Reindl 2015.

Study characteristics
Methods	RCT; parallel design Review comparison group: intramedullary devices versus DHS
Participants	Total number of randomised participants: 204 Inclusion criteria: unstable intertrochanteric hip fracture; ≥ 55 years of age; type 2 (AO/OTA 31 ‐ A2); isolated fracture; occurred < 2 weeks prior to the time of enrolment Exclusion criteria: fracture due to malignancy; inability to walk before the fracture; severe dementia; limited life expectancy due to substantial medical comorbidities; medical contraindication; inability to comply with rehab or complete the forms Setting: multicentre; 9 sites; Canada Baseline characteristics Intervention group 1 (intramedullary device) Age, mean (SD): 82 (± 8.6) years Gender, M/F: 57/55 Intervention group 2 (DHS) Age, mean (SD): 80 (± 9.9) years Gender, M/F: 31/61 Note: study authors did not report: smoking history, medication, BMI, comorbidities, mobility assessment cognitive status/dementia, preoperative waiting time
Interventions	General details: fracture table; attempted closed reduction; use of fluoroscopic guidance; clinical evaluations at 6 weeks, 3, 6 and 12 months Intervention group 1 A choice of 3 intramedullary devices: trochanteric fixation nail (TFN) (Synthes), Gamma nail (Stryker) or Trigen Intertan nail (Smith & Nephew); short nails; dynamic fixation proximally and all were distally locked Randomised = 112; devices: 42 = TFN, 48 = Intertan, 22 = Gamma nail; at 3 months = 96; at 12 months = 87 (13 died, 6 unwilling to continue, 5 unknown loss, 1 implant failure) Intervention group 2 DHS (Synthes); plate ranges in length from two to six holes at the surgeon’s discretion Randomised = 92; at 3 months = 85; at 12 months = 80 (6 died, 2 unwilling to continue, 2 unknown loss, 2 implant failure)
Outcomes	Outcomes measured/reported by study authors: available at 6 weeks, 3, 6 and 12 months: LEM; FIM; TUG; 2MWT; radiographic findings; implant position ‐ tip‐apex distance; femoral neck shortening; heterotopic ossification ‐ Brooker stage; complications; length of follow‐up: 12 months Outcomes relevant to the review: mortality and unplanned return to theatre (both 12 months)
Notes	Funding/sponsor/declarations of interest: the study was directed by the Canadian Orthopaedic Trauma Society (COTS) with no other conflicts reported Study dates: February 2007 to November 2012
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quotes: “Permuted block randomisation”,“randomly generated modality”
Allocation concealment (selection bias)	Unclear risk	Quote: “sealed envelopes” Comment: study authors do not report whether envelopes are opaque and sequentially numbered
Blinding of participants and personnel (performance bias) All outcomes	Low risk	It is not possible to blind surgeons to the type of intervention. We did not, however, expect that this would influence surgeon performance.
Other performance bias: surgeon experience of both implants	Unclear risk	Trial appears pragmatic in design. Multicentre trial with no information available on surgeon expertise
Blinding of outcome assessment: mortality (detection bias)	Low risk	We did not expect that lack of blinding of outcome assessors would influence objective outcome data.
Blinding of outcome assessment: unplanned return to theatre (detection bias)	High risk	It is not possible to blind surgeons to treatment groups. We judged that knowledge of the type of implant could influence judgments made by surgeons when assessing subjective outcomes.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Losses were balanced between groups and were mostly explained by death, which is expected in this population.
Selective reporting (reporting bias)	Unclear risk	Registration with clinical trials register NCT00597779: first registered in January 2008 although study commenced in February 2007. It was not feasible to effectively assess risk of reporting bias using retrospectively‐prepared documents. We noted that SF‐36 was listed as an outcome, but was later dropped from the outcome list on the clinical trials register and was not reported in the published study report.
Other bias	Low risk	We identified no other sources of bias.