Chen 2016.

Methods	Randomised using a computer‐generated randomisation sequence
Participants	60 participants (32 female, 28 male; mean age 66 years) with acute osteoporotic (bone mineral density (BMD) < ‐3.0) 4‐part proximal humeral fractures with or without fracture dislocation
Interventions	1. Intramedullary fibular allograft (IFA) with locking compression plate (LCP). Rehabilitation not described. 2. Hemiarthroplasty. Prosthesis (LINK, Germany). All had general anaesthesia and a deltopectoral approach to surgery.
Outcomes	Follow‐up: mean 36 months (range 24 to 48 months) DASH score, Constant score, subjective rating (excellent, good, fair, poor); pain (VAS) (not reported), complications, secondary surgery, time to union
Notes	BMD < ‐3.0 was a criterion for inclusion but mean BMD in both groups was ‐2.5. Thus criterion not followed. A request for further information and clarification sent to P Tang (by email 28 February 2020) was unsuccessful. The request sought details of the rehabilitation for the graft and plate group, checks on Table 4 in the trial report, especially the DASH scores which seem very low % of affected side, and what arthrolysis involved for the 4 participants treated for shoulder stiffness. Concerns over the analyses and data meant that we decided it was best to wait on clarification from the authors before inclusion.