A Decade of Peritoneal Dialysis Attrition: A Single-Centre Canadian Retrospective Cohort Study
Ashley Benchluch, Christopher Gayowsky, Bourne L. AugusteBackground
Peritoneal dialysis (PD) attrition is commonly used as a marker of program quality. However, all-cause attrition encompasses competing outcomes including transplantation, death, and technique failure, which carry fundamentally different clinical and programmatic implications. Cause-specific analysis is therefore essential for accurate program evaluation.
Objective
To characterize cause-specific PD attrition over a 10-year period and identify predictors of PD discontinuation using a competing risks framework.
Design
Single-centre retrospective cohort study.
Setting
Sunnybrook Health Sciences Centre, a tertiary academic centre in Toronto, Canada, with active referral relationships to university-affiliated kidney transplant centres.
Patients
All adult patients (age ≥18 years) who initiated PD between January 1, 2014 and December 31, 2024 with a minimum of one month of follow-up (N = 230).
Measurements
Cause-specific PD exit (transfer to hemodialysis [HD], kidney transplantation, death, facility transfer, kidney recovery). Among patients transferring to HD, clinical reasons for transfer were recorded. Predictors assessed included age, sex, diabetes status, PD modality, assisted PD use, and residual kidney function.
Methods
Cause-specific cumulative incidence functions were estimated using the Aalen-Johansen estimator. For clarity, only the three principal causes, transfer to hemodialysis, kidney transplantation, and death, are displayed. Multivariable Cox proportional hazards models were fit within a cause-specific hazard framework to evaluate predictors of PD discontinuation.
Results
During follow-up, 162 patients (70.4%) discontinued PD. In the early follow-up period, transfer to HD was the predominant cause of PD exit. Over longer follow-up, kidney transplantation emerged as the dominant cause of PD discontinuation, with its cumulative incidence exceeding that of HD transfer. Among 53 patients who transferred to HD, the most common reasons were peritonitis or other infection (37.7%), catheter non-function (20.8%), and under-dialysis (17.0%). In cause-specific analysis, male sex was independently associated with a higher hazard of technique failure (HR 3.15, 95% CI 1.62-6.12), whereas older age was associated with a lower hazard of both technique failure and transplantation but a higher hazard of death, offsetting effects that produced a null age association in a combined model. Kidney transplantation was the single most common cause of PD exit (54, 33.3%).
Limitations
Single-centre retrospective design with limited sample size. Important comorbidities including cardiovascular disease and frailty were not systematically captured. Non-medical contributors to attrition were not assessed. High transplant referral activity may limit generalizability.
Conclusions
In a Canadian PD program with active transplant referral relationships, kidney transplantation was the dominant driver of PD discontinuation over longer follow-up, underscoring that all-cause attrition cannot be interpreted uniformly across programs. Cause-specific, case-mix adjusted benchmarks are needed to accurately evaluate PD program performance.