Beyond "Median Waiting Time": Development and Validation of a Competing Risk Model to Predict Outcomes on the Kidney Transplant Waiting List

Abstract

Background: Median historical time to kidney transplant is misleading because it does not convey the competing risks of death or removal from the waiting list. We developed and validated a competing risk model to calculate likelihood of outcomes for kidney transplant candidates and demonstrate how this information differs from median time to transplant.

Methods: Data were obtained from the US Scientific Registry of Transplant Recipients. The retrospective cohort included 163 636 adults listed for kidney transplant before December 31, 2011. Predictors were age, sex, blood type, calculated panel-reactive antibodies, donation service area, dialysis duration, comorbid conditions, and body mass index. Outcomes were deceased or living donor transplant, death or removal from the list due to deteriorating medical condition, or removal due to other reasons. We calculated hazards for the possible outcomes, then the cumulative incidence function for a given candidate using competing risk methodology. Discrimination and calibration were assessed through C statistics and calibration plots for each cause-specific Cox proportional hazard model.

Results: C statistics ranged from 0.64 to 0.73. Calibration plots showed good calibration. The competing risk model shows probability of all possible outcomes for up to 12 years given a candidate's characteristics, contrasted with the median waiting time for that candidate's donation service area.

Conclusions: A competing risk model conveys more relevant information than the median waiting time for a given transplant center. This model will be updated to create a calculator reflecting the most recent outcomes and changes in allocation policy. It illustrates the conversations that should be initiated with transplant candidates.

FIGURE 1

For the 163 636 patients in the cohort: A, Kaplan-Meier survival curves for each of the possible outcomes on the waiting list; and B, cumulative incidence curve for each of the competing outcomes on the waiting list, contrasting the information provided by each method.

FIGURE 2

Sample waitlist calculator output in 2 different DSAs for a 50-year-old man with a BMI of 28 kg/m², PRA/CPRA of 0, blood type A, not on dialysis, with no comorbid conditions, with no living donors, who is not willing to accept an ECD kidney, the same theoretical patient shown in Table 4.