Response to: Early Refusal Pattern Phenotyping as a Surrogate for Organ Quality Assessment in Kidney Allocation

We appreciate the thoughtful commentary by Kashiv et al1 on our recent article in Transplantation.2 We find ourselves in full agreement with the points raised in their letter about how early behavioral indicators could augment current allocation workflows in India. We would like to highlight 2 of their observations on refining early offer phenotyping to further improve the allocation system. First, as Kashiv et al rightly suggest, organ-related refusal reasons or simultaneous multipatient refusals that arrive early in the allocation process could provide a window for more thorough immunologic assessment for high-risk donor-recipient pairs, ultimately enabling individualized, risk-stratified transplantation. Second, we agree with the authors’ emphasis on the translatability of these results. Tailoring these insights from refusal patterns to the unique donor demographics, allocation algorithms, and transplant center acceptance behaviors of different nations is essential for improving kidney utilization globally. Additionally, we echo Kashiv et al regarding need for more granular refusal codes, although this also presents several distinct challenges. In the United States, the Organ Procurement and Transplantation Network Data Advisory Committee is currently reviewing refusal codes “for greater specificity and standardization.”3 Yet, simply adopting these revised codes will not suffice; success depends on transplant centers using them effectively and having the flexibility to input multiple codes. When used transparently, these refusal phenotypes can help prevent kidney nonuse by informing real-time updates to the allocation priority list. Recorded refusal data can be integrated alongside existing clinical data to guide subsequent offers. Beyond coding, allocation systems can explore innovative ways to both capture both patient and transplant center preferences and enhance system efficiency. For example, a center declining an offer could be asked if they would still be interested in the organ for a different patient. However, these approaches should be managed with caution due to Goodhart’s Law: the principle that once a measure becomes a metric, it ceases to be an effective measure. Although current refusal codes provide significant value in identifying kidneys at risk of nonuse, their incorporation into formal performance metrics may incentivize strategic reporting. This behavior could compromise the integrity of the data; in this context, simultaneous multipatient refusals may be more robust than organ-related refusal reasons. Global improvements in deceased donor transplantation access depend on both collaborative dialogue of this nature and navigating the nuances of each country’s allocation system. Although we must remain mindful of current limitations in current refusal codes, the phenotyping of refusals—whether by organ versus patient-related refusal reasons or single- versus multipatient simultaneous refusals—provides a vital framework to capture expert clinical judgement early in the allocation process. Ultimately, acknowledging these complexities ensures that these findings can be effectively translated into clinical practice worldwide. ACKNOWLEDGMENTS The authors acknowledge Xingxing Cheng, Marc Melcher, Sanjit Neelam, Michael Rees, Alvin Roth, Paulo Somaini, and Joachim Studnia for their intellectual contributions to this letter.