Plantar fasciitis: ranking focused intensities and radial shockwaves · Review · 2012
Which kind and strength of shockwave works best for heel pain?
In this review of 12 trials, focused shockwave at medium or high strength led to somewhat more successful results than sham at 3 to 6 months. Radial shockwave ranked first in the combined analysis, but that rests on two very different trials, and no device or strength clearly beat another.
Helped more than sham at medium or high strength- trials
- 12
- adults with heel pain
- 1,431
- kinds compared: low, medium and high-strength focused, and radial
- 4
Good to know: Combines direct and indirect comparisons between trials.
Chang KV, Chen SY, Chen WS, et al. Arch Phys Med Rehabil, 2012 · Read the paper ↗
How this study worked
What happened
- studies
- 2
- Compared with
- Sham shockwave; the different kinds and strengths were compared with each other indirectly
Chance of a successful result vs sham
Odds ratio · above 1 means more successes with shockwave
Show the numbers
| Odds ratio | likely range | |
|---|---|---|
| Low-strength focused | 2.36 | 0.5 to 11.14 |
| Medium-strength focused | 1.56 | 1.23 to 1.96 |
| High-strength focused | 1.43 | 1.07 to 1.91 |
| Radial | 2.72 | 0.74 to 10.07 |
What the trials used
- Intensity groups
- Low ≤0.08 mJ/mm² (Rompe 2002, Haake 2003); medium 0.08–0.28 (Speed 0.12, Rompe 2003 0.16, Marks 0.16, Ogden 0.22, Gollwitzer 0.25); high ≥0.28 (Theodore 0.36, Kudo 0.64; Liang 0.56 versus 0.12). Radial: Gerdesmeyer 0.16, 2,000 ×3; Ibrahim 0.17, 2,000 ×2.
- Local anaesthesia
- Used in Haake 2003 (low), Ogden (medium), Theodore and Kudo (high) and the high-intensity arm of Liang; not used in the radial trials or other medium-intensity trials.
- Sham methods
- Sham in 11 trials (minimal or zero-energy pulses, polyethylene foil, sound-reflecting pad, Styrofoam block, air or foam cushion, placebo handpiece, heel clasp); one trial compared high with medium focused intensity. Radial and focused treatments were compared only indirectly.
- Unit of analysis
- Liang 2007 compared 43 versus 34 heels, not people.
Who took part
Adults aged 25–87 with heel pain for over 3 months, in 12 trials up to 2011; one trial counted heels rather than people.
Side effects
The review does not report side effects.
All the numbers and the full appraisal
The results, in numbers.
Adults with heel pain at the medial calcaneal origin of the plantar fascia for over 3 months in 12 RCTs (1,431 participants; age 25–87 years; BMI 20–39; pain 6–312 months; baseline VAS 3.0–8.9). One trial counted heels rather than people.
Success, medium-intensity focused (direct)9 reported outcomes
Success, medium-intensity focused (direct)
Treatment success as defined by each trial (Table 1) odds of treatment success
Shockwaves versus placebo
5 studies · I²=0%; heterogeneity P=0.743 (Appendix 1; 0.938 in Figure 3)
Fixed-effect pairwise meta-analysis (Stata)
Logarithmic ratio scale.
In 5 trials, more patients reached success with medium-intensity focused shockwaves than with sham, with consistent results.
Analysis details and source
Pairwise interval labelled “95% CrI” in the Results text but “95% CI” in Figure 3 and Appendix 1; recorded here as a confidence interval. Figure 3 per-trial values labelled ORs match ratios of success proportions from Table 1 for most trials (reviewer and extractor observation), so the pairwise “ORs” may be risk ratios; two trials (Speed, Rompe 2003) match neither.
Success, high-intensity focused (direct)
Treatment success as defined by each trial (Table 1) odds of treatment success
Shockwaves versus placebo
2 studies · I²=58.8% (Appendix 1; 60.0% in Figure 3); heterogeneity P=0.119 (Appendix 1; 0.114 in Figure 3)
Fixed-effect pairwise meta-analysis (Stata)
Logarithmic ratio scale.
In 2 trials, more patients reached success with high-intensity focused shockwaves than with sham.
Analysis details and source
Pairwise interval labelled “95% CrI” in the Results text but “95% CI” in Figure 3 and Appendix 1; recorded here as a confidence interval. Figure 3 per-trial values labelled ORs match ratios of success proportions from Table 1 for most trials (reviewer and extractor observation), so the pairwise “ORs” may be risk ratios; two trials (Speed, Rompe 2003) match neither. A fixed-effect model was used despite I² near 60%; the Results text gives OR 1.42.
Success, low-intensity focused (direct)
Treatment success as defined by each trial (Table 1) odds of treatment success
Shockwaves versus placebo
2 studies · I²=91% (Appendix 1; 91.0% in Figure 3); heterogeneity P printed as 0.001 in Appendix 1 and p=0.001 in Figure 3
Random-effects pairwise meta-analysis (Stata)
Logarithmic ratio scale.
The two trials both favoured shockwaves but differed greatly in size of effect (Rompe 2002 5.49; Haake 2003 1.12), so the pooled interval is wide and includes 1.
Analysis details and source
Pairwise interval labelled “95% CrI” in the Results text but “95% CI” in Figure 3 and Appendix 1; recorded here as a confidence interval. Figure 3 per-trial values labelled ORs match ratios of success proportions from Table 1 for most trials (reviewer and extractor observation), so the pairwise “ORs” may be risk ratios; two trials (Speed, Rompe 2003) match neither.
Success, radial (direct)
Treatment success as defined by each trial (Table 1) odds of treatment success
Shockwaves versus placebo
2 studies · I²=89.5% (Appendix 1; 88.8% in Figure 3); heterogeneity P printed as 0.000 (Appendix 1; 0.003 in Figure 3)
Random-effects pairwise meta-analysis (Stata)
Logarithmic ratio scale.
Two radial trials differed greatly (Gerdesmeyer 2008 OR 1.48; Ibrahim 2010 OR 5.67), so the pooled interval is wide and includes 1.
Analysis details and source
Pairwise interval labelled “95% CrI” in the Results text but “95% CI” in Figure 3 and Appendix 1; recorded here as a confidence interval. Figure 3 per-trial values labelled ORs match ratios of success proportions from Table 1 for most trials (reviewer and extractor observation), so the pairwise “ORs” may be risk ratios; two trials (Speed, Rompe 2003) match neither. Ibrahim 2010 had 25/25 successes versus 4/25 in Table 1; a true odds ratio for these counts would be far larger than the 5.67 shown, which fits a ratio of success proportions.
Success, radial (network)
Treatment success as defined by each trial (Table 1) odds of treatment success
Shockwaves versus placebo
Bayesian random-effects network meta-analysis (WinBUGS)
Bayesian credible interval. Logarithmic ratio scale.
In the network model radial treatment had the largest odds of success versus placebo and an 82.7% probability of ranking best; it was the only treatment whose interval excluded 1. The very wide interval reflects two heterogeneous trials.
Analysis details and source
Table 3 header says “95% confidence intervals”; Methods and text describe Bayesian credible intervals; the text gives 2.13–202.4. No direct radial-versus-focused trial was included.
Success, low-intensity focused (network)
Treatment success as defined by each trial (Table 1) odds of treatment success
Shockwaves versus placebo
Bayesian random-effects network meta-analysis (WinBUGS)
Bayesian credible interval. Logarithmic ratio scale.
Point estimate above 1 with an interval including 1; probability of ranking best 12.3%.
Analysis details and source
Network estimate; interval type as for the radial network estimate.
Success, medium-intensity focused (network)
Treatment success as defined by each trial (Table 1) odds of treatment success
Shockwaves versus placebo
Bayesian random-effects network meta-analysis (WinBUGS)
Bayesian credible interval. Logarithmic ratio scale.
Interval includes 1 in the network model, although the direct pairwise estimate excluded 1; probability of ranking best 1.7%.
Analysis details and source
Network estimate; the abstract lists the ranking as low, medium, high, whereas the ORs and probabilities place high above medium.
Success, high-intensity focused (network)
Treatment success as defined by each trial (Table 1) odds of treatment success
Shockwaves versus placebo
Bayesian random-effects network meta-analysis (WinBUGS)
Bayesian credible interval. Logarithmic ratio scale.
Interval includes 1; probability of ranking best 3.1%.
Analysis details and source
Network estimate.
Success, radial versus medium focused (network)
Treatment success as defined by each trial (Table 1) odds of treatment success
Radial versus medium-intensity focused shockwaves
Bayesian random-effects network meta-analysis (WinBUGS)
Bayesian credible interval. Logarithmic ratio scale.
The indirect comparison of radial with medium-intensity focused treatment had a very wide interval including 1; radial versus low (20.56, 0.27–119.60) and versus high (32.45, 0.61–181.30) also included 1.
Analysis details and source
Indirect evidence only. Among focused levels: medium versus low 1.22 (0.05–6.20), high versus low 1.43 (0.03–7.60), high versus medium 1.48 (0.14–6.07).
Pain reduction, medium focused (direct)8 reported outcomes
Pain reduction, medium focused (direct)
Reduction in VAS after intervention standardized difference (labelled SMD in Figure 4; text and Appendix 1 say VAS mean difference); scale uncertain
Shockwaves versus placebo
2 studies · I²=39.5%; heterogeneity P=0.199
Fixed-effect pairwise meta-analysis (Stata)
In 2 trials (Rompe 2003, Marks 2008) pain fell more with medium-intensity focused treatment than with sham.
Analysis details and source
Negative values mean greater pain reduction with shockwaves. Figure 4 labels these values SMD and the Table 2 data agree with a standardized difference, so they should not be read as centimetres, although the text and Appendix 1 call them VAS mean differences; this inconsistency is unresolved.
Pain reduction, high focused (direct)
Reduction in VAS after intervention standardized difference (labelled SMD in Figure 4; text and Appendix 1 say VAS mean difference); scale uncertain
Shockwaves versus placebo
2 studies · I²=0%; heterogeneity P=0.66
Fixed-effect pairwise meta-analysis (Stata)
In 2 trials (Theodore 2004, Kudo 2006) the difference was small but favoured high-intensity treatment.
Analysis details and source
Negative values mean greater pain reduction with shockwaves. Figure 4 labels these values SMD and the Table 2 data agree with a standardized difference, so they should not be read as centimetres, although the text and Appendix 1 call them VAS mean differences; this inconsistency is unresolved.
Pain reduction, low focused (direct)
Reduction in VAS after intervention standardized difference (labelled SMD in Figure 4; text and Appendix 1 say VAS mean difference); scale uncertain
Shockwaves versus placebo
2 studies · I²=98.7%; heterogeneity P printed as 0.000
Random-effects pairwise meta-analysis (Stata)
Two very different trials; the interval includes no difference.
Analysis details and source
Negative values mean greater pain reduction with shockwaves. Figure 4 labels these values SMD and the Table 2 data agree with a standardized difference, so they should not be read as centimetres, although the text and Appendix 1 call them VAS mean differences; this inconsistency is unresolved. Rompe 2002 measured night pain (Table 2), unlike the other trials.
Pain reduction, radial (direct)
Reduction in VAS after intervention standardized difference (labelled SMD in Figure 4; text and Appendix 1 say VAS mean difference); scale uncertain
Shockwaves versus placebo
2 studies · I²=98.9%; heterogeneity P printed as 0.000
Random-effects pairwise meta-analysis (Stata)
Two radial trials gave very different results; the interval is extremely wide and includes no difference.
Analysis details and source
Negative values mean greater pain reduction with shockwaves. Figure 4 labels these values SMD and the Table 2 data agree with a standardized difference, so they should not be read as centimetres, although the text and Appendix 1 call them VAS mean differences; this inconsistency is unresolved. Ibrahim 2010 contributes a single-trial standardized difference of −20.55 (−24.70 to −16.41), implausibly large because its change SDs in Table 2 are 0.2 and 0.4. Results text gives the pooled upper limit as 9.83.
Pain reduction, radial (network)
Reduction in VAS after intervention standardized difference (labelled SMD in Figure 4; text and Appendix 1 say VAS mean difference); scale uncertain
Shockwaves versus placebo
Bayesian random-effects network meta-analysis (WinBUGS)
Bayesian credible interval.
Radial treatment had the largest network pain reduction and a 91.0% probability of ranking best; it was the only interval excluding zero. This rests on the same two trials, one with implausible values.
Analysis details and source
Bayesian network estimate versus placebo (Appendix 1, “compared with placebo” column). The network presumably uses the same trial data as Figure 4, which are labelled SMD, so the scale is uncertain (extractor interpretation; the paper does not name the input scale). The Results text quotes the other column (e.g. radial −6.08, −8.29 to −3.41), and the appendix header says 95% CI while the text says CrI.
Pain reduction, low focused (network)
Reduction in VAS after intervention standardized difference (labelled SMD in Figure 4; text and Appendix 1 say VAS mean difference); scale uncertain
Shockwaves versus placebo
Bayesian random-effects network meta-analysis (WinBUGS)
Bayesian credible interval.
Interval includes zero; probability of ranking best 2.8%.
Analysis details and source
Bayesian network estimate versus placebo (Appendix 1, “compared with placebo” column). The network presumably uses the same trial data as Figure 4, which are labelled SMD, so the scale is uncertain (extractor interpretation; the paper does not name the input scale). The Results text quotes the other column (e.g. radial −6.08, −8.29 to −3.41), and the appendix header says 95% CI while the text says CrI.
Pain reduction, medium focused (network)
Reduction in VAS after intervention standardized difference (labelled SMD in Figure 4; text and Appendix 1 say VAS mean difference); scale uncertain
Shockwaves versus placebo
Bayesian random-effects network meta-analysis (WinBUGS)
Bayesian credible interval.
Interval includes zero; probability of ranking best 4.2%.
Analysis details and source
Bayesian network estimate versus placebo (Appendix 1, “compared with placebo” column). The network presumably uses the same trial data as Figure 4, which are labelled SMD, so the scale is uncertain (extractor interpretation; the paper does not name the input scale). The Results text quotes the other column (e.g. radial −6.08, −8.29 to −3.41), and the appendix header says 95% CI while the text says CrI.
Pain reduction, high focused (network)
Reduction in VAS after intervention standardized difference (labelled SMD in Figure 4; text and Appendix 1 say VAS mean difference); scale uncertain
Shockwaves versus placebo
Bayesian random-effects network meta-analysis (WinBUGS)
Bayesian credible interval.
Interval includes zero; probability of ranking best 1.8%.
Analysis details and source
Bayesian network estimate versus placebo (Appendix 1, “compared with placebo” column). The network presumably uses the same trial data as Figure 4, which are labelled SMD, so the scale is uncertain (extractor interpretation; the paper does not name the input scale). The Results text quotes the other column (e.g. radial −6.08, −8.29 to −3.41), and the appendix header says 95% CI while the text says CrI.
Intensity and outcome (meta-regression)1 reported outcome
Intensity and outcome (meta-regression)
qualitative finding
Across 9 focused-versus-placebo trials (0.08–0.64 mJ/mm²), intensity was not related to success: coefficient 0.06 (95% CrI −1.96 to 2.10), P=.94. In 6 trials with VAS change, the coefficient was −2.73 (−15.42 to 9.95), P=.582, which the authors read as a tendency toward more pain relief at higher energy.
Analysis details and source
Both intervals include zero, so the dose–pain tendency described in the abstract and conclusion was not demonstrated; local anaesthesia use differed by intensity.
How treatment was delivered
Low ≤0.08 mJ/mm² (Rompe 2002, Haake 2003); medium 0.08–0.28 (Speed 0.12, Rompe 2003 0.16, Marks 0.16, Ogden 0.22, Gollwitzer 0.25); high ≥0.28 (Theodore 0.36, Kudo 0.64; Liang 0.56 versus 0.12). Radial: Gerdesmeyer 0.16, 2,000 ×3; Ibrahim 0.17, 2,000 ×2.
Used in Haake 2003 (low), Ogden (medium), Theodore and Kudo (high) and the high-intensity arm of Liang; not used in the radial trials or other medium-intensity trials.
Sham in 11 trials (minimal or zero-energy pulses, polyethylene foil, sound-reflecting pad, Styrofoam block, air or foam cushion, placebo handpiece, heel clasp); one trial compared high with medium focused intensity. Radial and focused treatments were compared only indirectly.
Liang 2007 compared 43 versus 34 heels, not people.
Not reported. The review did not extract or analyse adverse events, so it gives no information on the safety of any intensity or of radial treatment.
Radial evidence: only 2 trials; Ibrahim 2010 reports 25/25 successes and contributes a standardized pain difference of −20.55, implausibly large because its change SDs are 0.2 and 0.4. Rankings favouring radial treatment depend on these data.
Reporting inconsistencies: pairwise intervals called CrI in the text but CI in figures; Table 3 header says CI for Bayesian estimates; Figure 4 labels SMD for values the text calls VAS differences; Figure 3 per-trial “ORs” mostly match ratios of success proportions (possible risk ratios; Speed and Rompe 2003 match neither); text versus table values differ (1.42/1.43; 2.13/2.14; 9.83/9.53; VAS network values from a different column); abstract ranking order differs from results; abstract says one reviewer extracted and another checked, Methods say two extracted independently; “trial duration 3–60 wk” versus 3–60 months in Table 1.
Comparability: success definitions varied (Roles and Maudsley, >50% or >60% pain decrease, Likert scale); Rompe 2002 analysed night pain; local anaesthesia clustered in medium and high intensity; a fixed-effect model was used for high intensity despite I² near 60%; Jadad scores of included trials are not shown. Malay 2006 was excluded for lacking a fixed intensity range.
Conclusions beyond the data: the authors recommend radial therapy “because of its lower price and likely better effectiveness”, but cost was not analysed and every radial-versus-focused interval included 1; meta-regression intervals included zero for pain and success. Author-reported limits: varied success criteria and the difference between traditional and network results.
Funding: National Science Council of Taiwan and an International Joint Grant from the Royal Society (UK) and the National Science Council; no commercial party with a financial interest. Search limited to MEDLINE and PubMed (January 1996 to June 2011) plus Cochrane, ClinicalTrials.gov and bibliographies.
Findings checked against the full source by a separate automated reviewer.
Limitations noted by the reviewers
- Radial evidence: only 2 trials; Ibrahim 2010 reports 25/25 successes and contributes a standardized pain difference of −20.55, implausibly large because its change SDs are 0.2 and 0.4. Rankings favouring radial treatment depend on these data.
- Reporting inconsistencies: pairwise intervals called CrI in the text but CI in figures; Table 3 header says CI for Bayesian estimates; Figure 4 labels SMD for values the text calls VAS differences; Figure 3 per-trial “ORs” mostly match ratios of success proportions (possible risk ratios; Speed and Rompe 2003 match neither); text versus table values differ (1.42/1.43; 2.13/2.14; 9.83/9.53; VAS network values from a different column); abstract ranking order differs from results; abstract says one reviewer extracted and another checked, Methods say two extracted independently; “trial duration 3–60 wk” versus 3–60 months in Table 1.
- Comparability: success definitions varied (Roles and Maudsley, >50% or >60% pain decrease, Likert scale); Rompe 2002 analysed night pain; local anaesthesia clustered in medium and high intensity; a fixed-effect model was used for high intensity despite I² near 60%; Jadad scores of included trials are not shown. Malay 2006 was excluded for lacking a fixed intensity range.
- Conclusions beyond the data: the authors recommend radial therapy “because of its lower price and likely better effectiveness”, but cost was not analysed and every radial-versus-focused interval included 1; meta-regression intervals included zero for pain and success. Author-reported limits: varied success criteria and the difference between traditional and network results.
- Funding: National Science Council of Taiwan and an International Joint Grant from the Royal Society (UK) and the National Science Council; no commercial party with a financial interest. Search limited to MEDLINE and PubMed (January 1996 to June 2011) plus Cochrane, ClinicalTrials.gov and bibliographies.
Source-linked critical reading
Systematic review with network meta-analysis
Technical source record
Complete local PDF read (all 10 pages): methods, results, Tables 1–3, Figures 1–5, the printed Appendix 1, discussion, limitations, conclusions, funding footnotes and references. No online supplement is referenced.
In sham-controlled trials of plantar heel pain lasting over 3 months, medium- and high-intensity focused shockwaves led to somewhat more treatment successes than sham at 3 to 6 months. The network analysis ranked radial shockwaves first, but that ranking rests on two very different trials, one of them small (Ibrahim 2010, 50 people), and no comparison between radial and focused treatment was clearly different. Side effects were not reviewed.
In plantar fasciitis, how do focused shockwaves of low, medium and high intensity and radial shockwaves compare with sham, and with each other, for treatment success and pain reduction?
Systematic review with pairwise meta-analysis (Stata) and Bayesian random-effects network meta-analysis (WinBUGS, three chains, 50,000 burn-in and 50,000 iterations). MEDLINE and PubMed January 1996 to June 2011 without language limits, plus Cochrane, ClinicalTrials.gov and bibliographies. RCTs with Jadad ≥3 and a fixed intensity range; 12 of 25 assessed trials included. Outcomes closest to 6 months.
Adults with heel pain at the medial calcaneal origin of the plantar fascia for over 3 months in 12 RCTs (1,431 participants; age 25–87 years; BMI 20–39; pain 6–312 months; baseline VAS 3.0–8.9). One trial counted heels rather than people.
Focused shockwaves grouped by energy flux density (low ≤0.08, medium 0.08–0.28, high ≥0.28 mJ/mm²; trials 0.08–0.64 mJ/mm²) and radial shockwaves (0.16 and 0.17 mJ/mm², 2,000 pulses, 2–3 sessions). Local anaesthesia was used in 5 trials.
Sham in 11 trials (minimal or zero-energy pulses, polyethylene foil, sound-reflecting pad, Styrofoam block, air or foam cushion, placebo handpiece, heel clasp); one trial compared high with medium focused intensity. Radial and focused treatments were compared only indirectly.
Primary: treatment success as defined by each trial (Roles and Maudsley grades, >50% or >60% pain decrease, Likert scale), reported at 3 or 6 months, as odds ratios. Secondary: reduction in VAS, available from 9 trials. Meta-regression of intensity against both outcomes. Adverse events were not assessed.
Direct comparisons with sham: success OR medium 1.56 (95% CI 1.23–1.96, 5 trials) and high 1.43 (1.07–1.91, 2 trials); low 2.36 (0.50–11.14) and radial 2.72 (0.74–10.07) had intervals including 1. Pain reduction, as standardized differences (Figure 4 labels SMD; the text says VAS), favoured medium (−1.21, −1.76 to −0.67) and high (−0.33, −0.58 to −0.08) intensity. Network: radial success OR 38.26 (95% CrI 2.14–202.40), probability best 82.7%; low 6.15, medium 2.51 and high 2.87, all including 1. Network pain difference versus placebo, on the same uncertain scale, was −6.15 for radial (−8.828 to −3.505), with focused levels including zero. All radial-versus-focused and between-intensity intervals included 1; meta-regression intervals included zero.
Not reported. The review did not extract or analyse adverse events, so it gives no information on the safety of any intensity or of radial treatment.
Radial results rest on two heterogeneous trials, one (Ibrahim 2010) with implausibly small SDs that inflate its standardized pain difference to −20.55. Interval types are mislabelled (CI versus CrI); Figure 4 labels pain differences as SMD while the text calls them VAS differences; Figure 3 per-trial “ORs” mostly match ratios of success proportions; and several numbers differ between text, tables and appendix. Success definitions varied, local anaesthesia clustered in the higher-intensity groups, and one trial counted heels. The recommendation of radial therapy for lower cost and likely better effectiveness goes beyond the data: cost was not analysed and no radial-versus-focused interval excluded 1. Harms were not reviewed. Public Taiwanese and UK funding; no commercial interest.
Applies to chronic plantar heel pain treated in trials up to 2011. The direct comparisons support modest success gains with medium- or high-intensity focused shockwaves versus sham; the network ranking does not establish that radial devices or any intensity are better, and the review is silent on safety.
Direct randomized comparisons of radial and focused devices and of intensity levels without differential anaesthesia, with verified variance data, harms and function outcomes.
Read the original article.
Comparative effectiveness of focused shock wave therapy of different intensity levels and radial shock wave therapy for treating plantar fasciitis: a systematic review and network meta-analysis.
Chang KV, Chen SY, Chen WS, Tu YK, Chien KL. · Arch Phys Med Rehabil · 2012
Open the publication record or available full-text files.
Checked against the full paper · Clinical review: Dr. Marcus Yu Bin Pai, · How we review →