Miller 2005.

Methods	Randomised trial for consenting participants (25), but a further patients enrolled and treated according to surgeon's preference (see Notes) Study period: June 1995 to July 1997
Participants	Children's Hospital, Boston, MA, USA 34 children with closed displaced metaphyseal fractures of the distal radius. Aged 10 years or over. Angulation > 30 º or complete fracture displacement Exclusion: open fracture, history of injury or surgery of the affected wrist, fractures requiring open reduction, swelling or neurovascular compromise precluding circumferential cast immobilisation. Skeletal maturity Sex: male 31 (91%) Age: mean 12.4 years, range 10 to 14 years Fracture type: displaced metaphyseal fractures; no information on ulna involvement Assigned: 16 (wire) / 18 (cast only) Analysed: ? / ? (25 followed up at mean 2.8 years)
Interventions	Closed reduction under general anaesthesia with fluoroscopic guidance 1. Percutaneous wire fixation. Small incision made over radial styloid. Wire directed proximally and ulnarly across fracture site engaging in opposite cortex. Optional second wire inserted through small dorsal incision. 6 participants (37.5%) required double‐pin fixation and 2 (12.5%) required transphyseal pin fixation.Then above‐elbow cast. Wires removed at 4 weeks 2. Above‐elbow cast. Above‐elbow cast comprised plaster cast overwrapped with fibreglass casting material All participants had above‐elbow cast for 4 weeks and then a further 2 weeks in a below‐elbow cast Repeat reduction was performed if participants showed > 25 º of angulation or complete loss of cortical contact at follow‐up
Outcomes	Length of follow‐up: 6 months (average 10½ weeks). Also 1, 2, 4 and 6 weeks (clinical evaluation and radiographs) Long‐term follow‐up: mean 2.8 years (numbers in each group not stated) Long‐term pain, limitations in range of motion, strength, or activities (none noted) Long‐term neurovascular compromise, growth arrest or deformity (none noted) Fracture alignment (post‐initial treatment and change between weeks 1 and 4) Loss of reduction and secondary procedures Nerve hyperaesthesia Tendon (extensor carpi ulnaris) irritation Wire migration Pin‐site (wire‐site) infection Failed closed reduction Non‐union Permanent nerve damage Compartment syndrome Cost analysis (see Notes)
Funding and declarations of interest	Funding source: "None of the authors received financial support for this study." Declarations of interest: reported, none
Notes	Separate data were not provided for the 9 children treated according to the surgeon's preference. Discrepancies between the 2 groups in initial dorsal angulation and shortening may have reflected some bias in the surgeon preference group A retrospective cost analysis was based on charges for operating room, anaesthesia services, orthopaedic surgery, office visits, radiology, plaster cast services
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: “Patients who agreed to study participation were randomised to a treatment group via the drawing of sealed envelopes. Patients who declined participation were treated according to the preference of the attending pediatric orthopaedic surgeon on call. Of the eight participating surgeons, four treated their patients primarily by closed reduction and casting, the other four primarily by pin fixation.” Separate data not available for the 25 properly randomised participants; thus, based this on the, at best, quasi‐randomised allocation for the other 9 participants
Allocation concealment (selection bias)	High risk	As above
Blinding of participants and personnel (performance bias) Objective outcomes	High risk	Not blinded – interventions obvious
Blinding of outcome assessment (detection bias) Subjective outcomes	Low risk	Limited subjective outcomes at long‐term follow‐up. Unlikely to be affected at long‐term follow‐up
Blinding of outcome assessment (detection bias) Objective outcomes	Unclear risk	Not blinded but criteria stated for displacement; other complications would have been self‐evident Long term was probably ‘low risk’
Incomplete outcome data (attrition bias) Subjective outcomes	Unclear risk	Just applies to the long term (2.8 years). However, loss to follow‐up (9: 26% of 34) of the already mixed population, with no separate denominators for the 2 groups
Incomplete outcome data (attrition bias) Objective outcomes	Unclear risk	Mixed population of randomised and non‐randomised. Although no losses to follow‐up, the flawed data handling is the key issue
Selective reporting (reporting bias)	High risk	No trial registration or published protocol. No explicit functional recovery data. The merging of data from the randomised and non‐randomised groups is inappropriate, even though it appears to have been approved by ethics
Other bias: major imbalance in baseline characteristics	High risk	Differences in the displacement: pinning lower dorsal angulation (17 º versus 30 º) but less shortening (6.9 versus 16.4 mm) Quote: “more patients in the pinning group had bayonet apposition of the fracture site; this reflected the element of the randomisation process based upon surgeon preference.” Bayonet apposition is where the 2 fracture fragments lie next to each other rather than in end‐to‐end contact
Other bias: performance bias	Unclear risk	No mention of expertise of clinicians
Other bias	Low risk	The issues relating to the inappropriate use of non‐RCT data are already covered in the above ratings