Post-stroke spasticity and motor function · Randomized trial · 2026
Does adding shockwave to rehab ease ankle stiffness after a stroke?
In this trial, adding six sessions of radial shockwave to intensive stroke rehab eased calf stiffness more than rehab alone, and the ankle could be moved about 10 degrees further, vs about 3. Both results were similar at 4 and 10 pulses a second. Leg movement scores changed little in any group.
Helped more than rehabilitation alone- adults with ankle stiffness after a stroke
- 45
- sessions over 3 weeks
- 6
- weeks: results measured at the end of treatment
- 3
Good to know: Small trial (45 people). Everyone had four hours of rehab a day; results were checked at the end of the 3-week course.
Ke M, Zhou P, Li D, et al. Journal of Neuroengineering and Rehabilitation, 2026 · Read the paper ↗
How this study worked
Conceptual animation: a radial wave spreads from the applicator and weakens with distance. Not to scale. Tap the field to send a pulse.
What happened
How far the ankle could be moved
Change in passive ankle range, in degrees · start to end of the 3-week course
Show the numbers
| Group | People | Change (degrees) |
|---|---|---|
| Shockwave 4 Hz + rehab | 14 | +9.5 |
| Shockwave 10 Hz + rehab | 14 | +9.5 |
| Rehab alone | 14 | +2.86 |
The treatment
Shockwave 4 Hz + rehab
Radial, twice a week for 3 weeks
- Sessions
- 6, two a week
- Pulses
- 2,000 per session · aimed at two spots on the calf muscle
- Strength
- 2bar · 4 per second
- Numbing
- Not mentioned
- Device
- HMJYS-A
- Everyone also got
- Intensive rehabilitation, 4 hours a day, 6 days a week
Shockwave 10 Hz + rehab
Radial, twice a week for 3 weeks
- Sessions
- 6, two a week
- Pulses
- 2,000 per session · aimed at two spots on the calf muscle
- Strength
- 2bar · 10 per second
- Numbing
- Not mentioned
- Device
- HMJYS-A
- Everyone also got
- Intensive rehabilitation, 4 hours a day, 6 days a week
Compared with: The same intensive rehabilitation without shockwave
Settings are what each study used, not a prescription.
Who took part
45 adults aged 30–75 with calf stiffness after a first stroke, at one rehabilitation centre in Shenzhen, China; 42 completed all checks.
hollow squares: no final score
Side effects
Five people felt mild pain during treatment that eased soon after resting. No serious or lasting problems.
All the numbers and the full appraisal
Every reported result
The results, in numbers.
Adults aged 30–75 years with first stroke and ankle plantar-flexor spasticity, at one Shenzhen rehabilitation department; 45 randomized and 42 completed all assessments, 14 per group.
One of 15 randomized participants was lost to follow-up; all groups received conventional rehabilitation.
One of 15 randomized participants was lost to follow-up; analyses used available data.
One of 15 randomized participants was lost to follow-up; analyses used available data.
Modified Ashworth Scale (MAS)1 reported outcome
Modified Ashworth Scale (MAS)
Modified Ashworth Scale (six coded levels: 0–5) ordinal scale category
After a single session, both ESWT groups improved from baseline; between the 4 and 10 Hz groups, no significant difference was reported. After three weeks, MAS change favored both ESWT groups over control (adjusted p=0.008 for 4 Hz vs control; p=0.025 for 10 Hz vs control); there was no 4 vs 10 Hz difference (>0.999). MAS is subjective, and grade I was the most common baseline grade (19 of 42).
Analysis details and source
Between-group comparisons of change; nonsignificance between frequencies does not show equivalence.
Ankle passive range of motion (PROM)1 reported outcome
Ankle passive range of motion (PROM)
Ankle PROM degrees as device-recorded; article does not state unit in result table
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Conventional rehabilitation control | 14 | — | 2.86 (SD 2.18) |
| 4 Hz radial ESWT plus rehabilitation | 14 | — | 9.5 (SD 7.65) |
| 10 Hz radial ESWT plus rehabilitation | 14 | — | 9.5 (SD 5.6) |
Values are group summaries identified in the caption; SD denotes standard deviation. CI is a confidence interval; IQR denotes the reported quartiles. A dash means no value was extracted for that cell.
At three weeks, PROM differed across groups; adjusted pairwise p=0.010 (4 Hz vs control), 0.002 (10 Hz vs control), and >0.999 (4 Hz vs 10 Hz).
Kruskal–Wallis with Dunn/Bonferroni pairwise tests
Mean (SD) change was 2.86 (2.18) in control, 9.50 (7.65) at 4 Hz and 9.50 (5.60) at 10 Hz. Both ESWT groups differed from control (adjusted p=0.010 and 0.002); 4 Hz and 10 Hz did not differ (>0.999).
Analysis details and source
Source reports change summaries; active-frequency nonsignificance is not equivalence.
Lateral gastrocnemius dynamic stiffness (LGS)3 reported outcomes
Lateral gastrocnemius dynamic stiffness (LGS)
MyotonPRO LGS N/m
After the first session, change was −24.33 (12.27) N/m at 4 Hz and −14.73 (8.53) at 10 Hz; between-frequency test p=0.024, Cohen’s d=−0.909. After three weeks, change was −37.22 (16.22) vs −16.24 (11.13) vs −2.67 (3.91) in 4 Hz, 10 Hz and control; overall p<0.001 and pairwise adjusted p<0.001 for 4 Hz vs control, 0.019 for 10 Hz vs control and 0.049 for 4 vs 10 Hz.
Analysis details and source
LGS is one objective muscle measure; authors interpret lower stiffness as reduced tone. Different assessment windows must not be conflated.
Lateral gastrocnemius dynamic stiffness (LGS)
MyotonPRO LGS N/m
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Conventional rehabilitation control | 14 | — | -2.67 (SD 3.91) |
| 4 Hz radial ESWT plus rehabilitation | 14 | — | -37.22 (SD 16.22) |
| 10 Hz radial ESWT plus rehabilitation | 14 | — | -16.24 (SD 11.13) |
Values are group summaries identified in the caption; SD denotes standard deviation. CI is a confidence interval; IQR denotes the reported quartiles. A dash means no value was extracted for that cell.
Three-week change across control, 4 Hz and 10 Hz ESWT
Kruskal–Wallis omnibus between-group test
Changes were −2.67 (3.91), −37.22 (16.22) and −16.24 (11.13) N/m. Overall p<0.001; adjusted p<0.001 for 4 Hz vs control, p=0.019 for 10 Hz vs control and p=0.049 for 4 vs 10 Hz. Tables 5 and 6 label the LGS row “mean (SD) or median (IQR)” without saying which applies to which group; the values are shown as printed, and the control value, whose within-group test was a Wilcoxon signed-rank test (Z = −1.540) rather than a t test, may be a median (IQR).
Analysis details and source
Table 5 reports within-group change summaries; Table 6 reports between-group tests and contrasts. Within-group change is not itself a causal treatment contrast. Median IQR widths are not represented as quartiles.
Lateral gastrocnemius dynamic stiffness (LGS), first session
MyotonPRO LGS N/m
| Group | n | Baseline | Reported change |
|---|---|---|---|
| 4 Hz radial ESWT plus rehabilitation | 14 | — | -24.33 (SD 12.27) |
| 10 Hz radial ESWT plus rehabilitation | 14 | — | -14.73 (SD 8.53) |
Values are group summaries identified in the caption; SD denotes standard deviation. CI is a confidence interval; IQR denotes the reported quartiles. A dash means no value was extracted for that cell.
4 Hz versus 10 Hz ESWT after one session
Independent-samples t test between ESWT groups
The reported change was −24.33 (SD 12.27) N/m at 4 Hz and −14.73 (SD 8.53) N/m at 10 Hz; the between-frequency test was p=0.024, meeting the table’s p<0.025 criterion. The trial does not establish frequency superiority on broader clinical outcomes.
Analysis details and source
Table 2 values are within-group changes; Table 3 supplies the between-group test. Preserve the published p<0.025 threshold. Values are as printed; in Table 2 the 4 Hz baseline and post-session means (314.17 and 281.95 N/m) differ by 32.22, not by the printed mean change of 24.33.
Pain during ESWT and other adverse events1 reported outcome
Pain during ESWT and other adverse events
VAS and active adverse-event questioning/examination VAS units; mild pain reports
Five patients reported mild pain during treatment, resolving shortly after rest. The paper reports no severe pain or serious/persistent complication.
Analysis details and source
Active questioning and examination were described; this small trial cannot rule out uncommon harms.
Modified Ashworth Scale distribution at three weeks1 reported outcome
Modified Ashworth Scale distribution at three weeks
Modified Ashworth Scale (ordinal grades 0, I, I+, II, III) participants in each ordinal category
Observed subgroup counts; these are not randomized treatment arms.
MAS change, control vs 4 Hz and 10 Hz ESWT
Kruskal–Wallis between-group comparison of change with Bonferroni-adjusted pairwise tests
At T2, ordinal MAS counts were control 0/9/3/2/0, 4 Hz 2/8/3/1/0 and 10 Hz 0/7/4/3/0 across grades 0/I/I+/II/III. The group comparison of change favored both ESWT groups versus control; there was no significant 4-vs-10-Hz contrast. These are ordinal categories, not interval-scale scores.
Analysis details and source
Distributions are source counts at T2; pairwise p-values compare change from baseline (Table 6). Do not infer equivalence from p>0.999.
MAS change, ordinal1 reported outcome
MAS change, ordinal
Modified Ashworth Scale (MAS) median ordinal points
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Conventional rehabilitation control | 14 | — | 0 |
| 4 Hz radial ESWT plus rehabilitation | 14 | — | -1 |
| 10 Hz radial ESWT plus rehabilitation | 14 | — | -1 |
Values are group summaries identified in the caption; SD denotes standard deviation. CI is a confidence interval; IQR denotes the reported quartiles. A dash means no value was extracted for that cell.
Median change was 0 (IQR 0) in control and −1 (IQR 1) in both ESWT groups. The overall between-group test was p=0.005; Bonferroni-adjusted comparisons: 4 Hz vs control p=0.008, 10 Hz vs control p=0.025, 4 vs 10 Hz p>0.999. IQR values are preserved in prose because the table reports IQR width, not quartile endpoints.
Analysis details and source
Table 5 reports within-group change summaries; Table 6 reports between-group tests and contrasts. Within-group change is not itself a causal treatment contrast. Median IQR widths are not represented as quartiles.
Ankle passive range of motion (PROM)1 reported outcome
Ankle passive range of motion (PROM)
Ankle passive range of motion degrees
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Conventional rehabilitation control | 14 | — | 2.86 (SD 2.18) |
| 4 Hz radial ESWT plus rehabilitation | 14 | — | 9.5 (SD 7.65) |
| 10 Hz radial ESWT plus rehabilitation | 14 | — | 9.5 (SD 5.6) |
Values are group summaries identified in the caption; SD denotes standard deviation. CI is a confidence interval; IQR denotes the reported quartiles. A dash means no value was extracted for that cell.
Three-week change across control, 4 Hz and 10 Hz ESWT
Kruskal–Wallis omnibus between-group test
Mean (SD) changes were 2.86 (2.18), 9.50 (7.65) and 9.50 (5.60) for control, 4 Hz and 10 Hz. Overall p=0.001; adjusted p=0.010 for 4 Hz vs control, 0.002 for 10 Hz vs control and >0.999 for 4 vs 10 Hz.
Analysis details and source
Table 5 reports within-group change summaries; Table 6 reports between-group tests and contrasts. Within-group change is not itself a causal treatment contrast. Median IQR widths are not represented as quartiles.
Ankle clonus score1 reported outcome
Ankle clonus score
Ankle clonus score ordinal score units
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Conventional rehabilitation control | 14 | — | 0 |
| 4 Hz radial ESWT plus rehabilitation | 14 | — | 0 |
| 10 Hz radial ESWT plus rehabilitation | 14 | — | -1 |
Values are group summaries identified in the caption; SD denotes standard deviation. CI is a confidence interval; IQR denotes the reported quartiles. A dash means no value was extracted for that cell.
Three-week change across control, 4 Hz and 10 Hz ESWT
Kruskal–Wallis omnibus between-group test
Median (IQR width) changes were 0 (0.25), 0 (1.00), and −1 (1.00) for control, 4 Hz and 10 Hz. The overall between-group comparison was not significant (p=0.194); no pairwise p-values were reported.
Analysis details and source
Table 5 reports within-group change summaries; Table 6 reports between-group tests and contrasts. Within-group change is not itself a causal treatment contrast. Median IQR widths are not represented as quartiles.
Medial gastrocnemius frequency (MGF)1 reported outcome
Medial gastrocnemius frequency (MGF)
MyotonPRO MGF Hz
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Conventional rehabilitation control | 14 | — | -0.15 (SD 1.26) |
| 4 Hz radial ESWT plus rehabilitation | 14 | — | -1.37 (SD 1.63) |
| 10 Hz radial ESWT plus rehabilitation | 14 | — | -1.16 (SD 0.88) |
Values are group summaries identified in the caption; SD denotes standard deviation. CI is a confidence interval; IQR denotes the reported quartiles. A dash means no value was extracted for that cell.
Three-week change across control, 4 Hz and 10 Hz ESWT
Kruskal–Wallis omnibus between-group test
Changes were −0.15 (1.26), −1.37 (1.63) and −1.16 (0.88) Hz for control, 4 Hz and 10 Hz. The overall between-group p-value was 0.010, but no adjusted pairwise contrasts were supplied. The paper is inconsistent here: Table 6 does not mark p=0.010 as significant although its threshold is p<0.025, and the Results text reports no MGF difference, while the Abstract states that both ESWT groups improved MGF significantly compared with control.
Analysis details and source
Table 5 reports within-group change summaries; Table 6 reports between-group tests and contrasts. Within-group change is not itself a causal treatment contrast. Median IQR widths are not represented as quartiles.
Lateral gastrocnemius frequency (LGF)1 reported outcome
Lateral gastrocnemius frequency (LGF)
MyotonPRO LGF Hz
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Conventional rehabilitation control | 14 | — | -0.98 (SD 0.92) |
| 4 Hz radial ESWT plus rehabilitation | 14 | — | -1.91 (SD 1.77) |
| 10 Hz radial ESWT plus rehabilitation | 14 | — | -1.78 (SD 1.36) |
Values are group summaries identified in the caption; SD denotes standard deviation. CI is a confidence interval; IQR denotes the reported quartiles. A dash means no value was extracted for that cell.
Three-week change across control, 4 Hz and 10 Hz ESWT
Kruskal–Wallis omnibus between-group test
Changes were −0.98 (0.92), −1.91 (1.77) and −1.78 (1.36) Hz. The overall between-group test was not significant (p=0.152); no adjusted pairwise comparisons were reported.
Analysis details and source
Table 5 reports within-group change summaries; Table 6 reports between-group tests and contrasts. Within-group change is not itself a causal treatment contrast. Median IQR widths are not represented as quartiles.
Medial gastrocnemius dynamic stiffness (MGS)1 reported outcome
Medial gastrocnemius dynamic stiffness (MGS)
MyotonPRO MGS N/m
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Conventional rehabilitation control | 14 | — | -2.74 (SD 5.45) |
| 4 Hz radial ESWT plus rehabilitation | 14 | — | -29.05 (SD 20.39) |
| 10 Hz radial ESWT plus rehabilitation | 14 | — | -10.52 (SD 11.63) |
Values are group summaries identified in the caption; SD denotes standard deviation. CI is a confidence interval; IQR denotes the reported quartiles. A dash means no value was extracted for that cell.
Three-week change across control, 4 Hz and 10 Hz ESWT
Kruskal–Wallis omnibus between-group test
Changes were −2.74 (5.45), −29.05 (20.39) and −10.52 (11.63) N/m. Overall p<0.001; adjusted p<0.001 for 4 Hz vs control, p=0.237 for 10 Hz vs control and p=0.113 for 4 vs 10 Hz.
Analysis details and source
Table 5 reports within-group change summaries; Table 6 reports between-group tests and contrasts. Within-group change is not itself a causal treatment contrast. Median IQR widths are not represented as quartiles.
Fugl–Meyer Assessment, lower extremity (FMA-LE)1 reported outcome
Fugl–Meyer Assessment, lower extremity (FMA-LE)
Fugl–Meyer Assessment, lower extremity median score points
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Conventional rehabilitation control | 14 | — | 1 |
| 4 Hz radial ESWT plus rehabilitation | 14 | — | 1.5 |
| 10 Hz radial ESWT plus rehabilitation | 14 | — | 1 |
Values are group summaries identified in the caption; SD denotes standard deviation. CI is a confidence interval; IQR denotes the reported quartiles. A dash means no value was extracted for that cell.
Three-week change across control, 4 Hz and 10 Hz ESWT
Kruskal–Wallis omnibus between-group test
Median (IQR width) changes were 1.00 (1.25), 1.50 (3.00) and 1.00 (3.25) for control, 4 Hz and 10 Hz. The between-group test was not significant (p=0.483); the source reports no pairwise comparisons.
Analysis details and source
Table 5 reports within-group change summaries; Table 6 reports between-group tests and contrasts. Within-group change is not itself a causal treatment contrast. Median IQR widths are not represented as quartiles.
Modified Ashworth Scale distribution after one session1 reported outcome
Modified Ashworth Scale distribution after one session
Modified Ashworth Scale, ordinal grade participants in each ordinal category
Observed subgroup counts; these are not randomized treatment arms.
4 Hz versus 10 Hz ESWT
Mann–Whitney U between-group comparison of change
At this single-session time point, the article provides ordinal MAS counts for the 4 Hz and 10 Hz groups only, not the control group. Counts in grade order 0/I/I+/II/III were 0/9/2/2/1 for 4 Hz and 0/3/8/3/0 for 10 Hz. The between-frequency comparison of change was not significant (Table 3 p=0.691).
Analysis details and source
Table 2 change data are within-group; Table 3 compares changes between the two ESWT frequencies. These ordinal grades should not be treated as interval-scale scores.
How treatment was delivered
HMJYS-A device; prone position; probe at the muscle–tendon junction and midpoint between the medial/lateral gastrocnemius heads; 2,000 pulses/session, 2 bar; 4 Hz or 10 Hz; six sessions over three weeks, twice weekly, with at least two days between sessions.
All groups received individualized conventional rehabilitation for three weeks, six days/week and four hours/day, including physical therapy, occupational therapy, speech training and acupuncture.
Baseline, after the first treatment, and at the end of the three-week course; VAS pain before and after each ESWT session.
Five participants reported mild treatment pain that resolved after rest; no severe pain or serious/persistent complication was observed in the two ESWT groups (28 completers).
The sample is small (14 analyzed per arm), follow-up ended with the three-week intervention, and intermediate frequencies were not tested.
The trial changed a prespecified secondary outcome during data collection and corrected the registered disease-duration criterion during manuscript preparation; these changes affect interpretation and should be reviewed alongside the protocol.
No long-term outcomes were collected; MAS grade I was the most common baseline grade (19 of 42) and only 2 participants had grade III, limiting inference for more severe spasticity.
Findings checked against the full source by a separate automated reviewer.
Limitations noted by the reviewers
- The sample is small (14 analyzed per arm), follow-up ended with the three-week intervention, and intermediate frequencies were not tested.
- The trial changed a prespecified secondary outcome during data collection and corrected the registered disease-duration criterion during manuscript preparation; these changes affect interpretation and should be reviewed alongside the protocol.
- No long-term outcomes were collected; MAS grade I was the most common baseline grade (19 of 42) and only 2 participants had grade III, limiting inference for more severe spasticity.
Source-linked critical reading
Randomized treatment-schedule comparison
Technical source record
Complete article XML read, including methods, results, tables, figure captions, discussion, conclusion and disclosures.
In this small study, adding shock-wave sessions to intensive rehabilitation improved some short-term measures. We do not yet know whether the changes last or which frequency is best for different levels of spasticity.
In post-stroke ankle plantar-flexor spasticity, do 4 Hz or 10 Hz radial ESWT added to conventional rehabilitation improve tone and function over rehabilitation alone?
Single-center prospective single-blind parallel RCT; 45 randomized 1:1:1 with concealed opaque envelopes and blinded outcome assessor. One participant per group was lost, leaving 42 (14/group). Analyses used available data.
Adults aged 30–75 with first stroke, hemiparesis and MAS grade I–III ankle plantar-flexor spasticity, 2–48 weeks after stroke. MAS grade I was the most common baseline grade (19 of 42); severe-grade subgroup inference is limited.
HMJYS-A radial device, 2,000 pulses at 2 bar; probe at gastrocnemius muscle–tendon junction and midpoint between heads. Six sessions over three weeks, twice weekly; 4 Hz or 10 Hz. ESWT groups also received the same conventional rehabilitation.
All groups received individualized conventional rehabilitation six days/week, four hours/day for three weeks. The control had rehabilitation without ESWT. Thus the question concerns an adjunct, not replacement.
Primary outcome was Modified Ashworth Scale; secondary outcomes included passive range of motion, ankle clonus, MyotonPRO muscle measures and Fugl-Meyer lower-extremity score. Immediate post-session and three-week results were reported. Between-group results were mixed by outcome.
At three weeks, MAS and PROM changes favored both ESWT groups over control; LGS differed across groups and between 4 and 10 Hz. Many other outcomes had no between-group differences. Nonsignificance does not prove equivalent frequencies. Table 6 reports adjusted pairwise tests.
Five patients reported mild pain that resolved after rest; no serious or persistent complication was reported. Active questioning and examination were performed after sessions, but the sample is too small to characterize uncommon harms.
Only 14 participants per arm were analyzed and no post-treatment long-term follow-up occurred. Frequency was limited to 4 and 10 Hz; MAS grade I was the most common baseline grade (19 of 42). MAS is subjective. A prespecified secondary outcome was discontinued and the eligibility duration criterion was changed during manuscript preparation.
The findings concern selected stroke patients receiving intensive rehabilitation at one center. They do not establish durable functional benefit, stand-alone ESWT effectiveness or an optimal frequency for severe spasticity.
Larger multisite trials should prespecify patient-important mobility outcomes, retain transparent protocol changes, include longer follow-up and broader severity, compare frequencies and doses prospectively, and report harms systematically.
Read the original article.
Efficacy of different frequencies of extracorporeal shockwave on plantar flexor spasticity of the ankle in patients with stroke: a single-center, prospective, single-blind, randomized controlled trial.
Ke M, Zhou P, Li D, Kong W, Xie C, Wang Y, Wang Z, Long J. · Journal of neuroengineering and rehabilitation · 2026
Open the publication record or available full-text files.
Checked against the full paper · Clinical review: Dr. Marcus Yu Bin Pai, · How we review →