Plantar heel pain: a small pooled difference in morning pain · Review · 2005
How much does shockwave ease morning heel pain compared with placebo?
In this review of 11 trials, morning heel pain at about 12 weeks was slightly lower with shockwave than with placebo or a very low dose: less than half a point on a 0–10 scale, across 6 trials. In the 4 better-run trials the difference was smaller and could be due to chance.
Close to placebo, under half a point apart- trials
- 11
- people with heel pain
- 1,290
- trials combined for morning pain
- 6
Good to know: Device makers supported 3 of the 11 trials.
Thomson CE, Crawford F, Murray GD BMC Musculoskelet Disord, 2005 · Read the paper ↗
How this study worked
What happened
- studies
- 6
- Compared with
- Placebo shockwave or a very low dose
Morning heel pain vs placebo
Difference on a 0–10 pain scale · above zero means less pain with shockwave
Show the numbers
| Difference (points on a 0–10 pain scale) | likely range | |
|---|---|---|
| All 6 trials | +0.42 | 0.02 to 0.83 |
| 4 better-run trials | +0.21 | −0.29 to 0.7 |
What the trials used
- Pooled trial regimens
- Abt: Ossatron 1,000 ×2, 0.08 mJ/mm²; Buch: Epos Ultra 3,800 total, 0.03–0.36 mJ/mm²; Buchbinder: Epos Ultra 2,000–2,500 ×3, 0.02–0.33 mJ/mm²; Haake: Epos Ultra 4,000 ×3, 0.08 mJ/mm²; Ogden: Ossatron 1,500 total, 0.22 mJ/mm²; Speed: Sonocur Plus 1,500 ×3, 0.06 mJ/mm².
- Controls
- Pooled trials: absorbent block (Abt), absorbent foil (Buch), 300 versus 6,000–7,500 impulses (Buchbinder), polythene foil barrier (Haake), Styrofoam block (Ogden), focus outside the patient (Speed).
- Local anaesthesia and guidance
- Local anaesthetic in both groups: Abt, Buch, Haake, Ogden (different for placebo); none in Buchbinder or Speed. Ultrasound guidance in Buch, Buchbinder and Speed, not in Ogden, not stated in Abt and Haake.
Who took part
1,290 adults with heel pain, usually for over 6 months, in 11 trials up to 2003; some trials enrolled runners or people with heel spurs.
Side effects
Mostly mild redness, swelling, numbness or pain; one person tore the plantar fascia after treatment, after many earlier steroid injections.
All the numbers and the full appraisal
The results, in numbers.
Adults with plantar heel pain in 11 RCTs published 1996–2003 (1,290 patients); 6 trials pooled for morning pain at about 12 weeks. Pain lasted over 6 months in 10 trials; some trials enrolled runners, people with heel spurs or people being considered for surgery.
Morning (first-step) heel pain4 reported outcomes
Morning (first-step) heel pain
10 cm VAS, morning (first-step) pain cm on a 10-cm VAS
Shockwaves versus placebo or low-dose shockwaves
6 studies · Chi²=8.92, df=5, P=0.11; I²=43.9%
Fixed-effect weighted mean difference (RevMan)
Pooled across 6 trials, the difference favoured shockwaves by 0.42 cm on a 10-cm scale; the authors considered this less than half a centimetre and not clinically meaningful. Individual trials ranged from −0.50 (Buchbinder) to 2.00 (Abt).
Analysis details and source
Positive values favour shockwaves per the Figure 2 axis. Figure 2 totals are 436 and 445 analysed, whereas the abstract states n=897; the participant count is left blank here. The figure does not state whether the means are final scores or improvements, and Haake values differ from Table 5. Abt SDs were derived from a p-value.
Morning pain, excluding Abt (imputed SDs)
10 cm VAS, morning (first-step) pain cm on a 10-cm VAS
Shockwaves versus control, Abt excluded
5 studies
Fixed-effect weighted mean difference (RevMan)
Without the only trial with imputed variance, the pooled difference was 0.30 cm with an interval including zero.
Analysis details and source
Sensitivity analysis on 5 trials; p-value not reported.
Morning pain, higher-quality trials
10 cm VAS, morning (first-step) pain cm on a 10-cm VAS
Shockwaves versus control, quality score ≥4
4 studies
Fixed-effect weighted mean difference (RevMan)
The four trials scoring 4 or more on the 6-point quality scale (Buchbinder, Haake, Speed, Buch) showed a small difference with an interval including zero.
Analysis details and source
Quality subgroup; fixed-effect model.
Morning pain, lower-quality trials
10 cm VAS, morning (first-step) pain qualitative finding
The two lower-quality trials (Abt and Ogden) gave a weighted mean difference reported as −0.90 (95% CI −1.62 to −0.19), p=0.01, “in favour of active treatment”.
Analysis details and source
Kept qualitative because the reported sign is opposite to the main analysis, where positive values favour shockwaves. The text calls these trials “less than four”, “less than three” and “three or above” in different places; both scored 3 in Table 1.
Secondary outcomes (not pooled)1 reported outcome
Secondary outcomes (not pooled)
qualitative finding
Walking, pressure, night and rest pain, Roles and Maudsley, AOFAS and Maryland foot scores, SF-12 and SF-36 could not be pooled because measures differed. Six trials reported a favourable walking-pain result; pain at rest favoured shockwaves in four of five trials; the authors describe the other outcomes as equivocal.
Analysis details and source
Narrative summary from Table 5; vote counting of p-values, not an effect estimate.
Trials not pooled1 reported outcome
Trials not pooled
qualitative finding
Five trials (about 310 patients) were not pooled for lack of usable data, different timing or outcomes. Rompe 2003 in running athletes reported a morning-pain difference of 2.60 (95% CI 1.37 to 3.83) at 6 months.
Analysis details and source
The authors suggest large effects in the two smallest trials may reflect chance, small-study bias, failed blinding or true subgroup effects.
How treatment was delivered
Abt: Ossatron 1,000 ×2, 0.08 mJ/mm²; Buch: Epos Ultra 3,800 total, 0.03–0.36 mJ/mm²; Buchbinder: Epos Ultra 2,000–2,500 ×3, 0.02–0.33 mJ/mm²; Haake: Epos Ultra 4,000 ×3, 0.08 mJ/mm²; Ogden: Ossatron 1,500 total, 0.22 mJ/mm²; Speed: Sonocur Plus 1,500 ×3, 0.06 mJ/mm².
Pooled trials: absorbent block (Abt), absorbent foil (Buch), 300 versus 6,000–7,500 impulses (Buchbinder), polythene foil barrier (Haake), Styrofoam block (Ogden), focus outside the patient (Speed).
Local anaesthetic in both groups: Abt, Buch, Haake, Ogden (different for placebo); none in Buchbinder or Speed. Ultrasound guidance in Buch, Buchbinder and Speed, not in Ogden, not stated in Abt and Haake.
Not pooled. Two trials did not report adverse events. Buchbinder: one person per arm had pain for a week; in the active arm one reported heat and numbness and one bruising; in the placebo arm one reported burning. Ogden: 38 procedure-related complications, 18 in the active arm, mostly mild numbness or tingling; one plantar fascia rupture 4 weeks after active treatment in a patient with multiple prior cortisone injections. Haake: more side effects with active treatment, OR 2.26 (95% CI 1.02 to 5.18), mainly skin reddening, pain and local swelling. Two trials described treatment as unpleasant; Krischek reported no adverse events.
Count inconsistency: the abstract and Discussion state 897 patients in the pooled analysis, while Figure 2 lists 436 shockwave and 445 control participants. Speed 2003 appears as 42 shockwave and 46 control in Figure 2, while Table 2 gives 26:20 women:men for treatment and 25:17 for control (extractor observation).
Direction and data: Figure 2 is captioned as VAS scores but plots positive differences as favouring treatment, while several shockwave means are higher than control; Haake values in Figure 2 differ from Table 5. The low-quality subgroup is reported with the opposite sign.
Methods: placebo and subtherapeutic-dose controls combined; local anaesthesia in some trials; Abt outcome at about 19 weeks with imputed SDs; quality score thresholds described inconsistently; five of 11 trials not pooled; searches to September 2004. The text lists Rompe [30] as a low-dose control and Rompe [32] as placebo, whereas Table 4 shows a 1-cm gap for [30] and 3,000 versus 30 impulses for [32].
Industry sponsorship: device makers supported three included trials (Haake supplied machine; Buch sponsored by Dornier; Ogden linked to HealthTronics), and Buch and Ogden were used for FDA approval; six trials made no declaration. The authors declared no competing interests; review funding is not stated.
Supplement: Additional file 1 (full EMBASE and MEDLINE search strategies) was not part of the local PDF and was not read.
Findings checked against the full source by a separate automated reviewer.
Limitations noted by the reviewers
- Count inconsistency: the abstract and Discussion state 897 patients in the pooled analysis, while Figure 2 lists 436 shockwave and 445 control participants. Speed 2003 appears as 42 shockwave and 46 control in Figure 2, while Table 2 gives 26:20 women:men for treatment and 25:17 for control (extractor observation).
- Direction and data: Figure 2 is captioned as VAS scores but plots positive differences as favouring treatment, while several shockwave means are higher than control; Haake values in Figure 2 differ from Table 5. The low-quality subgroup is reported with the opposite sign.
- Methods: placebo and subtherapeutic-dose controls combined; local anaesthesia in some trials; Abt outcome at about 19 weeks with imputed SDs; quality score thresholds described inconsistently; five of 11 trials not pooled; searches to September 2004. The text lists Rompe [30] as a low-dose control and Rompe [32] as placebo, whereas Table 4 shows a 1-cm gap for [30] and 3,000 versus 30 impulses for [32].
- Industry sponsorship: device makers supported three included trials (Haake supplied machine; Buch sponsored by Dornier; Ogden linked to HealthTronics), and Buch and Ogden were used for FDA approval; six trials made no declaration. The authors declared no competing interests; review funding is not stated.
- Supplement: Additional file 1 (full EMBASE and MEDLINE search strategies) was not part of the local PDF and was not read.
Source-linked critical reading
Systematic review and meta-analysis
Technical source record
Complete local PDF read (all 11 pages): methods, results, Tables 1–6, Figures 1–2, discussion, conclusion, competing interests and references. Additional file 1 (search strategies) was not available locally.
In randomized trials up to 2003, people with heel pain treated with shockwaves had slightly less morning pain at about 3 months than those given placebo or a very low dose, on average under half a centimetre on a 10-cm scale. In the better-conducted trials the difference was not clear, and the authors did not support routine use on this basis.
Compared with placebo or a low dose, do shockwaves reduce plantar heel pain, and by how much, across randomized trials?
Systematic review and meta-analysis of RCTs: Cochrane registers (2003), MEDLINE 1966, EMBASE and CINAHL 1982 to September 2004, plus reference lists; two German-language trials were translated. Two reviewers selected and extracted; 6-point quality score; fixed-effect RevMan pooling planned unless heterogeneity. 11 trials included, 6 pooled for the a priori primary outcome, morning pain.
Adults with plantar heel pain in 11 RCTs published 1996–2003 (1,290 patients); 6 trials pooled for morning pain at about 12 weeks. Pain lasted over 6 months in 10 trials; some trials enrolled runners, people with heel spurs or people being considered for surgery.
Shockwaves from several devices (Ossatron, Epos Ultra, Osteostar, Sonocur Plus, Orthima) at low or high energy, 500 to 4,000 impulses per session, 1 to 6 sessions; the review does not classify devices as focused or radial.
The text counts 8 placebo-controlled trials and 3 with a low, presumed subtherapeutic dose. Controls in Table 4: absorbent block or foil, polythene barrier, Styrofoam block, reflecting pad, 1-cm gap, focus outside the patient, or far fewer impulses (300 or 30). Local anaesthetic was given to both groups in 4 of the 6 pooled trials.
Primary: morning (first-step) pain on a 10-cm VAS at 12 weeks or nearest point. Secondary: walking, pressure and rest pain, disability, quality of life and adverse events; these could not be pooled.
Six trials: weighted mean difference 0.42 cm (95% CI 0.02 to 0.83), p=0.04, favouring shockwaves; I²=43.9%, fixed effect. Individual trials ranged from −0.50 to 2.00. Excluding the trial with imputed SDs: 0.30 (−0.12 to 0.72). Four higher-quality trials: 0.21 (−0.29 to 0.70), p=0.41. The two lower-quality trials gave a larger, significant difference (reported as −0.90, −1.62 to −0.19, with a reversed sign). Secondary outcomes were summarized only by p-values.
Not pooled. Two trials did not report adverse events. Buchbinder: one person per arm had pain for a week; in the active arm one reported heat and numbness and one bruising; in the placebo arm one reported burning. Ogden: 38 procedure-related complications, 18 in the active arm, mostly mild numbness or tingling; one plantar fascia rupture 4 weeks after active treatment in a patient with multiple prior cortisone injections. Haake: more side effects with active treatment, OR 2.26 (95% CI 1.02 to 5.18), mainly skin reddening, pain and local swelling. Two trials described treatment as unpleasant; Krischek reported no adverse events.
The 897 patients stated do not match the 436 and 445 in Figure 2; Figure 2 direction and Haake values conflict with the caption and Table 5; quality thresholds are described inconsistently and the low-quality estimate has a reversed sign. Placebo and low-dose controls were combined, local anaesthesia varied, and Abt contributed a 19-week outcome with imputed SDs. Three included trials had industry support. Authors declared no competing interests; Additional file 1 was not read.
Reflects trials of mostly long-standing heel pain up to 2003 with older devices and frequent local anaesthesia. It gives a pooled morning-pain difference at about 12 weeks; it says little about function, later follow-up, or newer regimens given without anaesthesia.
Trials reporting means with variance for morning pain and disability, separating sham from low-dose controls and device types, with prespecified clinically important differences.
Read the original article.
The effectiveness of extra corporeal shock wave therapy for plantar heel pain: a systematic review and meta-analysis.
Thomson CE, Crawford F, Murray GD. · BMC Musculoskelet Disord · 2005
Open the publication record or available full-text files.
Checked against the full paper · Clinical review: Dr. Marcus Yu Bin Pai, · How we review →