Knee osteoarthritis, Kellgren–Lawrence grade II–III · Randomized trial · 2013
Does shockwave help knee osteoarthritis pain?
In this trial, four weekly shockwave sessions brought pain on movement down from about 7.6 to 3.8 out of 10 by 12 weeks. With placebo, it fell from about 7.6 to 6.4. Knee function scores also improved more with shockwave.
Helped more than placebo- adults with grade II–III knee osteoarthritis
- 70
- sessions, one a week
- 4
- weeks of follow-up
- 12
Good to know: The report has some inconsistencies.
Zhao Z, Jing R, Shi Z, et al. The Journal of Surgical Research, 2013 · Read the paper ↗
How this study worked
What happened
Knee pain on movement
0–10 pain scale · start and after 12 weeks
StartAfter 12 weeks
Show the numbers
| Group | People | Start | After 12 weeks |
|---|---|---|---|
| Shockwave | 34 | 7.56 | 3.83 |
| Placebo | 36 | 7.55 | 6.41 |
The treatment
Shockwave
Shockwave, once a week for 4 weeks
- Sessions
- 4, one a week
- Pulses
- 4,000 per session
- Strength
- 0.25mJ/mm² · 6 per second
- Numbing
- Not mentioned
- Device
- Swiss DolorClast
- Everyone also got
- No other treatments; low activity advised for 48 hours after each session
Compared with: Placebo: the same device and procedure at zero energy
Settings are what each study used, not a prescription.
Who took part
70 adults aged 45 or older with grade II–III knee osteoarthritis and knee pain for at least 3 months, from two rheumatology clinics.
Side effects
Brief skin redness and swelling lasting a few days, mentioned without counts.
All the numbers and the full appraisal
The results, in numbers.
Adults aged 45 or older from two rheumatology clinics with symptomatic primary knee osteoarthritis by American College of Rheumatology criteria, knee pain for the previous 3 months and Kellgren–Lawrence grade II or III; one target knee per person. Mean ages 61.8 (placebo) and 59.9 years (shockwave); 25 of 36 and 20 of 34 were women; 26 per group had grade II. Baseline pain on walking was 7.6 on a 10-cm VAS in both groups. Spinal stenosis, neurological disease, inflammatory or metabolic arthritis, and knee surgery or injection in the previous 6 months were exclusions.
People, one target knee each. 34 in the intent-to-treat analysis; 30 completed 12 weeks (2 lost to follow-up or work commitments, 2 lack of efficacy). Imputation method not described.
People, one target knee each. 36 in the intent-to-treat analysis; 31 completed 12 weeks (2 increased pain and analgesic use, labelled adverse events in Figure 1; 3 lack of efficacy). Imputation method not described.
10-cm visual analog scale (VAS), primary outcome1 reported outcome
Pain on movement
10-cm visual analog scale (VAS), primary outcome cm (0–10)
| Group | n | Baseline | Follow-up |
|---|---|---|---|
| Shockwave therapy | 34 | 7.56 | 3.83 |
| Placebo at 0 mJ/mm² | 36 | 7.55 | 6.41 |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
Shockwave versus placebo, decrease in VAS at each period, intent-to-treat
Independent-samples t test or Mann–Whitney test on change from baseline (not specified)
Mean VAS fell 3.73 points with shockwaves (7.56 to 3.83) and 1.14 points with placebo (7.55 to 6.41); both within-group changes P<0.01. The between-group difference in decrease was reported as P<0.01 without an interval.
Analysis details and source
Means from the Results text in the intent-to-treat population; no SD or confidence interval given at 12 weeks. Table 1 reports baseline 'pain on walking' as 7.6 ± 1.1 (placebo) and 7.6 ± 1.3 (shockwave), whereas the primary outcome is called pain on movement. The Discussion's statement that baseline pain was better with shockwaves contradicts these values.
10-cm visual analog scale (VAS)1 reported outcome
Pain on movement, early visits
10-cm visual analog scale (VAS) cm (0–10)
The paper reports a greater VAS decrease with shockwaves than with placebo at each assessment (P<0.01). Values at 1 and 4 weeks are shown only as plotted mean changes in Figure 2.
Analysis details and source
Figure values were not extracted as numbers; no table reports them.
Lequesne index2 reported outcomes
Lequesne index
Lequesne index Points
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Shockwave therapy | 34 | — | -4.1 [CI -4.9; -3.3] |
| Placebo at 0 mJ/mm² | 36 | — | -2 [CI -2.9; -1] |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
Shockwave minus placebo
No adjustment reported
Disability on the Lequesne index fell more with shockwaves than with placebo over 12 weeks: difference −2.1 points (95% CI −3.4 to −0.9).
Analysis details and source
Mean change with 95% confidence interval as printed in Table 2; negative values mean improvement. Test was an independent-samples t test or Mann–Whitney test (which one is not specified per outcome). The intent-to-treat columns use 36 and 34 although nine patients withdrew; imputation is not described. Baseline 10.2 ± 2.3 (shockwave) and 10.1 ± 2.4 (placebo); maximum score 24.
Lequesne index, completers
Lequesne index Points
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Shockwave therapy | 30 | — | -4.6 [CI -5.4; -3.9] |
| Placebo at 0 mJ/mm² | 31 | — | -2.3 [CI -3.4; -1.2] |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
Shockwave minus placebo
No adjustment reported
Among completers the Lequesne difference was −2.3 points (95% CI −3.6 to −1.1), favoring shockwaves.
Analysis details and source
Mean change with 95% confidence interval as printed in Table 2; negative values mean improvement. Test was an independent-samples t test or Mann–Whitney test (which one is not specified per outcome). Completers only (31 placebo, 30 shockwave); exclusion of withdrawals can bias the comparison.
WOMAC total2 reported outcomes
WOMAC total
WOMAC total Points
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Shockwave therapy | 34 | — | -19.1 [CI -22.7; -15.6] |
| Placebo at 0 mJ/mm² | 36 | — | -8.5 [CI -12.4; -4.6] |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
Shockwave minus placebo
No adjustment reported
WOMAC total fell more with shockwaves than with placebo: difference −10.6 points (95% CI −15.8 to −5.4) on a 0–96 scale.
Analysis details and source
Mean change with 95% confidence interval as printed in Table 2; negative values mean improvement. Test was an independent-samples t test or Mann–Whitney test (which one is not specified per outcome). The intent-to-treat columns use 36 and 34 although nine patients withdrew; imputation is not described. Baseline 36.4 ± 10.3 (shockwave) and 32.8 ± 10.9 (placebo).
WOMAC total, completers
WOMAC total Points
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Shockwave therapy | 30 | — | -21.5 [CI -24.4; -18.5] |
| Placebo at 0 mJ/mm² | 31 | — | -10.3 [CI -14.3; -5.8] |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
Shockwave minus placebo
No adjustment reported
Among completers the WOMAC total difference was −11.4 points (95% CI −16.6 to −6.3), favoring shockwaves.
Analysis details and source
Mean change with 95% confidence interval as printed in Table 2; negative values mean improvement. Test was an independent-samples t test or Mann–Whitney test (which one is not specified per outcome). Completers only (31 placebo, 30 shockwave); exclusion of withdrawals can bias the comparison.
WOMAC pain1 reported outcome
WOMAC pain
WOMAC pain subscale Points
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Shockwave therapy | 34 | — | -4.5 [CI -5.4; -3.6] |
| Placebo at 0 mJ/mm² | 36 | — | -2.2 [CI -3.2; -1.2] |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
Shockwave minus placebo
No adjustment reported
WOMAC pain fell more with shockwaves: difference −2.3 points (95% CI −3.6 to −1.0).
Analysis details and source
Mean change with 95% confidence interval as printed in Table 2; negative values mean improvement. Test was an independent-samples t test or Mann–Whitney test (which one is not specified per outcome). The intent-to-treat columns use 36 and 34 although nine patients withdrew; imputation is not described. Per-protocol difference −2.0 (−3.4 to −0.5).
WOMAC function1 reported outcome
WOMAC function
WOMAC function subscale Points
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Shockwave therapy | 34 | — | -13.9 [CI -17.2; -10.6] |
| Placebo at 0 mJ/mm² | 36 | — | -6 [CI -9.2; -2.8] |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
Shockwave minus placebo
No adjustment reported
WOMAC function improved more with shockwaves: difference −7.9 points (95% CI −12.5 to −3.4).
Analysis details and source
Mean change with 95% confidence interval as printed in Table 2; negative values mean improvement. Test was an independent-samples t test or Mann–Whitney test (which one is not specified per outcome). The intent-to-treat columns use 36 and 34 although nine patients withdrew; imputation is not described. Per-protocol difference −9.0 (−13.3 to −4.8). Baseline function 25.7 ± 8.9 (shockwave) versus 22.3 ± 9.4 (placebo).
WOMAC stiffness1 reported outcome
WOMAC stiffness
WOMAC stiffness subscale Points
| Group | n | Baseline | Reported change |
|---|---|---|---|
| Shockwave therapy | 34 | — | -0.7 [CI -1; -0.4] |
| Placebo at 0 mJ/mm² | 36 | — | -0.3 [CI -0.8; 0.2] |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
Shockwave minus placebo
No adjustment reported
Stiffness did not differ between groups: difference −0.4 points (95% CI −1.0 to 0.2), an interval that includes no difference.
Analysis details and source
Mean change with 95% confidence interval as printed in Table 2; negative values mean improvement. Test was an independent-samples t test or Mann–Whitney test (which one is not specified per outcome). The intent-to-treat columns use 36 and 34 although nine patients withdrew; imputation is not described. Per-protocol difference −0.4 (−1.1 to 0.3), also P>0.05. A nonsignificant difference does not show equivalence.
Patient perception of clinical severity, 5-point Likert scale1 reported outcome
Patient-rated condition
Patient perception of clinical severity, 5-point Likert scale Points (1 very poor to 5 very good)
| Group | n | Baseline | Follow-up |
|---|---|---|---|
| Shockwave therapy | — | 3.09 (SD 0.67) | 3.94 (SD 0.92) |
| Placebo at 0 mJ/mm² | — | 3.11 (SD 0.67) | 3.42 (SD 0.81) |
SD: standard deviation. CI: confidence interval. IQR: interquartile range. —: not extracted.
Shockwave versus placebo, change in patient perception at 12 weeks
Independent-samples t test or Mann–Whitney test (not specified)
Patients' rating of their knee condition rose from 3.09 to 3.94 with shockwaves and from 3.11 to 3.42 with placebo; the paper reports a better rating with shockwaves (P<0.01).
Analysis details and source
Means ± SD from the Results text; the number analysed is not stated for these values. Both the Results text and Table 2 give the changes as −0.3 and −0.9 (difference −0.6, 95% CI −0.9 to −0.2), although the reported means rose on a scale where higher is better; the sign of the change conflicts with the means. The per-protocol placebo interval (−0.3, −0.6 to −0.8) does not contain its estimate.
Participant flow1 reported outcome
Withdrawal by 12 weeks
Participant flow People
Dropout rate, shockwave versus placebo
Not specified
Four of 34 shockwave and 5 of 36 placebo patients withdrew; the dropout rate did not differ (P>0.05).
Analysis details and source
Reasons: shockwave, 2 lack of efficacy and 2 work commitments (text) or lost to follow-up (Figure 1); placebo, 3 lack of efficacy and 2 increased pain with analgesic use (text) or adverse events (Figure 1).
How treatment was delivered
Swiss DolorClast, Electro Medical Systems (Nyon, Switzerland). Radial or focused modality not stated in the paper.
4,000 pulses at an energy flux density of 0.25 mJ/mm², frequency 6 Hz.
Four sessions at weekly intervals, given by a technologist.
Supine, knee straight or flexed to 90°. Probe held stationary on a trigger point around the knee or at the patellofemoral and tibiofemoral borders, avoiding the peroneal nerve and vessels; aqueous coupling gel applied in circular motions.
Anesthesia not reported. No bed rest; low physical activity recommended for 48 hours. Evaluations done 1 day after treatment.
Same device, same area and manner at 0 mJ/mm²; patients could not see the device parameters. Blinding success not assessed.
Patients were not offered other therapies, except the nine who withdrew.
The abstract states that no adverse events occurred during or after shockwave treatment, but the Discussion reports transient skin reddening after treatment and swelling for several days, without counts. In the flow chart two placebo withdrawals are labelled adverse events, while the text attributes them to increased pain and analgesic use. Systematic adverse-event collection is not described. Funded by the National Natural Science Foundation of China and the China Postdoctoral Science Foundation; the authors declare no conflict of interest.
Author-reported: dose, intensity and frequency were not studied; parameters came from the authors' experience; long-term effects need study; no untreated group, so the authors cannot exclude the effect of placebo itself.
Extractor assessment: allocation concealment is unclear (block list, yet assignment by odd and even numbers), blinding success was not checked, the intent-to-treat analysis keeps 70 patients despite nine withdrawals without describing imputation, the test used per outcome is unspecified, and no registration is reported.
Internal inconsistencies: the Discussion says baseline pain was better with shockwaves, but Table 1 shows 7.6 in both groups; the Results text and Table 2 give negative patient-perception changes although reported means rose on a higher-is-better scale; the per-protocol placebo perception interval (−0.3, −0.6 to −0.8) excludes its estimate; the abstract reports no adverse events while the Discussion reports transient skin reddening and swelling; withdrawal reasons differ between text and Figure 1.
Findings checked against the full source by a separate automated reviewer.
Limitations noted by the reviewers
- Author-reported: dose, intensity and frequency were not studied; parameters came from the authors' experience; long-term effects need study; no untreated group, so the authors cannot exclude the effect of placebo itself.
- Extractor assessment: allocation concealment is unclear (block list, yet assignment by odd and even numbers), blinding success was not checked, the intent-to-treat analysis keeps 70 patients despite nine withdrawals without describing imputation, the test used per outcome is unspecified, and no registration is reported.
- Internal inconsistencies: the Discussion says baseline pain was better with shockwaves, but Table 1 shows 7.6 in both groups; the Results text and Table 2 give negative patient-perception changes although reported means rose on a higher-is-better scale; the per-protocol placebo perception interval (−0.3, −0.6 to −0.8) excludes its estimate; the abstract reports no adverse events while the Discussion reports transient skin reddening and swelling; withdrawal reasons differ between text and Figure 1.
Source-linked critical reading
Single-center, single-blinded, placebo-controlled randomized trial in China. 70 of 186 screened patients randomized, 36 placebo and 34 shockwave; 31 and 30 completed 12 weeks. Allocation concealment not described; no registration reported.
Technical source record
Full local PDF read; All 6 PDF pages (journal pp. 661–666); Abstract; Introduction; Materials and methods; Results; Discussion; Conclusions; Table 1; Table 2; Figure 1 flow chart; Figures 2–3 captions and plots (plotted values not extracted); Acknowledgment (funding); conflict of interest statement; references
In this trial of 70 people with grade II–III knee osteoarthritis, four weekly shockwave sessions were followed by less pain on movement and better knee function than placebo over 12 weeks. It is a single-center study with some reporting inconsistencies and no follow-up beyond 12 weeks.
In adults with Kellgren–Lawrence grade II–III knee osteoarthritis, do four weekly shockwave sessions reduce pain on movement and improve knee function over 12 weeks more than a placebo application at zero energy?
Single-center, single-blinded, placebo-controlled randomized trial in Beijing, China (July 2011 to February 2012). Of 186 people screened, 70 were randomized: 36 to placebo and 34 to shockwaves. Five placebo and four shockwave patients withdrew; 31 and 30 completed 12 weeks. Allocation used a computer-generated block list, but the paper also describes assignment by odd and even numbers and does not describe allocation concealment. No trial registration is reported.
Adults aged 45 or older from two rheumatology clinics with symptomatic primary knee osteoarthritis by American College of Rheumatology criteria, knee pain for the previous 3 months and Kellgren–Lawrence grade II or III; one target knee per person. Mean ages 61.8 (placebo) and 59.9 years (shockwave); 25 of 36 and 20 of 34 were women; 26 per group had grade II. Baseline pain on walking was 7.6 on a 10-cm VAS in both groups. Spinal stenosis, neurological disease, inflammatory or metabolic arthritis, and knee surgery or injection in the previous 6 months were exclusions.
Swiss DolorClast device (EMS, Nyon, Switzerland), applied by a technologist: 4,000 pulses at 0.25 mJ/mm² and 6 Hz per session, once a week for 4 weeks. The probe was held stationary on a trigger point around the knee or at the patellofemoral and tibiofemoral borders, avoiding the peroneal nerve and vessels, with coupling gel. Low physical activity was recommended for 48 hours. The paper does not state whether the waves were radial or focused and does not report anesthesia.
Placebo: the same device applied to the same area in the same manner at 0 mJ/mm². Patients could not see the device settings, and a physician not involved in selection or treatment assessed outcomes. Whether patients could tell the treatments apart was not tested. No other therapies were offered, except to patients who withdrew.
Primary: pain on movement on a 10-cm VAS (0 no pain, 10 maximal pain). Secondary: Lequesne index (maximum 24, higher worse), WOMAC (0 best to 96 worst, with pain, stiffness and function subscales) and patient perception of severity on a 5-point scale (1 very poor to 5 very good). Assessments at baseline and at 1, 4 and 12 weeks, one day after the sessions. Table 2 reports 12-week mean changes with 95% confidence intervals for the intent-to-treat and per-protocol populations.
At 12 weeks mean pain on movement fell from 7.56 to 3.83 with shockwaves and from 7.55 to 6.41 with placebo; the paper reports a greater decrease with shockwaves at 1, 4 and 12 weeks (P<0.01) but gives no between-group interval or 12-week SD for VAS. In the intent-to-treat table, differences in 12-week change versus placebo were −2.1 Lequesne points (95% CI −3.4 to −0.9), −10.6 WOMAC total points (−15.8 to −5.4), −2.3 WOMAC pain (−3.6 to −1.0) and −7.9 WOMAC function (−12.5 to −3.4), all favoring shockwaves. WOMAC stiffness did not differ (−0.4, −1.0 to 0.2). Per-protocol results were similar. Patient-rated condition rose from 3.09 to 3.94 with shockwaves and from 3.11 to 3.42 with placebo (P<0.01).
The abstract states that no adverse events occurred during or after shockwave treatment, but the Discussion reports transient skin reddening after treatment and swelling for several days, without counts. In the flow chart two placebo withdrawals are labelled adverse events, while the text attributes them to increased pain and analgesic use. Systematic adverse-event collection is not described. Funded by the National Natural Science Foundation of China and the China Postdoctoral Science Foundation; the authors declare no conflict of interest.
Single center, modest size and 12-week follow-up. Allocation concealment is unclear and patient blinding was not checked. Table 2 analyses all 70 patients as intent-to-treat although nine withdrew, and the handling of missing data is not described. The paper has internal inconsistencies: the Discussion says baseline pain was better with shockwaves although Table 1 shows 7.6 in both groups; the Results text and Table 2 give negative changes in patient perception although the reported mean scores rose and higher scores mean better; one per-protocol interval (−0.3, −0.6 to −0.8) does not contain its estimate; withdrawal reasons differ between the text and Figure 1. VAS values at 1 and 4 weeks appear only in Figure 2. The authors note that dose was not studied and that a no-treatment group is lacking.
Applies to adults aged 45 or older with grade II–III knee osteoarthritis, treated with this device and dose without other therapies, over 12 weeks. It does not address grade IV arthritis, effects beyond 12 weeks, other energy levels, or comparison with exercise, medication or injections.
A registered multicenter trial with concealed allocation, a checked sham, prespecified handling of missing data, follow-up of at least 6 to 12 months, systematic adverse-event recording, and comparison with or addition to exercise-based care.
Read the original article.
Efficacy of extracorporeal shockwave therapy for knee osteoarthritis: a randomized controlled trial.
Zhao Z, Jing R, Shi Z, Zhao B, Ai Q, Xing G. · The Journal of surgical research · 2013
Open the publication record or available full-text files.
Checked against the full paper · Clinical review: Dr. Marcus Yu Bin Pai, · How we review →