Kukkonen 2014.

Methods	Design: multi‐centre, parallel, three‐arm, randomised controlled trial Setting: 2 tertiary and 1 secondary care hospitals in Finland Timing: October 2007 and December 2012 Interventions: rotator cuff repair + subacromial decompression + exercises versus subacromial decompression + exercises versus exercises. Sample size: the power calculations were based on the assumed statistical behaviour of the Constant score. The mean score value at baseline was assumed to be 50 (SD 10). The score in the best treatment group at follow‐up was assumed to be 70, and in the worst treatment group 60.25. The correlation between the measurements during the follow‐up was estimated to be 0.40 to 0.50 (SD 20). By analysis of variance (ANOVA), with alpha = 0.05 and power = 85%, statistical significance could be reached with 51 participants per treatment group. The expected drop‐out rate was set at 15%, so the number of participants per group was 60. Analysis: per protocol analysis. The authors excluded participants with no full‐thickness tear after allocation and data from these participants were not analysed at follow‐up time points. Participants who crossed over, were analysed as allocated.
Participants	Number of participants 370 screened clinically, 271 scanned with MRI 180 eligible 180 randomised (60 repair, decompression and exercises, 60 to decompression + exercises and 60 to exercises alone) 174 received treatment as randomised At 3 months, data available for 170 (55/59 (93%) for repair, decompression and exercises, 58/59 (98%) for decompression and exercises, and 57/58 (98%) for exercises alone) At 6 months, data available for 158 (47/59 (80%) for repair, decompression and exercises, 54/59 (92%) for decompression and exercises and 57/58 (98%) for exercises alone At 12 months, data available for 167 (55/59 (93%) for repair, decompression and exercises, 57/59 (97%) for decompression and exercises, and 55/58 (95%) for exercises alone) Inclusion criteria Age > 55 years Atraumatic symptomatic supraspinatus tendon tear comprising < 75% of the tendon insertion and documented with MRI Full range of motion of the shoulder Exclusion criteria Age < 55 years History of trauma relating to the onset of symptoms A massive tendon tear involving the whole supraspinatus tendon and/or combined tear of two to three tendons, i.e., supraspinatus with infraspinatus or subscapularis tendon tear Stiffness of the glenohumeral joint (passive external rotation < 30° ± elevation < 120°) Glenohumeral osteoarthritis with present osteophytes in radiographs Systemic corticosteroid or antimetabolite medication Significant malignant, hematological, endocrine, metabolic, rheumatoid or gastrointestinal disease History of alcoholism, drug abuse, psychological or other emotional problems that are likely to invalidate informed consent Previous surgery of same shoulder Baseline data Repair, decompression and exercise group Mean (SD) age 65: (6.0) years Number (%) females: 26 (47) Number (%) participants with right side affected n (%): 36 (65) Number (%) participants working: 23 (42) Number (%) participants on sick leave: 1 (2) Number (%) participants retired: 30 (55) Number (%) smokers: 8 (15) Number (%) participants with prior corticosteroid injection: 31 (56) Mean (SD) duration of symptoms: 28 (9.5) months Mean (95% CI) Constant score: 58 (54.1‐61.9) Mean (95% CI) pain (VAS) 2.6 (2.0 to 3.2) Mean (SD) tear size in mm: 8.5 (4.0) Number (%) participants with biceps pathology present: 16 (29) Number (%) participants with OA changes in GH‐joint: 23 (42) Number (%) participants with OA in AC‐joint: 51 (93) Number (%) participants with grade II fatty degeneration 26/55 (47) Number (%) participants with grade III fatty degeneration 2/55 (4) Number (%) participants with muscle atrophy 2/57(4) Decompression + exercise group Mean (SD) age 65 (5.1) years Number (%)females: 28 (49) Number (%) participants with right side affected: 33 (58) Number (%) participants working: 10 (18) Number (%) participants on sick leave: 1(2) Number (%) participants retired: 42 (74) Number (%) smokers: 5 (9) Number (%) participants with prior corticosteroid injection: 33 (58) Mean (SD) duration of symptoms: 28 (9.7) months Mean (95% CI) Constant score: 59.6 (55.8‐63.4) Mean (95% CI) pain (VAS) 2.5 (1.8 to 3.2) Mean (SD) tear size in mm: 9.3 (5.3) Number (%) participants with biceps pathology present: 21 (37) Number (%) participants with OA changes in GH‐joint: 28 (49) Number (%) participants with OA in AC‐joint: 52 (89) Number (%) participants with grade II fatty degeneration 26/57 (46) Number (%) participants with grade III fatty degeneration 5/57 (9) Number (%) participants with muscle atrophy 5/60 (8) Exercise group Mean (SD) age: 65 (5.8) years Number (%) females: 31 (56) Number (%) participants with right side affected: 41 (75) Number (%) participants working: 17 (31) Number (%) participants with on sick leave: 5 (9) Number (%) participants retired: 32 (58) Number (%) smokers: 10 (18) Number (%) participants with prior corticosteroid injection: 39 (71) Mean duration of symptoms: 26 (9.9) months Mean Constant score (95% CI): 57.8 (53.9‐61.7) Mean (SD) tear size in mm: 9.6 (5.2) Number (%) participants with biceps pathology present: 16 (29) Number (%) participants with OA changes in GH‐joint: 19 (35) Number (%) participants with OA in AC‐joint: 48 (87) Number (%) participants with grade 2 fatty degeneration 29/55 (53) Number (%) participants with grade 3 fatty degeneration 1/55 (2) Mean (95% CI) pain (VAS) 2.7 (1.9 to 3.5) Number (%) participants with muscle atrophy 1/60 (2)
Interventions	All operations were performed arthroscopically in a standardised manner by four experienced shoulder surgeons. A physiotherapist trained in shoulder rehabilitation gave the patient written information and guided the patient in how to perform a standardised training exercise protocol at home. Exercise group A physiotherapist trained in shoulder rehabilitation gave the patient written information and guidance for exercises to be conducted at home. The exercise protocol was standardised and started with exercises aimed at improving glenohumeral motion and active scapular retraction for the first six weeks. Then static and dynamic exercises for the scapular and glenohumeral musculature were gradually increased from six weeks to 12 weeks, after which the participant increased resistance and strength training up to six months. In addition to written instructions the patient was referred for ten sessions of physiotherapy in an outpatient health care facility where their progress was monitored. Subacromial decompression group Subacromial debridement and an arthroscopic acromioplasty were carried out by smoothing the inferior surface of the acromion from a postero‐anterior direction. The sagittal size of the supraspinatus tear was measured with a probe. In addition, 6 mm of the acromioclavicular (AC) joint was resected if it had been painful before surgery and if there were severe degenerative changes in the AC joint on the MRI. If the long head of the biceps tendon was unstable or frayed, a biceps tenotomy was also performed. After the operation a physiotherapist gave the patient guidance on how to exercise to improve free glenohumeral motion and how to retract the scapula actively. After three weeks the physiotherapist controlled the progress of rehabilitation and gave the patient written information for movement and gradual resistance exercises to be conducted at home and sessions of physiotherapy, as for exercise group. Subacromial decompression and rotator cuff repair group Subacromial debridement and acromioplasty were performed arthroscopically. The sagittal size of the supraspinatus tear was measured with a probe. The rotator cuff was repaired anatomically using standard titanium bone anchors with non‐absorbable sutures (Corkscrew FT II; Arthrex Inc., Naples, Florida or Twinfix; Smith‐Nephew, Andover, Massachusetts). AC resection and/or biceps tenotomy were per‐ formed, if indicated. After the operation the arm was immobilised in a sling for three weeks after which the rehabilitation followed the same regime as exercise group.
Outcomes	Outcomes were assessed at 3 months, 6 months and 12 months Primary outcome: Constant score (0 to 100, higher score indicates better function) Secondary outcomes: Pain in VAS (0 to 10, higher score indicates worse pain) Satisfaction with two questions: 1) better or worse compared with pre‐operative state and 2) satisfied or dissatisfied with the treatment outcome Costs for patients and for society (direct treatment related in €). Outcomes used in this review Mean pain; pain in VAS Mean function; Constant score Satisfaction: proportion of patients who were satisfied with the treatment outcome
Source of funding	The authors report no conflicts of interest that could have affected the report.
Notes	Trial registration: NCT01116518 Comparison: we included the following comparisons: repair with subacromial decompression versus non‐operative treatment, repair with subacromial decompression versus decompression alone. We excluded subacromial decompression versus non‐operative treatment as it did not meet inclusion criteria. Data analysis: for function at 6 months, we extracted the Constant score from figure; at 12 months the Constant score was reported; at 24 months, the change from baseline was reported (Analysis 1.2). For pain, we extracted values from figure (Analysis 1.1). Withdrawals: in exercise group, 3/60 (5%); in subacromial decompression group 1/57(2%); in subacromial decompression and repair group 5/60 (8%). Data missing: 2/60 (3%) in exercise; 2/60 (3%) in subacromial decompression; 0/60(0%) in decompression and repair group at 12 months. The authors excluded participants who withdrew or had missing data. Cross overs: in exercise group 4/60 (7%) had rotator cuff repair and in subacromial decompression group 1/60 (2%) had rotator cuff repair by 12 months. Their data were analysed in their allocated groups. Two participants were excluded after allocation because they were found to have intact supraspinatus tendons at surgery. Adverse events: no treatment related complications in any of the groups Serious adverse events: the authors did not report if there were any serious adverse events or not.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Random sequence generation not reported.
Allocation concealment (selection bias)	Unclear risk	Quote:"After consent, the study nurse randomised the patients into one of the three treatment groups using sequentially numbered, opaque, sealed envelopes. The randomisation process was stratified according to participating hospital". No further information was given about allocation concealment.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote:"After randomisation, the patient and the treating physician were openly informed of the treatment group." Participants and personnel were not blinded.
Blinding of outcome assessment for self‐reported outcomes including pain, function and global assessment (detection bias)	High risk	Participants were not blinded.
Blinding of outcome assessment for incidence of full‐thickness tears at follow‐up (detection bias)	High risk	Constant scores were recorded by an independent study nurse at each time point. It is not stated if this nurse was blinded to the allocation. All MRI images were re‐evaluated at the end of follow‐up by two musculoskeletal radiologists (KTM and EKJT) blinded to patient data but they could not have been reliably blinded to treatment intervention.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Withdrawals: in exercise group, 3/60 (5%); in subacromial decompression (ASD) group 1/57(2%); in ASD and repair group 5/60 (8%). Data missing at 6 months:1/60 (2%) in exercise; 5/60 (8%) in ASD group; 8/60 (13%) in ASD and repair group. Data missing at 12 months: 2/60 (3%) in exercise; 2/60 (3%) in ASD; 0/60 (0%) in ASD and repair group. Data missing at 24 months: 2/60 (3%) in exercise group; 1/60 (2%) in ASD group; 1/60 (2%) in ASD and repair group. The authors have excluded participants who withdrew or had missing data at each time point. Small numbers dropped out and reasons for dropout reported and unlikely to influence results.
Selective reporting (reporting bias)	Unclear risk	Adverse events not reported and unclear if they were measured. Global satisfaction partially reported.
Other bias	Unclear risk	By one year, 4/60(7%) and by two years, 7/60 (12%) participants from the exercise group had rotator cuff repair during follow‐up but were analysed according to their allocated treatment groups. One participant from the subacromial decompression had rotator cuff repair during follow‐up. All participants were analysed according to their allocated treatment groups. An additional AC resection was performed in 7/57 (12%) in decompression group and 8/55 (15%) in decompression and repair group. An additional biceps tenotomy was performed in 29/57 (51%) in subacromial decompression group and in 23/55 (42%) in subacromial decompression and repair group. The cross overs and side interventions could bias the results regarding surgery versus non‐operative treatment but unclear if to an important level.