The unit of randomization was the treated limb, assigned by a computer-generated random sequence. Some participants underwent bilateral procedures and could contribute limbs to different arms; within-participant correlation was addressed with patient-clustered robust standard errors and a one-limb-per-patient sensitivity analysis. The primary endpoint was clinically significant (AVF class ≥II) EHIT on the first post-ablation duplex, which was performed and interpreted by two registered vascular technologists blinded to treatment allocation, with adjudication by the treating physician.
