Long-term Mortality Risk Following a Transition from Peritoneal Dialysis to Hemodialysis: Analysis of the Canadian Organ Replacement Register

Transitions to facility hemodialysis (HD) are a major cause of peritoneal dialysis (PD) termination. These transitions are challenging for patients and associated with a high mortality in the immediate aftermath. Yet, long-term clinical outcomes after these transitions are poorly known. We hence aimed at evaluating the mortality incidence and risk factors following a PD-to-HD transition in Canada.

Analysis of the Canadian Organ Replacement Registry (CORR). We included patients who initiated PD (defined as at least 30 consecutive days of PD) between January 1st 2005 and December 31st 2018. A transition to facility HD was defined as the receipt of at least 90 consecutive days of HD after a PD episode. Patients were followed for mortality until December 31st 2019 and were censored if they received a kidney transplant, returned to PD, transitioned to home hemodialysis, or experienced kidney recovery. All PD episodes were considered for patients with multiple PD-to-HD transitions. Deaths observed in the first 90 days of HD were attributed to PD and were not considered for this study. Mortality was examined using cumulative incidence and instantaneous mortality rates. Risk factors for post-transition mortality were examined using mixed-effect Cox models with random effects at center- and individual-level. Potential risk factors included demographics, comorbidities, dialysis vintage, center size, and hospitalizations while on PD in the year preceding the transition.

From 63,327 participants in the CORR, 18,207 (29%) initiated PD. During a median follow-up of 3.9 years after PD initiation, 6,111 patients experienced a total of 6,464 transition to facility HD. PD-to-HD occurred after 1.4 years on PD [IQR 0.6 to 2.7] and patients were followed for 1.4 years [IQR 0.6 to 3.2] in facility HD. Cumulative mortality incidence were respectively of 13% and 56% at 1- and 5-years after the PD-to-HD transition. Instant mortality rates (Figure) reached their maximal value at start of follow-up (0.26 deaths/patient-year in the fourth month after the transition) and their minimal value at the 14th month following the transfer (0.16 deaths/patient-year) after which they started to increase again. In fully adjusted Cox mixed-effects models, advanced age (HR= 1.37 95th CI [1.31-1.43] by decade), facility HD prior to PD initiation (1.26 [1.14-1.39]), kidney replacement therapy vintage (1.13 [1.07-1.13] per year), coronary artery disease (1.26 [1.13-1.41]), diabetes (1.34 [1.14-1.58]), peripheral artery disease (1.25 [1.10-1.43]), prior stroke (1.24 [1.08-1.43]) and experiencing two or more hospitalizations in the year before the transfer (1.60 [1.42-1.80]; compared to no hospitalization) were associated with increased mortality. In contrast, experiencing one hospitalization in the year preceding the transfer and center size (categorized by annualized incident PD patients) were not associated with mortality after the transition.

Transitions from PD to HD are associated with heightened mortality for up to a year. In addition to known mortality risk factors, we identified repeated hospitalizations prior to transfer as a novel predictor of mortality after a PD-to-HD transition.