Dialysis modality, vascular access and mortality in end-stage kidney disease: A bi-national registry-based cohort study

Abstract

Aim: There remains debate on which dialysis modality offers better survival outcomes for patients. We compare the survival of patients undergoing home haemodialysis (HD) with a permanent vascular access, facility HD with a permanent vascular access, facility HD with a central venous catheter or peritoneal dialysis.

Methods: We considered adult patients from the Australia and New Zealand Dialysis and Transplant Registry who commenced dialysis between 1 October 2003 and 31 December 2011. Patients were followed until death, transplant, loss to follow-up or 31 December 2011. Marginal structural models for mortality were used to account for time-varying treatment, comorbidities and baseline covariates. Unmeasured differences between treatment groups may remain even after adjustment for measured differences, so the potential effects of unmeasured confounding were explicitly modelled.

Results: There were 20,191 patients who underwent ≥90 days of dialysis (median 2.25 years, interquartile range 1-3.75 years). There were significant differences in age, gender, comorbidities and other variables between treatment groups at baseline. Thirty per cent of patients had at least one treatment change. Relative to facility HD with permanent access, the risk of death for home HD patients with a permanent access was lower in the first year (at 9 months: hazard ratio 0.41, 95% CI 0.25-0.67, adjusted for all baseline covariates). Findings were robust to unmeasured confounding within plausible ranges.

Conclusion: Relative to facility HD with permanent vascular access, home HD conferred better survival prospects, while peritoneal dialysis was associated with a higher risk and facility HD with a catheter the highest risk, especially within the first year of dialysis.

Keywords: haemodialysis; marginal structural model; mortality; peritoneal dialysis.