Robinson 2013a.

Methods	Study design: multicentre parallel group, 2‐arm RCT (3 centres) Duration of the study: September 2007‐July 2010 Protocol was published before recruitment of participants: unclear; trial was registered but protocol methods could not be found. Details of trial registration: yes. The study was registered on the National Research Register of the National Health Service Institute for Health Research (Number N0256199069). Funding sources: none of the study authors received payments or services, either directly or indirectly.
Participants	Place of study: New Royal Infirmary of Edinburgh, Glasgow Royal Infirmary, and Frenchay Hospital, Bristol; UK Number of participants assigned: 200 participants (95 surgical; 105 conservative) Number of participants assessed: 178 participants (86 surgical; 92 conservative) Inclusion criteria Aged 16‐60 years An isolated and completely displaced fracture in the middle third of the clavicle Fresh fracture, treatment within 2 weeks after injury No previous shoulder abnormalities No pathological fracture No open fracture No neurovascular injury No head injury No contraindications to general anaesthesia No cognitive impairment Informed consent Age: Surgical group (mean (95% CI)): 32.3 (30.1 to 34.5) years Conservative group (mean (95% CI)): 32.5 (30.1 to 34.9) years Gender of participants assigned (male/female) Surgical group: 83/12 Conservative group: 92/13 Classification of injury: fractures were classified according to the Robinson's Classification (Robinson 1998).
Interventions	Timing of intervention: participants underwent surgery within 2 weeks after the injury Type of surgical intervention: open reduction and plate fixation Type of conservative intervention: collar and cuff for 3 weeks Rehabilitation Surgical group: same as the conservative group Conservative group: total shoulder immobilisation for 3 weeks, after that physiotherapy for ROM exercises. Followed by strengthening starting at 6 weeks and it was recommended to not practice sports for 4 months Any co‐interventions: not reported
Outcomes	Length of follow‐up Follow‐up was 1 year Participants were evaluated at 6 weeks, and at 3, 6 and 12 months Loss to follow‐up: 22/200 participants were lost to follow‐up at 1 year Surgical group: 9/95 participants were lost at 1 year Conservative group: 13/105 participants were lost at 1 year Primary outcomes Function or disability measured by DASH and Constant score Treatment failure measured for symptomatic non‐union, malunion or other complications Secondary outcomes Other treatment failure measured by: cosmetic result: dissatisfaction with shoulder droop, shoulder asymmetry, or local bump at fracture site asymptomatic non‐union Adverse events measured by: infection symptoms of discomfort related to the implant, requiring removal; skin and nerve problems; stiffness/restricted range of shoulder movement Health‐related quality of life measured by: Short Form‐12 (physical and mental scores) Numbers of participants returning to previous activities (work and sport).
Notes	When conservative method failed after 6 months (26.1%, 24 participants), 13 participants chose to undergo a secondary open reduction and plate fixation. We tried unsuccessfully to contact the study authors to obtain further information. Composite adverse events: surgery: superficial wound infection (n = 2); dehiscence (n = 1); local sensitivity/irritation (n = 17); local numbness (n = 15); plate removal (n = 10); rotator cuff impingement (n = 2); adhesive capsulitis (n = 1). Conservative: symptomatic malunion (n = 1); local sensitivity/irritation (n = 11); local numbness (n = 4); rotator cuff impingement (n = 1). In Analysis 1.14 we used event rates 32 for surgery and 15 for conservative. Cosmetic result events: surgery: dissatisfaction with shoulder droop (n = 1), shoulder asymmetry (n = 2), bump at fracture site (n = 5). Conservative: dissatisfaction with shoulder droop (n = 15), shoulder asymmetry (n = 17), bump at fracture site (n = 26). These are unlikely to be mutually exclusIve. Analysis 1.11 we used event rates 5 for surgery and 26 for conservative.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The sequence generation was performed by a computer random number generator.
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	A research physiotherapist, who was blinded by having the participant wear a t‐shirt, assessed ROM and strength in both limbs. High risk of bias for self‐reported outcomes (function and satisfaction) as participants were aware of treatment group. Unclear if unblinded assessment by radiographer of treatment failure (e.g. symptomatic non‐union) and asymptomatic non‐union was subject to bias.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Only 11% of participants were lost to follow‐up. Missing outcome data were balanced in numbers across intervention groups. The study authors said that there were no significant demographic differences between the participants who were lost to follow‐up and those who were followed to 1 year.
Selective reporting (reporting bias)	Unclear risk	Despite the study protocol being registered, the pre‐specified outcomes were not recorded and were not available.
Other bias	Low risk	The study appears to be free of other potential sources of bias.