Abstract
BACKGROUND: Prior nonrandomized studies have suggested nephroprotective effects of arteriovenous fistula (AVF) formation, but these are plausibly susceptible to immortal time and selection biases. METHODS: We studied patients attending nephrology clinics in the West of Scotland during 2010-22 with an estimated glomerular filtration rate (eGFR) ≤15 mL/min/1.73 m(2) and no prior AVF. Using target trial emulation and a sequential trial design, we simulated a hypothetical trial that would randomize patients to either undergo AVF formation immediately or not to undergo AVF formation. The primary outcome was the difference in eGFR slope for the first 6 months of follow-up, estimated using a mixed-effects model. The secondary outcomes were 5-year absolute risks of dialysis and death, estimated using the Aalen-Johansen and Kaplan-Meier estimators respectively. RESULTS: A total of 1364 unique patients (mean age 51.1 years, 55.7% male) contributed 3125 person-trials, with 561 in the AVF and 2564 in the no AVF group. Mean eGFR was 12.6 mL/min/1.73 m(2) and the median number of eGFR measurements per person-trial was 7 (interquartile range 4-12). Slope of eGFR decline did not differ significantly between the AVF and no AVF groups (between-group difference -0.67 mL/min/1.73 m(2)/year, 95% CI -1.43, 0.10). The 5-year absolute risk of dialysis was 87% (95% CI 84, 91) in the AVF group and 75% (95% CI 73, 77) in the no AVF group, and the 5-year survival probability was 77% (95% CI 70, 83) in the AVF group and 67% (95% CI 64, 69) in the no AVF group. CONCLUSIONS: In this study of patients with advanced chronic kidney disease, there was no evidence of a nephroprotective effect of AVF formation.