Brox 1993

Methods	Randomised controlled trial set in hospital surgery and physiotherapy departments in Norway Randomisation method: random permuted blocks Allocation concealment: unclear Blinding: outcome assessors blinded, participants not blinded Loss to follow up: 5 (4%) Intention to treat analysis: analysed in group allocated, but excluding n=5 lost to follow‐up (i.e. available case analysis) Overall validity: high risk of bias
Participants	125 participants, 66; mean age 48 years (no variance reported) Inclusion criteria: aged 18‐66; pain in shoulder > 3 months; resistant to physiotherapy, NSAIDS, steroids; dysfunction or pain on abduction; normal passive gleno‐humeral ROM; pain during two of three isometric‐eccentric tests (abduction at 0 degrees and 30 degrees and external rotation); positive results in tests for impingement; positive response to subacromial injection of local anaesthetic into the subacromial space Exclusion criteria: arthritis of acromioclavicular joint; cervical syndrome; rupture of the rotator cuff; glenohumeral instability; bilateral muscular pain with tenderness; severely decreased ability to relax shoulder, neck, and temporo‐mandibular joints on examination; reluctant to accept one or more study treatments
Interventions	Group 1 (N = 45): arthroscopic subacromial decompression Group 2 (N = 50): exercise regime over three to six months supervised by one experienced physiotherapist Group 3 (N = 30): placebo laser‐ 12 sessions of detuned soft laser treatment over 6 weeks
Outcomes	Assessed at baseline, 3, 6 months Primary outcome measure was change in overall Neer score from baseline to six months 1) Neer score 2) Self assessed degree of pain on 9‐point VAS scale (1 = no pain, 9 = worst possible) with activity, at rest and at night; 3) Emotional distress on the Hopkins symptom checklist (25 items) (only reported at baseline in trial)
Notes	Outcomes reported as medians, and thus we could not extract data into the Comparisons and Data tables; primary outcome in Additional tables 08. Trial authors also report components of primary outcome separately; we have not reported the components in this review. Authors performed an interim analysis of 68 participants who completed 6 months follow up and found that surgery or exercises were superior to placebo, and thus stopped allocating participants to placebo (hence the smaller number in placebo); the authors did not appear to statistically adjust for the interim analysis in the final analysis. No differences were found between the three groups in duration of sick leave to six months (median 3 months) and daily intake of analgesics. Participants in both groups that received active treatment improved significantly more that those in the placebo group at 6 months: median differences between exercises (13.0 (95% CI 7 to 20)) and surgery (19.5 (95% CI 12 to 27)) compared with placebo (mean change in Neer score ‐0.3 with placebo compared with 10.8 in the exercise group and 20.2 in the surgery group). Treatment costs were higher for those given surgery (720 pounds) versus those given supervised exercises (390 pounds) reported to be due to hospitalisation in the surgical group. Adverse outcomes: Nil Funding: Norwegian Research Council A second paper by Brox was written in 1999 assessing these patients at 2.5 years but the outcome assessment was no longer blinded, and the results have not been included in this review
Risk of bias
Item	Authors' judgement	Description
Allocation concealment?	Unclear	B ‐ Unclear