Estimating risk of encapsulating peritoneal sclerosis accounting for the competing risk of death

Abstract

Background: Risk of encapsulating peritoneal sclerosis (EPS) is strongly associated with the duration of peritoneal dialysis (PD), such that patients who have been on PD for some time may consider elective transfer to haemodialysis to mitigate the risk of EPS. There is a need to determine this risk to better inform clinical decision making, but previous studies have not allowed for the competing risk of death.

Methods: This study included new adult PD patients in Australia and New Zealand (ANZ; 1990-2010) or Scotland (2000-08) followed until 2012. Age, time on PD, primary renal disease, gender, data set and diabetic status were evaluated as predictors at the start of PD, then at 3 and 5 years after starting PD using flexible parametric competing risks models.

Results: In 17 396 patients (16 162 ANZ, 1234 Scotland), EPS was observed in 99 (0.57%) patients, less frequently in ANZ patients (n = 65; 0.4%) than in Scottish patients (n = 34; 2.8%). The estimated risk of EPS was much lower when the competing risk of death was taken into account (1 Kaplan-Meier = 0.0126, cumulative incidence function = 0.0054). Strong predictors of EPS included age, primary renal disease and time on PD. The risk of EPS was reasonably discriminated at the start of PD (C-statistic = 0.74-0.79) and this improved at 3 and 5 years after starting PD (C-statistic = 0.81-0.92).

Conclusions: EPS risk estimates are lower when calculated using competing risk of death analyses. A patient's estimated risk of EPS is country-specific and can be predicted using age, primary renal disease and duration of PD.

Keywords: age; encapsulating peritoneal sclerosis; peritoneal dialysis; peritoneal membrane; prognosis.

FIGURE 1

Risk of EPS calculated using ‘standard’ and competing risks approaches. The figure shows the non-parametric estimates of EPS risk over cumulative PD exposure. The standard approach, using the Kaplan–Meier estimate, is shown by the broken line. The competing risks approach, using the cumulative incidence function estimate, is shown by the solid line.

FIGURE 2

ANZDATA risks of EPS and death over time. Cumulative risk of EPS (black area) and death (grey area) over time in patient groups at low risk of death (left-hand column) and high risk of death (right-hand column) demonstrated for patients starting PD (top) and patients 5 years after starting PD (bottom).

FIGURE 3

SSR risks of EPS and death over time. Cumulative risk of EPS (black area) and death (grey area) over time in patient groups at low risk of death (left-hand column) and high risk of death (right-hand column) demonstrated for patients starting PD (top) and patients 5 years after starting PD (bottom).