Knee tendon and soft-tissue disorders: large pooled effects, high heterogeneity · Review · 2018
Does shockwave help knee tendon and soft-tissue problems?
In this review of 19 trials, mostly of patellar tendon pain, shockwave on average eased pain and improved function more than comparison treatments. But the trials differed greatly and some results were unusually large. Beyond one month, success was not clearly better.
Helped more than comparison groups, on average- trials
- 19
- people with knee tendon or soft-tissue problems
- 1,189
- trials of patellar tendon pain
- 8
Good to know: Against acupuncture, iontophoresis or a PRP injection, pain gaps were not clear.
Liao CD, Xie GM, Tsauo JY, et al. BMC Musculoskeletal Disorders, 2018 · Read the paper ↗
How this study worked
What happened
- studies
- 16
- people
- 892
- Compared with
- Sham or no shockwave, exercise, physiotherapy, medicines, injections, acupuncture or other treatments
Knee pain compared with other treatments
Standardised difference (no units) · below zero means less pain with shockwave
Show the numbers
| Difference (standardised difference (no units)) | likely range | |
|---|---|---|
| Up to 1 month | −1.18 | −1.67 to −0.68 |
| 1 to 3 months | −1.07 | −1.84 to −0.31 |
| 3 to 6 months | −1.98 | −3.32 to −0.64 |
What the trials used
- Shockwave regimens
- Focused shockwaves in 8 trials (5 high-energy, 3 low-energy) and radial in 11 (5 high, 6 low), with high energy defined as ≥0.2 mJ/mm². Table 3 ranges: 0.03–0.58 mJ/mm², 1,000–4,600 impulses per session, 1 to 36 sessions over 1–12 weeks. Local anaesthesia was used only in one ACL reconstruction trial, where a single session was given during surgical anaesthesia.
- Comparators and co-interventions
- Nine trials used sham or no shockwaves; others compared exercise, physiotherapy, medicines, iontophoresis, ultrasound diathermy, PRP, acupuncture or injection. Twelve trials used co-interventions (physiotherapy, acupuncture, exercise, manual therapy or medication), not always the same in both arms. In four "placebo" radial trials the control simply received no shockwaves.
- Review methods
- Search to June 2018 in MEDLINE, PubMed, EMBASE, Cochrane, PEDro, CNKI and Google Scholar without language limits; PEDro and Cochrane risk-of-bias assessment; RevMan 5.3; fixed or random effects by heterogeneity; subgroups by wave type, energy (cut-off 0.2 mJ/mm²), duration (<1 versus ≥1 month), control, population, disease and co-intervention.
Who took part
1,189 people, average age about 35, mostly with patellar tendon pain; also pes anserine pain, knee stiffness after injury, Osgood–Schlatter disease and others.
Side effects
The review reports no serious side effects in the included trials.
All the numbers and the full appraisal
The results, in numbers.
Nineteen randomized trials, 1,189 patients (562 shockwave, 627 control), mean age 34.7 years: patellar tendinopathy in eight, pes anserine tendinopathy in two, ACL injury or reconstruction in two, post-traumatic knee stiffness in two, and one each of Osgood–Schlatter disease, iliotibial band syndrome, infrapatellar fat pad injury, traumatic synovitis and popliteal cyamella. Many trials were Chinese-language reports.
Treatment success, overall1 reported outcome
Treatment success, overall
Odds ratio
Shockwaves versus trial-specific controls
16 studies · 842 participants in this estimate · I²=60%
Pooled estimate; random effects where heterogeneity was significant
Logarithmic ratio scale.
Across 16 trials, shockwave-treated patients had higher odds of a successful rating than controls (OR 3.36, 95% CI 1.84–6.12), graded moderate by the authors.
Analysis details and source
Treatment success was mostly a Likert or Roles–Maudsley rating; definitions varied between trials. OR above 1 favours shockwaves.
Treatment success, ≤1 month1 reported outcome
Treatment success, ≤1 month
Odds ratio
Shockwaves versus trial-specific controls
11 studies · 518 participants in this estimate · I²=63%
Pooled estimate; random effects where heterogeneity was significant
Logarithmic ratio scale.
At immediate follow-up success odds were higher with shockwaves (OR 3.09). The authors report no significant combined effect on success at short-, medium- or long-term follow-up.
Analysis details and source
Treatment success was mostly a Likert or Roles–Maudsley rating; definitions varied between trials. OR above 1 favours shockwaves.
Treatment success after 1 month1 reported outcome
Treatment success after 1 month
Odds ratio
For all shockwave types combined, success did not differ significantly from controls beyond one month. In Table 5 focused trials gave ORs of 1.02, 0.49 and 0.33 at short, medium and long term (all non-significant), while radial trials gave significant ORs at short and long term.
Analysis details and source
Combined estimates for these periods are only in Additional file 4, which was not read. The text says radial success was significant at medium term, but Table 5 lists no radial trials for that period.
Pain reduction, overall7 reported outcomes
Pain reduction, overall
VAS pain Standardized mean difference
Shockwaves versus trial-specific controls
18 studies · 1,084 participants in this estimate · I²=95%
Pooled estimate; random effects where heterogeneity was significant
Pain improved more with shockwaves than with controls on average (SMD −1.49), but heterogeneity was extreme (I²=95%), so the size of the effect is uncertain.
Analysis details and source
Pain was converted to 0–100 mm and analysed as change from baseline; negative SMD favours shockwaves. Missing SDs of change were imputed with an assumed correlation of 0.8, and medians were converted to means.
Pain reduction, ≤1 month
VAS pain Standardized mean difference
Shockwaves versus trial-specific controls
16 studies · 892 participants in this estimate · I²=91%
Pooled estimate; random effects where heterogeneity was significant
Immediate pain relief favoured shockwaves (SMD −1.18; 16 trials).
Analysis details and source
Pain was converted to 0–100 mm and analysed as change from baseline; negative SMD favours shockwaves. Missing SDs of change were imputed with an assumed correlation of 0.8, and medians were converted to means.
Pain reduction, >1–3 months
VAS pain Standardized mean difference
Shockwaves versus trial-specific controls
10 studies · 469 participants in this estimate · I²=93%
Pooled estimate; random effects where heterogeneity was significant
Short-term pain also favoured shockwaves (SMD −1.07; 10 trials).
Analysis details and source
Pain was converted to 0–100 mm and analysed as change from baseline; negative SMD favours shockwaves. Missing SDs of change were imputed with an assumed correlation of 0.8, and medians were converted to means.
Pain reduction, >3–6 months
VAS pain Standardized mean difference
Shockwaves versus trial-specific controls
6 studies · 466 participants in this estimate · I²=97%
Pooled estimate; random effects where heterogeneity was significant
Medium-term pain favoured shockwaves (SMD −1.98; 6 trials) with very high heterogeneity.
Analysis details and source
Pain was converted to 0–100 mm and analysed as change from baseline; negative SMD favours shockwaves. Missing SDs of change were imputed with an assumed correlation of 0.8, and medians were converted to means.
Pain, focused waves >6 months
VAS pain Standardized mean difference
Focused shockwaves versus comparator (single trial)
1 study · 46 participants in this estimate · Not applicable (one trial)
Single-trial estimate reported in the subgroup table
The only focused trial with pain beyond six months (46 patients) favoured the comparator (SMD 0.71, 95% CI 0.11 to 1.31). The text instead says focused treatment reduced pain at every follow-up except short term.
Analysis details and source
Trial identity is not named in Table 5; Figure 6 places Vetrano 2013 (focused shockwaves versus PRP) in the long-term group (extractor reading). Positive SMD favours the comparator.
Pain versus invasive comparators (radial)
VAS pain Standardized mean difference
Radial shockwaves versus acupuncture (Zhou 2015) or iontophoresis (Liu 2016), both classed by the review as invasive
2 studies · 160 participants in this estimate · I²=99%
Pooled estimate; random effects where heterogeneity was significant
Against invasive comparators, neither radial (SMD −0.80, 95% CI −3.85 to 2.24) nor focused treatment (one trial against PRP injection, SMD 0.36, −0.22 to 0.95) differed significantly in pain.
Analysis details and source
Pain was converted to 0–100 mm and analysed as change from baseline; negative SMD favours shockwaves. Missing SDs of change were imputed with an assumed correlation of 0.8, and medians were converted to means.
Pain versus sham or no waves (focused)
VAS pain Standardized mean difference
Focused shockwaves versus "placebo" control
5 studies · 234 participants in this estimate · I²=97%
Pooled estimate; random effects where heterogeneity was significant
Against the "placebo" group, focused trials favoured shockwaves (SMD −3.22) and radial trials did too (SMD −1.14, 95% CI −1.42 to −0.86; 4 trials, 232 patients).
Analysis details and source
Per Table 2, the focused "placebo" pain set includes four sham-controlled trials plus Chen 2014 (popliteal cyamella), whose control arm received no shockwaves; the focused placebo success set instead includes Wang 2014, a no-shockwave control under surgical anaesthesia. All four radial "placebo" trials used no-shockwave controls with shared treatment (extractor reading). An SMD of −3.22 is unusually large.
Patient-reported function1 reported outcome
Patient-reported function
Standardized mean difference
Shockwaves versus trial-specific controls
13 studies · 728 participants in this estimate · I²=96%
Pooled estimate; random effects where heterogeneity was significant
Questionnaire-based function (mostly VISA-P, plus Lysholm, IKDC, Lequesne, HSS and others) favoured shockwaves on average (SMD 2.03), with extreme heterogeneity.
Analysis details and source
Different questionnaires were combined as SMD; positive values favour shockwaves.
Knee range of motion1 reported outcome
Knee range of motion
Standardized mean difference
Shockwaves versus trial-specific controls
Not stated for the overall estimate
Overall estimate given only in the abstract
Range of motion favoured shockwaves (abstract SMD 1.76). The body reports it separately for focused (2 trials, 117 patients, SMD 2.61) and radial treatment (2 trials, 90 patients, SMD 1.09), from four trials in total.
Analysis details and source
The overall 1.76 estimate does not appear in the main text; its forest plot is in Additional file 9, which was not read.
Adverse events1 reported outcome
Adverse events
Narrative
The review states that no clinically relevant adverse events or severe complications were reported in the included trials. Losses to follow-up in four focused trials ranged from 1 to 7 patients (3.2–31.8%).
Analysis details and source
Adverse events were not prespecified review outcomes and trial-level reporting is in Additional file 10 (not read); unreported events are not zero.
How treatment was delivered
Focused shockwaves in 8 trials (5 high-energy, 3 low-energy) and radial in 11 (5 high, 6 low), with high energy defined as ≥0.2 mJ/mm². Table 3 ranges: 0.03–0.58 mJ/mm², 1,000–4,600 impulses per session, 1 to 36 sessions over 1–12 weeks. Local anaesthesia was used only in one ACL reconstruction trial, where a single session was given during surgical anaesthesia.
Nine trials used sham or no shockwaves; others compared exercise, physiotherapy, medicines, iontophoresis, ultrasound diathermy, PRP, acupuncture or injection. Twelve trials used co-interventions (physiotherapy, acupuncture, exercise, manual therapy or medication), not always the same in both arms. In four "placebo" radial trials the control simply received no shockwaves.
Search to June 2018 in MEDLINE, PubMed, EMBASE, Cochrane, PEDro, CNKI and Google Scholar without language limits; PEDro and Cochrane risk-of-bias assessment; RevMan 5.3; fixed or random effects by heterogeneity; subgroups by wave type, energy (cut-off 0.2 mJ/mm²), duration (<1 versus ≥1 month), control, population, disease and co-intervention.
The review reports no clinically relevant adverse events, side effects or severe complications in the included trials, but adverse events were not prespecified and their trial-level summary is in an unread supplement. The Conclusions instead mention "a few minor side effects", citing an elbow trial that is not among the 19 included, so minor effects may have gone unsummarized; not reported does not mean zero. The authors received no funding and declared no competing interests.
Heterogeneity was extreme for pain and function (I² 91–97%), and some SMDs are implausibly large (for example −5.17 for pain with focused monotherapy in two trials, and 6.98 for function in one trial against a noninvasive control). Change-score SDs were imputed with an assumed correlation of 0.8 and medians converted, which can inflate SMDs (extractor interpretation).
Text and tables conflict: the text says focused treatment reduced pain at all follow-ups except short term, while Table 5 shows the >6-month focused result favouring the comparator (SMD 0.71); radial success is called significant at medium term although Table 5 lists no trials; I² for focused success is 24% in the text and 25% in Table 5; 18 trials are said to use 3–6 sessions over 2–6 weeks, yet Table 3 lists 36, 18 and 8 sessions in three trials; the Discussion refers to 20 trials.
The "placebo" subgroups mix four true sham-controlled trials with trials whose controls received no shockwaves alongside shared treatment (Table 2), so they are not all sham comparisons.
Risk of bias: participants were not blinded in 14 trials and therapists in 16; fewer than half reported sequence generation or concealment; outcome assessors were blinded in six trials and explicitly unblinded in one. The authors nonetheless rated all trials medium or high quality (PEDro median 6, range 5–9). Evidence levels used a best-evidence synthesis, not GRADE.
Eleven of the 19 trials were non-English (Chinese-journal) reports, and English titles were translated by software where missing; most outcomes were patient-reported and only five trials measured performance. Ten trials did not report funding. Additional files 1–10 (search strategies, funding, forest-plot data, adverse events) were not read.
Findings checked against the full source by a separate automated reviewer.
Limitations noted by the reviewers
- Heterogeneity was extreme for pain and function (I² 91–97%), and some SMDs are implausibly large (for example −5.17 for pain with focused monotherapy in two trials, and 6.98 for function in one trial against a noninvasive control). Change-score SDs were imputed with an assumed correlation of 0.8 and medians converted, which can inflate SMDs (extractor interpretation).
- Text and tables conflict: the text says focused treatment reduced pain at all follow-ups except short term, while Table 5 shows the >6-month focused result favouring the comparator (SMD 0.71); radial success is called significant at medium term although Table 5 lists no trials; I² for focused success is 24% in the text and 25% in Table 5; 18 trials are said to use 3–6 sessions over 2–6 weeks, yet Table 3 lists 36, 18 and 8 sessions in three trials; the Discussion refers to 20 trials.
- The "placebo" subgroups mix four true sham-controlled trials with trials whose controls received no shockwaves alongside shared treatment (Table 2), so they are not all sham comparisons.
- Risk of bias: participants were not blinded in 14 trials and therapists in 16; fewer than half reported sequence generation or concealment; outcome assessors were blinded in six trials and explicitly unblinded in one. The authors nonetheless rated all trials medium or high quality (PEDro median 6, range 5–9). Evidence levels used a best-evidence synthesis, not GRADE.
- Eleven of the 19 trials were non-English (Chinese-journal) reports, and English titles were translated by software where missing; most outcomes were patient-reported and only five trials measured performance. Ten trials did not report funding. Additional files 1–10 (search strategies, funding, forest-plot data, adverse events) were not read.
Source-linked critical reading
clinical synthesis
Technical source record
Complete local PDF read: all article sections, Tables 1–5, Figures 1–6, funding and disclosures. Additional files 1–10 were not available and were not read.
Pooled averages favoured shockwaves for knee tendon and soft-tissue pain and function, but the trials differed greatly and some effects were unusually large. Later success and several conditions showed no clear difference. Against acupuncture or iontophoresis (two trials) and in one PRP-injection trial, the differences in pain were not statistically significant, and that PRP trial favoured PRP after six months.
In knee tendinopathies and other non-cartilage knee soft-tissue disorders, how do shockwaves compare with sham or other treatments for success, pain and function, and do wave type, energy or course length matter?
Systematic review and meta-analysis of randomized trials following PRISMA, searched to June 2018 without language limits (873 database plus 146 Google Scholar records; 54 full texts; 19 trials). Effects were pooled as odds ratios and standardized mean differences of change, with subgroup analyses and a best-evidence grading of evidence levels.
Nineteen randomized trials, 1,189 patients (562 shockwave, 627 control), mean age 34.7 years: patellar tendinopathy in eight, pes anserine tendinopathy in two, ACL injury or reconstruction in two, post-traumatic knee stiffness in two, and one each of Osgood–Schlatter disease, iliotibial band syndrome, infrapatellar fat pad injury, traumatic synovitis and popliteal cyamella. Many trials were Chinese-language reports.
Focused shockwaves in 8 trials (5 high-energy, 3 low-energy) and radial in 11 (5 high, 6 low), with high energy defined as ≥0.2 mJ/mm². Table 3 ranges: 0.03–0.58 mJ/mm², 1,000–4,600 impulses per session, 1 to 36 sessions over 1–12 weeks. Local anaesthesia was used only in one ACL reconstruction trial, where a single session was given during surgical anaesthesia.
Nine trials used sham or no shockwaves; others compared exercise, physiotherapy, medicines, iontophoresis, ultrasound diathermy, PRP, acupuncture or injection. Twelve trials used co-interventions (physiotherapy, acupuncture, exercise, manual therapy or medication), not always the same in both arms. In four "placebo" radial trials the control simply received no shockwaves.
Primary outcomes were treatment success (Likert or Roles–Maudsley ratings) and pain change on VAS; secondary outcomes were questionnaire-based function and performance tests (range of motion, vertical jump). Follow-up was grouped as immediate (≤1 month), short (>1–3 months), medium (>3–6 months) and long term (>6 months); 9 trials followed patients for 6 or 12 months.
Success OR 3.36 (95% CI 1.84–6.12; 16 trials, 842 patients), significant only at ≤1 month in the combined analysis (OR 3.09). Pain SMD −1.49 (−2.11 to −0.87; 18 trials, I²=95%), with similar results at ≤1, >1–3 and >3–6 months; the single focused trial beyond six months favoured the comparator (SMD 0.71). Function SMD 2.03 (1.09–2.96; 13 trials). Against invasive comparators (one focused trial against PRP injection, two radial trials against acupuncture or iontophoresis) pain did not differ significantly. Courses shorter than one month showed no significant pain effect.
The review reports no clinically relevant adverse events, side effects or severe complications in the included trials, but adverse events were not prespecified and their trial-level summary is in an unread supplement. The Conclusions instead mention "a few minor side effects", citing an elbow trial that is not among the 19 included, so minor effects may have gone unsummarized; not reported does not mean zero. The authors received no funding and declared no competing interests.
Heterogeneity reached I²=91–97% for pain and function, and pooled SMDs reached −5.17. Changes were imputed with an assumed correlation of 0.8. "Placebo" subgroups mixed true sham with no-shockwave controls. Several text statements conflict with Table 5. Blinding was poor in most trials, and harms were not prespecified. Supplements were not read. The text also claims a medium-term radial success effect that Table 5 does not contain, and mentions 20 trials in the Discussion.
Mostly patellar tendinopathy and other knee soft-tissue problems, often with shared rehabilitation. The pooled sizes should not be read as expected individual benefit; long-term success, comparisons with invasive treatments (acupuncture, iontophoresis and one 46-patient PRP-injection trial) and rare conditions remain uncertain.
Large blinded sham-controlled trials in single knee conditions, reporting change scores with SDs, harms and follow-up beyond six months; reviews should separate sham from no-treatment controls and recheck text against tables.
Read the original article.
Efficacy of extracorporeal shock wave therapy for knee tendinopathies and other soft tissue disorders: a meta-analysis of randomized controlled trials.
Liao CD, Xie GM, Tsauo JY, Chen HC, Liou TH. · BMC musculoskeletal disorders · 2018
Open the publication record or available full-text files.
Checked against the full paper · Clinical review: Dr. Marcus Yu Bin Pai, · How we review →