Fuji 2015.

Methods	Randomised, open‐label, controlled study
Participants	Total number of participants: 43 Number of participants allocated to each group: fondaparinux (FX) group: 21, edoxaban group: 22 Number of participants excluded and/or lost to follow‐up: FX group: 3 participants discontinued, edoxaban group: 2 participants discontinued Inclusion: patients ≥ 20 years of age, with serious renal injury (creatinine clearance ≥ 20 mL/min to < 30 mL/min) who were undergoing unilateral TKA or THA (excluding revision surgeries) or hip fracture surgery for medial or lateral femoral neck fracture (trochanteric or subtrochanteric section of the femur) within 10 days of presurgical examination. Informed consent was obtained from all participants. Exclusion: Presurgical exclusion criteria included, but were not limited to, patients undergoing or possibly undergoing haemodialysis; risk of bleeding; risk of thromboembolism; and hepatic dysfunction. Postsurgical exclusion criteria included, but were not limited to, creatinine clearance < 15 mL/min; abnormal bleeding at the site of spinal anaesthesia; abnormal or excessive bleeding during or immediately after surgery; and inability to take oral medication.
Interventions	Fondaparinux: subcutaneous fondaparinux 1.5 mg sc once daily Edoxaban: oral edoxaban 15 mg once daily
Outcomes	Primary efficacy outcome: incidence of symptomatic VTE (composite of symptomatic DVT or PE) during treatment period Primary safety outcome: major bleeding defined as fatal bleeding; clinically overt bleeding accompanied by a decrease in haemoglobin > 2 g/dL or requiring a transfusion of > 4 units of blood (1 unit = ˜200 mL); retroperitoneal, intracranial, intraocular or intrathecal bleeding; or bleeding requiring repeat surgery
Notes	Use of adjunctive anticoagulative methods: Concomitant physiotherapy (intermittent pneumatic compression devices or elastic stockings) was permitted throughout the treatment period.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomised, controlled study applied permuted block method with SAS software to generate random sequence. Comment: probably done
Allocation concealment (selection bias)	Low risk	Randomised, controlled study applied permuted block method with SAS software to generate random sequence. Comment: probably done
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Open‐label study Comment: probably not done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	All thromboembolic events were assessed by a thromboembolic event assessor, who was blinded to treatment group, on the basis of imaging results. Comment: probably done
Incomplete outcome data (attrition bias) All outcomes	Low risk	18/21 and 20/22 participants in the fondaparinux and edoxaban groups, respectively finished their treatment. Comment: probably done
Selective reporting (reporting bias)	Low risk	All primary efficacy and safety outcomes listed in the Methods section were reported. Comment: low risk of bias
Other bias	Unclear risk	This study was supported by a pharmaceutical company that was involved in the study design and analysis of the data and provided writing and editorial support.