Background: Previous studies comparing the outcomes in haemodialysis (HD) with those in peritoneal dialysis (PD) have yielded conflicting results.
Methods: The aim of the study was to compare the survival of planned HD versus PD patients in a cohort of adult incident patients who started renal replacement therapy (RRT) between 2006 and 2008 in the nationwide REIN registry (Réseau Epidémiologie et Information en Néphrologie). Patients who started RRT in emergency or stopped RRT within 2 months were excluded. Adjusted Cox models, propensity score matching and marginal structural models (MSMs) were used to compensate for the lack of randomization and provide causal inference from longitudinal data with time-dependent treatments and confounders including transplant censorship, modality change over time and time-varying covariates.
Results: Among a total of 13 767 dialysis patients, 13% were on PD at initiation of RRT and 87% were on HD. The median survival times were 53.5 months or 4.45 years and 38.6 months or 3.21 years for patients starting on HD and PD, respectively. Regardless of the model used, there was a consistent advantage in terms of survival for HD patients: hazard ratio (HR) 0.76 [95% confidence interval (95% CI) 0.69-0.84] with the Cox model using propensity score; HR 0.67 (95% CI 0.62-0.73) in the Cox model with censorship for each treatment change; and HR 0.82 (95% CI 0.69-0.97) with MSMs. However, MSMs tended to reduce the survival gap between PD and HD patients.
Conclusion: This large cohort study using various statistical methods to minimize the bias appears to demonstrate a better survival in planned HD than in PD.