Differences in initial hemodialysis vascular access use among glomerulonephritis subtypes in the United States

Background The type of vascular access used for hemodialysis affects patient morbidity and mortality. Whether vascular access types differ by glomerulonephritis (GN) subtype in the US hemodialysis population has not been investigated. Study Design Cross-sectional observational study. Setting & Participants We identified all adult (aged ≥ 18 years) patients within the US Renal Data System who initiated hemodialysis therapy from July 2005 through December 2011 with a diagnosis of end-stage renal disease attributed to any of 4 primary (focal segmental glomerulosclerosis, immunoglobulin A nephropathy [reference group], membranous nephropathy, and membranoproliferative GN) or 2 secondary (lupus nephritis and vasculitis) GN subtypes. Predictor GN subtype. Outcomes ORs with 95% CIs for arteriovenous fistula versus central venous catheter (CVC) use and for arteriovenous graft versus CVC use were computed using multinomial logistic regression, with adjustment for demographic, socioeconomic, comorbidity, and duration of nephrology care covariates. Results Among 29,015 patients, CVC use at initiation of hemodialysis therapy was substantially higher in patients with lupus nephritis (89.2%) or vasculitis (91.2%) compared with patients with primary GN subtypes (72.7%-79.8%). After adjustment and compared with patients with immunoglobulin A nephropathy, patients with lupus nephritis or vasculitis were as likely to have used an arteriovenous graft (ORs of 0.94 [95% CI, 0.70-1.27] and 0.80 [95% CI, 0.56-1.13], respectively) but significantly less likely to have used an arteriovenous fistula (ORs of 0.66 [95% CI, 0.57-0.76] and 0.54 [95% CI, 0.45-0.63], respectively), whereas patients with any comparator primary GN subtype were at least as likely to have used either of these 2 access types. Limitations Potential misclassification of exposure; residual confounding by unmeasured covariates; inability to determine causes of observed associations; lacking longitudinal data for vascular access use. Conclusions Significant differences in vascular access distributions at initiation of hemodialysis therapy are apparent among GN subtypes. The unacceptably high use of CVCs in patients with lupus nephritis and vasculitis is particularly concerning. Further studies are needed to identify any potentially modifiable factors underlying these findings.