Is the Kidney Donor Risk Index a Useful Tool in Non-US Patients?

Abstract

Background:Deceased donor kidney allocation in the United States is guided by the Kidney Donor Risk Index (KDRI). The generalizability of the KDRI beyond the United States has not been widely studied.Objective:To assess the generalizability of the KDRI in a cohort of non-US (Canadian) deceased donor kidney transplant recipients.Design:Population-based retrospective cohort study.Setting:Ontario, Canada.Patients:Recipients of deceased donor kidneys from January 1, 2005, to March 31, 2011.Methods:Using administrative data, we analyzed a cohort of deceased donor kidney recipients in Ontario, Canada. The Kaplan-Meier method and Cox proportional hazards models were used to assess the relationship between KDRI and the outcomes of graft loss and death. KDRI was modeled continuously and categorically. The ability of models with KDRI to predict recipient outcomes beyond donor age was also explored. Model discrimination was assessed using c-statistics, evaluated at 5 years of follow-up.Results:A total of 1299 consecutive deceased donor kidney transplant recipients were included. The median follow-up was 5.5 years. Mean donor age increased from 27 to 64 years across ascending KDRI quintiles. The adjusted relative hazards (95% confidence interval) for total graft loss from Q2 to Q5 (referent = Q1) were 1.27 (0.89-1.80), 1.58 (1.13-2.22), 1.43 (1.01-2.02), and 2.15 (1.54-2.99), respectively. Increased relative hazards across KDRI quintiles were also observed for death-censored graft loss, but not death with graft function. All-cause mortality was increased for the highest KDRI quintile only. In this cohort, a model with KDRI performed better than a model with donor age alone (P = .009).Limitations:Large health care databases may have precluded the complete capture of covariate data.Conclusions:In conclusion, the KDRI is generalizable to Canadian patients in Ontario and may help inform risk assessment beyond donor age. The performance of KDRI in other non-US settings, and the need for additional model refinement, warrants further study.