The process of organ transplantation is the culmination of many unique and differentiated individual decisions. Examples of individual decisions on the patient side are the decision to be placed on an organ waiting list, the surgeon’s decision to select desired organ characteristics [ie, Extended Criteria Donor (ECD) organ], the decision to accept or reject an organ that is offered and the type of posttransplant care. Examples on the donor side are the decision to be placed on the donor registry, the next-of-kin donation decisions and the donation decisions of living donors. Developing a better understanding of these individual decisions will facilitate the creation of a more efficient organ transplantation process. A number of neoclassical and behavioral economic principles can be used to address this need. This primer serves as an illustration of how applying these economic principles can be used to advance our understanding of the organ transplant process.
The foundation of neoclassical, or traditional, economic theory is expected utility theory.1 However, a number of behavioral anomalies (ie, Allias Paradox2) have arisen which cannot be explained by expected utility theory and the field of behavioral economics has arisen to address these behavioral anomalies (i.e., prospect theory3). In our discussion, we will first discuss the tools of neoclassical economic theory and their application to organ transplantation and then expand the discussion to theories in behavioral economics that may enrich our understanding.
Neoclassical or Traditional Economic Theory
The first step in applying economic principles to individual decision-making is to understand the neoclassical terms of utility, indifference curves, and utility maximization. Utility is an ordinal measure of well-being that defines the benefit an individual assigns to a given outcome. The concept of utility was first introduced by Bernoulli4 and the work of von Neumann and Morgenstern1 axiomatically formalized expected utility theory in their development of the expected utility theorem. An indifference curve is a graphical representation of preferences that indicates all consumption bundles that yield the same level of utility for an individual.5Figure 1 illustrates both of these functions using information that is relevant to organ transplantation.5
The utility function illustrated in Figure 1 represents the surgeon's utility function when they are tasked with the decision to accept or reject an organ for their patient. For simplification, we will assume that the surgeon serves as a perfect agent for the patient (ie, makes decisions that are in the best interest of the patient) and therefore the surgeon's utility function captures not only the surgeon’s preferences but also the patient's preferences as well. The surgeon's utility function is defined over the Kidney Donor Profile Index (KDPI) used by the United Network for Organ Sharing (UNOS) to measure the quality of a donated organ. The KDPI is derived from the Kidney Donor Risk Index developed by Rao et al6 and based on estimates of graft survival using donor-specific covariates in a Cox proportional hazard model. In this sense, the measure is highly objective and it provides rapid information to a surgeon on the quality of the organ relative to the other organs obtained in the previous year. By construction, the lower the KDPI the higher the quality of the organ as a KDPI of “X” value indicates that only “X” percent of the organs obtained within the last year have a higher rate of graft survival. The lowest quality organ possesses a KDPI value of 100 and subsequently has the lowest level of utility assigned to it. A KDPI value of 100 indicates that 100% of the donated kidneys have a higher graft survival rate than the kidney in question. As the quality of the donated organ increases so does the surgeon's utility. The surgeon's utility function also illustrates another important feature of many utility functions used in economics, the presence of decreasing marginal utility. As the quality of the donated organ increases the additional utility derived from using the organ increases very rapidly at higher KDPI values (poorer organ quality), but as we begin to reach higher quality organs the additional gains in utility derived from a small increase in the quality diminishes. Given this shape, we would expect that a surgeon would be much more responsive to small increases in organ quality when the perceived quality is low versus when it is high.
The indifference curves illustrated (purely illustrative) in the right panel illustrate all possible combinations of 2 important variables in transplantation, histocompatibility (antigen mismatches), and the KDPI that generate the same level of utility for a surgeon. They also illustrate the ordinal nature of utility, a tenant of expected utility theory,1 as when both the histocompatibility and KDPI increase in quality the surgeon’s utility increases. The slope of the indifference curve is referred to as the marginal rate of substitution, and it indicates an individual's willingness to trade one good for another while maintaining the same level of utility. This trade-off also varies depending on where one is on the indifference curve. For instance, if we focus on the indifference curve for a utility level of one, we can see that when the histocompatibility is very high (close to 0), yet the KDPI indicates a very poor organ (close to 90), the surgeon is willing to trade off a high level of histocompatibility for a small increase in the quality of the donated organ. From a neoclassical economics perspective, we assume that surgeons make these trade-offs, and we endeavor to estimate these utility trade-offs to develop a more robust profile of surgeon preferences.
Based on expected utility theory,1 neoclassical economics stipulates that rational agents make choices to maximize their level of utility subject to a budget constraint. However, McFadden7 illustrated that the discrete choices made by agents can also be framed in the context of utility maximization. For instance, the transplant surgeon's decision to accept or reject an organ that has been offered to a patient is a discrete choice that is consistent with utility maximization. A surgeon will accept the organ for a patient if and only if the utility they derive from that decision exceeds the utility derived from not accepting the organ. This might seem to imply that a surgeon should always accept an organ for their patient but this is not necessarily true. When a surgeon is offered an organ for a patient, the utility of accepting the organ includes a multitude of factors, such as the expected gains in the patient's health status and the expected length of time that the graft will survive, which is based on the organ quality and its compatibility with the patient. The utility of not accepting the organ also includes a multitude of factors, such as the patient's health, the expected length of time until another organ will be offered to the patient, the expected quality of the organ that will be offered in the future, and the surgeon's expectations regarding how others on the organ waiting list may benefit from the organ offered. Whenever a surgeon elects to not accept an organ, they are implicitly stating that the utility of not accepting the organ exceeds the utility of accepting the organ.
The expectations that a surgeon must form to maximize utility indicate that there exists a substantial degree of uncertainty in the organ utilization process. The presence of uncertainty introduces another important neoclassical economic concept, risk aversion.5,8,9 Risk aversion is based on the premise that individuals are not risk-neutral agents in the presence of uncertainty. A simple example of risk aversion can be illustrated using a gamble based on an unbiased coin flip. In the gamble, an individual earns $10 if a coin flip comes up heads and $0 if it comes up tails. The expected value of the gamble is $5. However, if an individual is only willing to pay less than $5 to take the gamble, they are said to be a risk-averse agent. If the individual is willing to pay more than $5 they are risk-loving and if exactly $5 they are risk-neutral. Risk aversion arises in neoclassical economic terms because the utility of expected wealth from taking the gamble is greater than the expected utility of wealth.5 This is a result of concavity in an agent's utility function. Figure 2 provides an illustration of risk aversion as it may apply to the organ utilization process.2
The graphical illustration of risk aversion in organ utilization is based on the expectations that a transplant surgeon must form with regard to the expected graft survival of an organ. For simplification, we suppose that there are 2 possible outcomes, the graft survives 2 years and the graft survives 10 years, which occur with equal probability. The expected number of years is 6 years, but the utility that the surgeon derives from 6 years of graft survival, U(E(years)), is greater than the expected utility in the presence of this uncertainty, E(U(years)). This directly implies that the surgeon is risk averse because they would prefer to accept an organ with a certain graft survival length slightly less than 6 years (ie, 5.5 years) versus being uncertain over the length of the graft survival. This is not to say that all surgeons are risk averse; they could also exhibit risk-neutral or risk-loving behavior. A risk-neutral surgeon would value E(U(years)) and U(E(years)) equally. A risk-loving surgeon would illustrate preferences that are the opposite of those depicted in Figure 2 as E(U(years)) would be greater than U(E(years)) and the utility function would be convex. Whether or not a surgeon is risk-averse, risk-loving, or risk-neutral is an empirical question that can be answered using data on a surgeon's utilization decisions. Moreover, ignoring the possibility of these preferences may dramatically impact our assumptions regarding organ utilization in transplant policy.
Behavioral Economic Theory
As mentioned earlier, there are a number of behavioral anomalies that cannot be explained by neoclassical economic theory (ie, Allias Paradox2), and behavioral economic theories were created to fill this void, one of which is prospect theory.3 Under prospect theory, decision agents use behavioral reference points to compare outcomes. This generates a notion of “losses” or “gains” relative to the reference point. Prospect theory states that agents make decisions in 2 stages defined as the editing and evaluation phases.3 The application of a behavioral reference point arises in the editing phase of the decision process where agents will determine whether or not the risky choice involves a gain or loss relative to the reference point. After the editing phase, agents evaluate the prospects through the weighting of probabilities and the assignment of value, note not utility, to each outcome relative to the reference point.3 Furthermore, under prospect theory, it is hypothesized that the value functions are concave over the gains and convex over losses, thus inducing the risk aversion in the gains domain and risk-seeking behavior in the loss domain as well as a decreasing marginal effect on value as the levels rise.3
An example of behavioral reference points in transplantation can be observed by looking at the organ utilization process. The process by which a surgeon forms expectations on the suitability of an organ is based on the agglomeration of many pieces of information that is presented to them at the time the organ is being offered (ie, DonorNet). This is a massive flow of information that must be processed rapidly. Behavioral economics would posit that surgeons use behavioral reference points to facilitate this process and develop heuristics that assist in the processing of information. A behavioral reference point is a focal point in the decision space at which the surgeon's utility function possesses a discontinuous shift in marginal utility and is a cornerstone of prospect theory.3,10 There are a multitude of potential reference points that may exist. For instance, Schnier et al.11 posit that the regulations used by the Centers for Medicare and Medicaid Services (CMS) to monitor the performance of transplant centers may in fact generate behavioral reference points. Alternatively, a physician's expectation regarding the quality of an unknown future organ offer may also provide a reference point (Schnier et al, 2015, unpublished data). Figure 3 graphically illustrates a potential organ quality reference point using the KDPI variable.12
The hypothesized reference point is the KDPI value of 50 because this is the mean quality of donated organs within the United States (Schnier KE et al, 2015, unpublished data). This reference point is based on a surgeon's expectation of quality for an unknown future organ offer that is consistent with the rational expectations approach to reference points proposed by Köszegi and Rabin.13 The figure illustrates that organs offered with a KDPI greater than 50 are viewed as a “loss” relative to the mean donor, whereas for a KDPI less than 50, the organs are viewed as a “gain” relative to the mean donor. The benefit of the KDPI is that it objectively allows surgeons to evaluate the marginal increases in the probability of a graft failure. However, the presence of a reference point indicates that surgeons may focus on behavioral triggers to indicate the quality of an organ with the reference point informing them if the organ offered is better or worse than the reference donor. For instance, if a surgeon is offered an organ with a KDPI of 60, this transforms to a 10-point “loss” in the organ quality, whereas if they are offered an organ with a KDPI of 40 this is a 10-point “gain” in organ quality. Knowledge of these reference points can be used to develop alternative mechanisms of providing information that do not generate stark behavioral responses.
The utility function in Figure 3 graphically illustrates another important behavioral economics concept that may be important to organ transplantation, the concept of loss aversion.3,10 Loss aversion stipulates that the utility function is steeper in the loss domain than in the gains domain. Therefore, losses are weighted more than an equivalent gain. In Figure 3, the utility assigned to the highest quality donor is 2, for the reference donor it is 0, and to lowest quality donor, it is negative 3. The difference in the KDPI between the highest and lowest quality organ relative to the reference organ is the same, but the utility of the loss is weighted more than the respective gain in quality. The presence of loss aversion alters the shape of an individual's indifference curve, and it can apply to many of the attributes that a decision agent has preferences over. Figure 4 graphically illustrates a set of revised indifference curves for a decision agent who has 2 reference points as well as exhibits loss aversion for both attributes.10 To simplify the illustration, we have linearized the indifference curves.10
The indifference curves in Figure 4 are contextualized so that the reference donor has a histocompatibility of 3 and a KDPI of 50. Loss aversion implies that the indifference curves are steeper in the loss domain than the gains domain. For instance, if we focus on the utility curve U = 1, we can see how being in the loss and gains domains for histocompatibility and the KDPI impact the shape of the indifference curves. Along line segment AB, the decision agent is only in the loss domain for the KDPI; along BC, they are in the loss domain for both histocompatibility and the KDPI; and along CD, they are only in the loss domain for histocompatibility. The steepness of the indifference curves reflects when the decision agent is above or below the respective reference point for either histocompatibility or the KDPI. For instance, segment EF on the U = 3 indifference curve has the least steep slope of the other indifference curves, and this is because in this region, the donor is above both the histocompatibility and the KDPI reference points. An interesting feature of this model is that it is possible for a decision agent to possess multiple overlapping reference points and exhibit corresponding loss aversion when they are below the reference points. In the context of transplantation, this would imply that surgeons may decompose the attributes of an offered organ (ie, ECD status, histocompatibility, PRA, KDPI, and so on) and construct reference points for each of them. The existence of these overlapping reference points and the potential for loss aversion may have a profound impact on our assumptions regarding surgeon behavior.
The application of the neoclassical and behavioral economic theories outlined in this primer provide only a small number of theoretical tools that can be taken from economics to advance our understanding of the individual decisions made in organ transplantation. It is our expectations that many transplant surgeons apply these theories to the decisions they make for themselves and their patients. Given this, there is a growing need for considerable research at the intersection of economics and transplantation so that we can learn from these behavioral theories and develop new institutions that incorporate these behavioral responses. In a follow-up to this primer, we will be discussing some additional behavioral economic concepts that may influence the transplant process. More specifically, we will discuss fairness and other regarding preferences, framing effects, the law of small numbers and hyperbolic and quasi-hyperbolic discounting. A limitation of this current discussion is that is has predominately focused on economic efficiency based on individual preferences. An additional concern that is beyond the scope of this discussion is equity. However, even if equity concerns dominate over efficiency, this would manifest itself in the preferences one exhibits, and expansions of the theory may incorporate these preferences. This will become more evident in our follow-up primer. It is our goal that through a broad description of these terms, we can advance our understanding of how some of the key decisions in the transplant process (ie, organ utilization, organ donation) are made and advance public policy by making the decisions of individuals an endogenous component of the analysis.
1. von Neumann J, Morgenstern O. Theory of Games and Economic Behavior
. 1st ed. Princeton University Press: Princeton, NJ; 1944.
2. Allias M. LeComportement de l'Homme Rationnel devant le Risque: Critique des Postulats et Axiomes de l'Ecole Americaine. Econometrica
. 1953; 21: 503–546.
3. Kahneman D, Tversky A. Prospect theory: an analysis of decisions under risk. Econometrica
. 1979; 47: 263–291.
4. Bernoulli D. Specimen Theoriae Novae de Mensura Sortis. Commentarii Academiae Scientiarum Imperialis Petropolitanae
. 1738; 5: 175–192.
5. Varian HR. Intermediate Microeconomics—A Modern Approach
. W.W. Norton & Company, Inc: New York, NY; 1987.
6. Rao PS, Schaubel DE, Guidinger MK, et al. A comprehensive risk quantification score for deceased donor kidneys: the Kidney Donor Risk Index. Transplantation
. 2009; 88: 231–236.
7. McFadden D. Conditional Logit Analysis of Qualitative Choice Behavior. In: P. Zarembka, ed. Frontiers in Econometrics
. Academic Press; New York, NY: 1974: 105–142.
8. Arrow KJ. Aspects of the Theory of Risk Bearing. In: The Theory of Risk Aversion.
Helsinki: Yrjo Jahnssonin Saatio. 1965.
9. Pratt JW. Risk aversion in the small and in the large. Econometrica
. 1964; 32: 122–136.
10. Tversky A, Kahneman D. Loss aversion in riskless choice: a reference-dependent model. Q J Econ
. 1991; 106: 1039–1061.
11. Schnier KE, Cox JC, McIntyre C, et al. Transplantation at the nexus of behavioral economics and healthcare delivery. Am J Transplant
. 2013; 13: 31–35.
12. Tversky A, Kahneman D. Advances in prospect theory: cumulative representation of uncertainty. J Risk Uncertain
. 1992; 5: 297–323.
13. Köszegi B, Rabin M. A model of reference-dependent preferences. Q J Econ
. 2006; 121: 1133–1165.