A kidney offer acceptance decision tool to inform the decision to accept an offer or wait for a better kidney

Abstract

We developed a kidney offer acceptance decision tool to predict the probability of graft survival and patient survival for first-time kidney-alone candidates after an offer is accepted or declined, and we characterized the effect of restricting the donor pool with a maximum acceptable kidney donor profile index (KDPI). For accepted offers, Cox proportional hazards models estimated these probabilities using transplanted kidneys. For declined offers, these probabilities were estimated by considering the experience of similar candidates who declined offers and the probability that declining would lead to these outcomes. We randomly selected 5000 declined offers and estimated these probabilities 3 years post-offer had the offers been accepted or declined. Predicted outcomes for declined offers were well calibrated (<3% error) with good predictive accuracy (area under the curve: graft survival, 0.69; patient survival, 0.69). Had the offers been accepted, the probabilities of graft survival and patient survival were typically higher. However, these advantages attenuated or disappeared with higher KDPI, candidate priority, and local donor supply. Donor pool restrictions were associated with worse 3-year outcomes, especially for candidates with high allocation priority. The kidney offer acceptance decision tool could inform offer acceptance by characterizing the potential risk-benefit trade-off associated with accepting or declining an offer.

Keywords: clinical decision-making; donors and donation; health services and outcomes research; kidney (allograft) function/dysfunction; kidney transplantation/nephrology; organ acceptance; organ procurement and allocation.

Figure 1

An illustration of the process for estimating graft survival and patient mortality with the offer acceptance decision tool. For example, after an offer is declined, we consider the likelihood of deceased donor transplant, living donor transplant, death, and removal from the list for other reasons. We combine the likelihood of being removed from the waiting list with the likelihood that the corresponding removal results in graft survival or patient mortality. Patient survival is equal to 1 minus patient mortality. DD, deceased donor; GS, graft survival; LD, living donor; PM, patient mortality; PS, patient survival.

Figure 2

The calibration and estimated predicted error for the approach used to estimate the probability of graft survival and patient survival 3 years after declined offers. Each circle represents the estimate within a 5% quantile bin of offer number. The solid lines correspond to penalized splines of the estimated effects within each bin.

Figure 3

The estimated differences in predicted graft survival and patient survival between accepting versus declining an offer, 3 years after the offer. The dashed lines represent the 95% confidence intervals for the mean difference across offer number. The evaluated offers were 5000 randomly selected declined offers. Estimated differences above the dotted horizontal line indicated that acceptance led to better predicted outcomes than declining the offer. KDPI, kidney donor profile index; GS, graft survival; PS, patient survival.

Figure 4

The estimated differences in predicted graft survival and patient survival 3 years after an offer with and without restrictions on the donor pool. The restrictions consist of KPDI cutoffs of 70%, 80% and 90% with a maximum CIT of 30 hours. The dashed lines represent the 95% confidence intervals for the mean difference across offer number. The evaluated offers were 5000 randomly selected declined offers. Estimated differences below the dotted horizontal line indicated that the donor pool restrictions reduced the probability of graft survival and patient survival. CIT, cold ischemia time; KDPI, kidney donor profile index; GS, graft survival; PS, patient survival.

Figure 5

Offer acceptance decision tool screen shot showing that acceptance led to an 85% probability of graft survival 3 years after the offer, compared with a 46% probability if the offer was declined.

Figure 6

Offer acceptance decision tool screen shot showing that restricting the donor pool with a maximum KDPI of 85% and CIT of 20 hours deceased the probability of graft survival to 33% 3 years after the offer was declined.