Post-stroke spasticity and motor function · Randomized trial · 2026
Can shockwave help arm movement months after a stroke?
In this trial, arm movement scores rose from about 22 to about 27 out of 66, a month after four sessions of radial shockwave, one a week, while the sham group stayed near 14. Gains were clearest at the shoulder and elbow.
Possible early help for arm movement- adults 6–24 months after a first stroke
- 23
- sessions, one a week
- 4
- weeks from start to final check
- 8
Good to know: No one had other treatment during the study.
Imamura M, Sugawara AT, Santos ACA, et al. Clinics (Sao Paulo, Brazil), 2026 · Read the paper ↗
How this study worked
What happened
Arm movement after a stroke
Fugl–Meyer arm score, 0–66 · start and week 8
StartWeek 8
Show the numbers
| Group | People | Start | Week 8 |
|---|---|---|---|
| Radial shockwave | 9 | 22.33 | 27.44 |
| Sham shockwave | 7 | 14.14 | 13.86 |
The treatment
Radial shockwave
Radial, once a week for 4 weeks
- Sessions
- 4, one a week
- Pulses
- 5,000 per session · spread over the forearm and hand muscles
- Strength
- 3.5–4.0bar about 0.07 mJ/mm²; set to what each person could tolerate · 15 per second
- Numbing
- Not mentioned
- Device
- Swiss Dolorclast
Compared with: Sham shockwave: a handpiece that looked and sounded the same but gave no pulses
Settings are what each study used, not a prescription.
Who took part
23 adults 6–24 months after a first stroke; none had a measurable brain-to-arm nerve signal at the start.
hollow squares: no final score
Side effects
The study does not report side effects.
All the numbers and the full appraisal
The results, in numbers.
Adults with first ischemic stroke in middle cerebral artery territory, 6–24 months prior, imaging-confirmed; all 23 recruited participants lacked detectable MEP at baseline.
The article reports 23 total randomized. Table 1 lists 12 active-arm and 11 sham-arm participants; final per-protocol comparison used 9 and 7, respectively.
The article reports 23 total randomized. Table 1 lists 12 active-arm and 11 sham-arm participants; final per-protocol comparison used 9 and 7, respectively.
Fugl-Meyer Upper Extremity total score (primary)1 reported outcome
Fugl-Meyer Upper Extremity total score (primary)
FMA-UE global score
| Group | n | Baseline | Follow-up |
|---|---|---|---|
| Radial ESWT | 9 | 22.333 | 27.444 |
| Sham radial ESWT | 7 | 14.143 | 13.857 |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
rESWT minus sham
1 study · 16 participants in this estimate
ANCOVA adjusted for baseline; per-protocol
rESWT versus sham, per-protocol
ANCOVA; alternative tests: Student p=0.096, Welch p=0.091, Hotelling T² p=0.037, Johansen T² p=0.033
ANCOVA and the two T² methods favor rESWT, but Student/Welch comparisons of change were not significant. No adjusted CI.
Analysis details and source
Per-protocol n=9 rESWT and n=7 sham. Primary and secondary inferential tests are method-dependent; no confidence interval for adjusted estimates is reported.
Proximal FMA-UE score1 reported outcome
Proximal FMA-UE score
FMA-UE proximal score
| Group | n | Baseline | Follow-up |
|---|---|---|---|
| Radial ESWT | 9 | 10.444 | 13 |
| Sham radial ESWT | 7 | 6.286 | 5.143 |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
rESWT minus sham
1 study · 16 participants in this estimate
ANCOVA adjusted for baseline; per-protocol
rESWT versus sham, per-protocol
ANCOVA; alternative tests: Student p=0.100, Welch p=0.106, Hotelling T² p=0.027, Johansen T² p=0.028
Adjusted and T² tests favor rESWT, while unadjusted change t tests are not significant. This is a decomposed outcome the authors say was proposed after recognizing proximal/distal responses may differ.
Analysis details and source
Per-protocol n=9 rESWT and n=7 sham. Primary and secondary inferential tests are method-dependent; no confidence interval for adjusted estimates is reported. The abstract reports p=0.004, whereas Table 5 reports ANCOVA p=0.030 and T² p=0.027/0.028; unresolved reporting discrepancy.
Distal FMA-UE score1 reported outcome
Distal FMA-UE score
FMA-UE distal score
| Group | n | Baseline | Follow-up |
|---|---|---|---|
| Radial ESWT | 9 | 5.222 | 8.667 |
| Sham radial ESWT | 7 | 2.714 | 2.714 |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
rESWT minus sham
1 study · 16 participants in this estimate
ANCOVA adjusted for baseline; per-protocol
rESWT versus sham, per-protocol
ANCOVA; alternative tests: Student p=0.161, Welch p=0.140, Hotelling T² p=0.168, Johansen T² p=0.134
No statistically significant between-group effect across reported tests.
Analysis details and source
Per-protocol n=9 rESWT and n=7 sham. Primary and secondary inferential tests are method-dependent; no confidence interval for adjusted estimates is reported.
Muscle strength (secondary, exploratory)1 reported outcome
Muscle strength (secondary, exploratory)
MRC Scale score
| Group | n | Baseline | Follow-up |
|---|---|---|---|
| Radial ESWT | 9 | 1.802 | 2.37 |
| Sham radial ESWT | 7 | 1.413 | 1.206 |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
rESWT minus sham
1 study · 16 participants in this estimate
ANCOVA adjusted for baseline; per-protocol
rESWT versus sham, per-protocol
ANCOVA; alternative tests: Student p=0.030, Welch p=0.025, Hotelling T² p=0.030, Johansen T² p=0.021
Between-group tests favor rESWT; the authors call this secondary and exploratory. No adjusted CI or validated aggregate scale range is reported.
Analysis details and source
Per-protocol n=9 rESWT and n=7 sham. Primary and secondary inferential tests are method-dependent; no confidence interval for adjusted estimates is reported.
Flexor muscle spasticity1 reported outcome
Flexor muscle spasticity
Modified Ashworth Scale, flexors score
| Group | n | Baseline | Follow-up |
|---|---|---|---|
| Radial ESWT | 9 | 1.259 | 0.833 |
| Sham radial ESWT | 7 | 0.667 | 0.762 |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
rESWT versus sham, per-protocol
Johansen T²; authors identify it as the most reliable test after assumptions for other methods were violated; p=0.584
The author-preferred Johansen test found no between-group difference. Student/Welch and ANCOVA results differ, and univariate normality, bivariate homoscedasticity, dissociation and parallelism were rejected; do not infer benefit from a single test.
Analysis details and source
Per-protocol n=9 rESWT and n=7 sham. Primary and secondary inferential tests are method-dependent; no confidence interval for adjusted estimates is reported.
Extensor muscle spasticity1 reported outcome
Extensor muscle spasticity
Modified Ashworth Scale, extensors score
| Group | n | Baseline | Follow-up |
|---|---|---|---|
| Radial ESWT | 9 | 2.259 | 2.148 |
| Sham radial ESWT | 7 | 2.143 | 2.19 |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
rESWT versus sham, per-protocol
Reported comparative tests: Welch p=0.618, ANCOVA p=0.714, Hotelling T² p=0.908, Johansen T² p=0.903
Reported tests found no between-group difference.
Analysis details and source
Per-protocol n=9 rESWT and n=7 sham. Primary and secondary inferential tests are method-dependent; no confidence interval for adjusted estimates is reported.
Grip strength1 reported outcome
Grip strength
Grip dynamometer kg-force
| Group | n | Baseline | Follow-up |
|---|---|---|---|
| Radial ESWT | 9 | 5.611 | 7.556 |
| Sham radial ESWT | 6 | 2.571 | 3 |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
rESWT versus sham, per-protocol
Reported tests: Welch p=0.347, ANCOVA p=0.535, Hotelling T² p=0.960, Johansen T² p=0.946
No reliable between-group difference established; sham-group linearity test did not converge and assumptions were not fully met.
Analysis details and source
Per-protocol n=9 rESWT and n=7 sham, but Table 4 marks sham participant 19's post-treatment grip value as missing: the sham follow-up mean (3.000) and the between-group tests (Student t(13), ANCOVA t(12), T²(2,14)) use 6 sham participants, while the printed sham baseline mean (2.571) includes all 7. The printed sham difference 1.083 is therefore not 3.000 − 2.571 (0.429); it matches the mean change of the 6 sham participants with both values in Table 4. Values are kept as printed. Primary and secondary inferential tests are method-dependent; no confidence interval for adjusted estimates is reported.
Pinch strength1 reported outcome
Pinch strength
Pinch dynamometer kg-force
| Group | n | Baseline | Follow-up |
|---|---|---|---|
| Radial ESWT | 9 | 1.322 | 2.05 |
| Sham radial ESWT | 6 | 0.457 | 0.583 |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
rESWT versus sham, per-protocol
Reported tests: Welch p=0.160, ANCOVA p=0.139, Hotelling T² p=0.307, Johansen T² p=0.228
No reliable between-group difference established; repeated zero values and rejected normality make the findings inconclusive.
Analysis details and source
Per-protocol n=9 rESWT and n=7 sham, but Table 4 marks sham participant 19's post-treatment pinch value as missing: the sham follow-up mean (0.583) and the between-group tests (Student t(13), ANCOVA t(12), T²(2,14)) use 6 sham participants, while the printed sham baseline mean (0.457) includes all 7. The printed sham difference 0.217 is therefore not 0.583 − 0.457 (0.126); it matches the mean change of the 6 sham participants with both values in Table 4. Values are kept as printed. Primary and secondary inferential tests are method-dependent; no confidence interval for adjusted estimates is reported.
How treatment was delivered
Swiss Dolorclast; 36-mm applicator with Power+ handpiece; 5,000 continuous pulses per session at 15 Hz; 3.5–4.0 bar, approximately 0.07 mJ/mm²; pressure/energy individually adjusted to maximum tolerance; 4 weekly sessions.
Diffuse treatment of elbow and wrist flexors in the forearm and hand interossei.
Handpiece appeared, felt, and sounded the same but emitted no shockwave or tactile sensation; active/sham applied at four weekly visits.
No other intervention was given during the study period.
No systematic arm-specific adverse-event counts are reported. Seven participants were not included in the final per-protocol analysis due to botulinum toxin, withdrawal, or missing follow-up; the article does not establish that these were all treatment-related events.
Only 16/23 randomized were included in the final per-protocol analysis; the article does not provide an intention-to-treat effect estimate.
The primary statistical inference is method-dependent: t tests of change were not significant, while baseline-adjusted ANCOVA and T² approaches were significant for total/proximal FMA-UE. No confidence intervals are reported.
The per-protocol analysis, short post-treatment follow-up and restriction to absent-baseline-MEP patients limit generalizability.
Several secondary outcomes violate model assumptions; authors identify the Johansen T² test as most reliable for flexor MAS, which found no difference. Pinch results are inconclusive because of zeros and distribution/linearity issues.
The sham did not provide tactile sensation, creating a possible clue to allocation; the paper nevertheless describes patients as blinded.
The paper discloses an author’s consultancy relationship with the device manufacturer; authors state the manufacturer had no role in the trial.
Findings checked against the full source by a separate automated reviewer.
Limitations noted by the reviewers
- Only 16/23 randomized were included in the final per-protocol analysis; the article does not provide an intention-to-treat effect estimate.
- The primary statistical inference is method-dependent: t tests of change were not significant, while baseline-adjusted ANCOVA and T² approaches were significant for total/proximal FMA-UE. No confidence intervals are reported.
- The per-protocol analysis, short post-treatment follow-up and restriction to absent-baseline-MEP patients limit generalizability.
- Several secondary outcomes violate model assumptions; authors identify the Johansen T² test as most reliable for flexor MAS, which found no difference. Pinch results are inconclusive because of zeros and distribution/linearity issues.
- The sham did not provide tactile sensation, creating a possible clue to allocation; the paper nevertheless describes patients as blinded.
- The paper discloses an author’s consultancy relationship with the device manufacturer; authors state the manufacturer had no role in the trial.
Source-linked critical reading
Randomized sham-controlled trial; per-protocol primary analysis
Technical source record
Full article XML read: Abstract plus all 27 sections, five tables, four figure captions and author/declaration sections. Figure image files are absent from the retrieved package; graphical figures were not visually inspected.
A trial reported possible short-term improvement in some upper-limb measures with radial shockwave, but results vary across statistical methods and several secondary analyses are unreliable. The findings need confirmation and do not establish long-term benefit.
In chronic ischemic middle cerebral artery stroke with absent baseline MEPs, does radial ESWT improve upper-limb function compared with sham?
Randomized double-blind sham-controlled trial at a single rehabilitation center in São Paulo; 23 randomized, 16 analyzed per protocol (9 active, 7 sham). Planned sample was 10 per arm.
Adults ≥18 years with imaging-confirmed middle cerebral artery stroke 6–24 months earlier. All 23 recruited patients had absent MEPs at baseline, an unexpected finding not required by eligibility criteria.
Swiss Dolorclast, 36-mm applicator on Power+ handpiece; 4 weekly sessions; 5,000 continuous shocks/session at 15 Hz, 3.5–4.0 bar, approximately 0.07 mJ/mm²; pressure/energy adjusted to individual tolerance. Applied diffusely to elbow/wrist flexors in forearm and hand interossei.
The sham handpiece looked, felt and sounded like the active one but emitted no ESWT and provided no tactile sensation. Patients and assessors were described as blinded; the no-sensation difference could reveal allocation.
Primary outcome: global, proximal and distal FMA-UE at 4 weeks after the final treatment (8 weeks from baseline). Secondary outcomes: MRC strength, MAS flexor/extensor tone, grip and pinch strength, plus TMS measures. No usable comparative results for ICI/ICF/CSP are reported in the tables.
For global FMA-UE, ANCOVA (baseline adjusted) and Hotelling/Johansen T² tests were significant, but Student/Welch tests of change were not; no adjusted CI was reported. Proximal FMA-UE had method-dependent findings. The abstract reports p=0.004, while Table 5 reports ANCOVA p=0.030 and Hotelling/Johansen p=0.027/0.028; the discrepancy is unresolved. Distal FMA-UE tests were nonsignificant. MRC favored active treatment but was explicitly exploratory. MAS flexor showed no between-group difference in the authors’ preferred Johansen test; grip and pinch results were inconclusive due to assumption violations.
The article does not provide systematic arm-specific adverse-event counts. Reasons for discontinuation included botulinum toxin use, withdrawal, or missed follow-up; these should not all be attributed to treatment harms.
Per-protocol analysis, with 7/23 randomized not included in the final analysis, and final evaluation only 4 weeks after treatment. All participants lacked MEP at baseline, limiting generalizability. Multiple tests produce inconsistent inference; several secondary tests violate assumptions. The sham provided no sensation. The manufacturer’s consultant relationship is disclosed, though authors state the manufacturer had no study role. Figure pixels were unavailable for visual inspection.
Findings concern a selected chronic stroke subgroup with absent MEPs at one Brazilian center. They do not establish durable benefit, benefit in patients with preserved MEPs, or general effectiveness for stroke rehabilitation.
Conduct a larger preregistered multicenter trial with intention-to-treat analysis, a credible sensory-matched sham, longer follow-up, prespecified primary/global and segmental outcomes, correction for multiplicity, transparent assumptions and confidence intervals, systematic harms, and stratification by baseline MEP status.
Read the original article.
Radial extracorporeal shock wave therapy improves upper limb function in chronic stroke patients: a randomized sham-controlled trial.
Imamura M, Sugawara AT, Santos ACA, Siqueira JO, Silveira PSP, Schmitz C, Battistella LR. · Clinics (Sao Paulo, Brazil) · 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 →