Beyer‐Westendorf 2017.

Methods	Open‐label, masked endpoint, randomised, non‐inferiority phase 3b trial.
Participants	472 participants aged ≥ 18 years with symptomatic ST (≥ 5 cm in a supragenual superficial‐vein segment) and ≥ 1 additional risk factor for thromboembolic complications (aged ≥ 65 years, male sex, previous VTE, cancer, autoimmune disease, thrombosis of non‐varicose veins).
Interventions	Rivaroxaban (10 mg oral od). Fondaparinux (2.5 mg sc od). Study treatment given for 45 days.
Outcomes	Primary efficacy outcome: composite of symptomatic DVT or PE, progression or recurrence of ST, and all‐cause mortality at 45 days. Secondary efficacy outcomes: composite primary efficacy outcome at 90 days, incidence of each component of the primary efficacy outcome at 45 and 90 days; major VTE (symptomatic PE, symptomatic proximal DVT, or VTE‐related death) at days 45 and 90; and surgery for ST within 45 and 90 days of initiation of study drug treatment. Primary safety outcome: major bleeding. Secondary safety outcomes: clinically relevant non‐major bleeding, and minor bleeding within 45 days of initiation of treatment with study drug, censored 2 days after the last dose of study drug.
Notes	Funding: GWT‐TUD and Bayer Vital.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Adequate method of randomisation. Quote: "Randomisation was done with a central block randomisation process, with a random block sequence of four numbers per block."
Allocation concealment (selection bias)	Low risk	Participants and investigators could not foresee treatment allocation. Quote: "The generation of the allocation sequence to assign all patients to the two treatment groups and the allocation were done at the sponsor’s site by the study manager, who had no role in the trial or in data collection or analysis."
Blinding (performance bias and detection bias) All outcomes	High risk	Open‐label study with blinded outcome assessment. Quote: "An independent committee, whose members were unaware of study group assignment, adjudicated the qualifying diagnosis, the anatomical extent of the initial superficial‐vein thrombosis, and all suspected outcomes."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Primary analysis originally planned as ITT, but modified into a per‐protocol analysis during study. 36/471 (7.6%) participants randomised were excluded from the primary analysis but the authors reported results also according to the ITT principle. Quote: "The original study protocol specified that the primary efficacy analysis would be done in the intention‐to‐treat population; however, the protocol was changed by the steering committee on June 15, 2016, to specify analysis in the per‐protocol population."
Selective reporting (reporting bias)	Low risk	All outcome described in the methods and protocol were presented in the results.