Twenty-eight RCTs were included. It was not possible to calculate the total number of patients because of potential double-counting in studies with more than one treatment group.
In terms of study quality, the most common methodological flaws were absence of allocation concealment and the lack of intention-to-treat analysis.
Twenty-four RCT assessed only short-term outcomes (less than 6 weeks). Only 8 RCTs followed up patients beyond 6 months.
Exercise (1 RCT): the study found that exercises may improve pain, but not MGS, compared with US.
Manipulation (3 RCTs): a meta-analysis of 2 RCTs showed that manipulation improved PFGS (SMD 1.28, 95% CI: 0.84, 1.73) and pressure pain threshold (SMD 0.49, 95% CI: 0.08, 0.90) post-intervention in comparison with placebo, but there was no longer term follow-up. The other RCT showed no significant difference between a combination treatment that included exercise and control for global improvement, PVAS, MGS or pressure pain threshold.
Orthotics and taping (3 RCTs): the studies used different control groups and assessed outcomes at different times. One RCT found taping significantly improved PFGS and pressure pain threshold post-treatment. Other studies found no significant difference between an orthotic and corticosteroid injection (1 RCT) or between an off-the-shelf orthotic, two placebo braces and control (1 RCT).
Acupuncture (4 RCTs): 2 of 3 RCTs comparing acupuncture with placebo found a significant improvement with acupuncture post-treatment, but no significant different between treatments at 2 months; the other placebo-controlled RCT found acupuncture significantly increased the duration of pain relief and improved success. The fourth RCT found no significant difference between acupuncture and US post-treatment or at 1 month.
Laser (6 RCTs): the short-term results for laser versus placebo were mixed. Four RCTs found no significant difference between treatments, while 1 RCT found a significant improvement with laser. The meta-analysis showed no statistically significant difference between laser and placebo for PVAS, PFGS and global improvement at 3 months, 6 months or 1 year (the results were reported).
ESWT (2 RCTs): the meta-analysis showed no statistically significant difference between ESWT and placebo for PVAS (SMD 0.02, 95% CI: -0.19, 0.24) or global improvement (RR 1.01, 95% CI: 0.78, 1.57) at 4 to 6 weeks. One high-quality RCT found that ESWT significantly increased side-effects (RR 2.64, 95% CI: 1.98, 3.53).
Electromagnetic field therapy and ionisation (4 RCTs): a meta-analysis of the 2 placebo-controlled RCTs of ionisation showed no significant difference between treatments for patient-rated global improvement at 1 to 3 months (RR 1.03, 95% CI: 0.83, 1.28). One RCT found that ionisation with high- or low-dose NSAID significantly improved PVAS compared with sham ionisation. The one electromagnetic study presented insufficient data to calculate an effect size.
US and phonophoresis (5 RCTs): a meta-analysis of the 2 detuned placebo-controlled RCTs of US ionisation showed no significant difference between treatments for global improvement at 3 months (RR 1.01, 95% CI: 0.62, 1.65). Other RCTs found no significant difference between US and acupuncture at 4 weeks (1 RCT) or for US plus either transverse friction massage or hydrocortisone coupling compared with US alone post-treatment (1 RCT). The fifth RCT compared US with exercise (see 'Exercise' above for details). Combined physical interventions (2 RCTs): one RCT used manipulation as the control (see 'Manipulation' above for details). The other RCT found corticosteroid injection improved outcomes at 6 weeks in comparison with combined treatment (massage, US and exercise), but the physical intervention significantly improved outcomes at 6 and 12 months. It found no significant difference between combined treatment and a no treatment control.