What Matters for the Productivity of Kidney Exchange?

Abstract

Kidney exchange platforms serve patients who need a kidney transplant and who have a willing, but incompatible, donor. These platforms match patients and donors to produce transplants. This paper documents operational details of the three largest platforms in the United States. It then uses the framework developed in Agarwal et al. (2017) to examine how practical details influence platform productivity. The results show that reducing frictions in accepting proposed matches, frequent matching, and encouraging altruistic donors are important ways in which a platform can increase its productivity.

Market Failure in Kidney Exchange

“…9 See Abraham et al (2007), Ashlagi et al (2016), Anderson et al (2014), Dickerson et al (2012), and Agarwal et al (2018).…”

“…This step results in frictions within the system that reduce transplantation rates (Agarwal et al 2018). The parameters that govern these frictions are the time required for each of the two approval steps, the probability that a proposed transplant is abandoned in each step, and the duration for which a bridge donor is retained in the pool before donating her kidney to a patient on the deceased donor list.…”

Kidney Exchange: An Operations Perspective

“…In simulation studies, Agarwal et al (2018) and Ashlagi et al (2018b) measure the effect that batching policies have on efficiency (measured by the fraction of matched pairs and waiting times). The study uses APKD and MSTH data, which have different pool compositions.…”

“…Match offers are presently declined at a rate of about 25%-35% at the APKD and UNOS and around 20% at the NKR (Hanto et al 2008, Fumo et al 2015, Ashlagi et al 2018b, Agarwal et al 2018. This suggests that the data and preference criteria are too coarse.…”

“…The addition of easy-to-match pairs (or NDDs) reduces the need for long chains and makes short cycles more common (Ashlagi et al 2019a, Blum andMansour 2020). 57 Agarwal et al 2018 andAshlagi et al 2018b) Note. The left and right plots are generated without and with delays, respectively (delays due to receiving match offer responses and crossmatch tests).…”