAT robotic-assisted training decreases ankle clonus.
There is level 4 evidence (from one pre-post study; Adams et al. 2011) that use of tilt-table standing decreases extensor spasms and body-weight support treadmill training results in a reduction in passive resistance to movement and flexor spasms.
There was level 4 evidence (from one pre-post study; Boutilier et al. 2012) that showed use of a Segway device for dynamic standing results in a reduction of spasticity.
There is conflicting level 4 evidence (from two case series; Kressler et al. 2014; Del-Ama et al. 2014) that use of an exoskeleton walking device results in a reduction in spasticity.
There is level 4 evidence (from one pre-post study; Cortes et al. 2013) that use of robotic training of the wrist does not improve upper limb spasticity.
There is level 4 evidence (from one pre-post study; Kesiktas et al. 2004) that hydrotherapy is not more effective in producing a short-term reduction in spasticity than conventional rehabilitation alone.
Resistance training is not deleterious but does not decrease spasticity as evidenced by one level 1b RCT (Bye et al 2017).
There is level 2 evidence (from 2 RCTs; Martinez et al 2018, Estes et al 2017, further supported by 1 small level 4 pre-post study by Gant et al 2018)) that combination therapies do not consistently reduce spasticity. This is slightly challenged by level 4 evidence (small pre-post, Mazzoleri et al 2017) that FES cycling followed by robotic exoskeleton training may reduce spasticity.
Active exercise interventions such as hydrotherapy, FES-assisted cycling and walking and robot-assisted exercise (including specific exercises combined) may produce short-term reductions in spasticity.