When Rao, et al, published the Kidney Donor Risk Index model (KDRI) (Transplantation 2009;88:231-236), they found coefficients for HLA match, cold ischemic time, en-bloc (EB) , and dual kidney in addition to the 10 variables included in the UNOS Kidney Donor Profile Index (KDPI) score. When UNOS implemented KDPI, it removed these variables without recalculating the model.

Methods: To evaluate the impact of omitting EB from KDPI, we compared the distribution of UNOS KDPI scores for 74 EB transplants conducted at our center from 1/1/13-9/30/14 to what the scores would have been had they included the EB covariate (-0.364) from the original model, as well as the impact of the difference under consent and allocation policies.

Results: The 74 PEB transplants had a median UNOS KDPI of 82.5 (SD 9.33, Range 58-98). When each KDPI was recalculated to include the EB coefficient, median score was 50.5 (SD 13.8, Range 22-84). Mean decrease in score was 30.5 (SD 4.9, Range 14-36). The adjustment to hazard ratio for graft failure(GF) for each is by definition .695 (e^-.364), so the UNOS KDPI model overstates the GF risk by 44% (1/(e^-.364) – 1) compared to the original model.

Distributions:

UNOS KDPI

EB-Adjusted KDPI

29 of 74 (39.2%) kidneys had KDPI > 85, which would have required additional consent and been allocated regionally had they been transplanted after 12/4/14. After adjusting for EB, none had KDPI > 85.

Conclusions: Centers including ours have shown excellent outcomes for pediatric EB transplants not usable as single kidneys, as the original KDRI score reflected in its coefficient predictive of less GF. In removing that variable without any other adjustment to the model, the UNOS KDPI score for EBs considerably overestimates the KDPI score and GF risk, subjecting those transplants to additional list consent and alternative allocation.

« Back to 2015 American Transplant Congress