Comparative studies of dialysis therapies should reflect real world decision-making

The incidence and prevalence of end-stage renal disease (ESRD) continues to rise. While transplantation is the preferred therapy for kidney failure, there is a shortage of donor organs, and the majority of patients will be treated with either peritoneal dialysis (PD) or hemodialysis (HD). Randomized controlled trials comparing patient outcomes on PD and HD are not likely to be successful, as individuals who are educated about their treatment options generally develop a strong preference for one therapy over the other and will not consent to randomization. As a result, prospective cohort studies are frequently the strongest study design available to compare outcomes between dialysis modalities. Previous studies have provided important insights into the relative merits of the 2 therapies. However, they have examined outcomes in relatively heterogeneous groups of ESRD patients and are generally not designed in a manner that mirrors clinical decision-making. We explore several key methodological challenges in the design of observational research in ESRD with a focus on minimizing selection bias and making studies more relevant to the practicing nephrologist. We emphasize that incident patients are preferred in most comparative studies of dialysis modalities. We argue that analyses comparing the outcomes of renal replacement therapy (RRT) modalities should include patients eligible for the therapies being compared and that the way that patients are assigned to treatment groups should reflect decision-making in clinical practice. Finally, the point at which baseline characteristics are measured and we begin tracking patients for the occurrence of outcomes should be chosen carefully.