Patient Preferences for Waiting Time and Kidney Quality : Clinical Journal of the American Society of Nephrology

Journal Logo

Original Article: Transplantation

Patient Preferences for Waiting Time and Kidney Quality

Mehrotra, Sanjay1; Gonzalez, Juan Marcos2; Schantz, Karolina1; Yang, Jui-Chen2; Friedewald, John J.3; Knight, Richard4

Author Information
CJASN 17(9):p 1363-1371, September 2022. | DOI: 10.2215/CJN.01480222



Approximately 20% of deceased donor kidneys are discarded in the United States each year (1). Studies suggest that utilizing more of these kidneys would benefit patients who are waiting for a transplant (2–7). In the current US transplant system, deceased donor kidneys are allocated to patients on the basis of their waiting time, calculated panel reactive antibody (CPRA), prior living donation, zero or single HLA-DR mismatch, and geographic proximity to the donor hospital (8). Additionally, deceased donor kidneys are risk stratified using the Kidney Donor Profile Index (KDPI), with kidneys in the lowest-quality range (KDPI >85) allocated only to patients who have given prior informed consent (9). A model assessing the 3-year risk-benefit trade-off for declined offers found that acceptance of these offers would likely have resulted in better outcomes for patients lower on the match run (those with longer expected waiting time) (3). Compared with staying on dialysis, even a KDPI=99 kidney appears to offer greater longevity for most transplant candidates (4). Studies also suggest that a significant reduction in kidney discard is possible. A study applying French kidney acceptance practices to US deceased donor kidneys found that 62% of the kidneys discarded in the United States between 2004 and 2014 would have been transplanted under the French system (2).

Although studies suggest that patients who are waitlisted could benefit from policies to reduce kidney discard, less is understood about patient preferences with regard to accepting offers of lower-quality kidneys. A survey of patient allocation policy preferences found that 74% “somewhat” or “completely” accepted a policy for fast-tracking kidneys at risk of discard and that 71% personally would accept a marginal kidney expected to have 2–3 weeks of delayed graft function to avoid 2 additional years on the waitlist (10). However, currently, patients are not the main drivers of kidney acceptance decisions; these decisions are primarily made by clinicians (11). One study found that patients who died on the waiting list between 2008 and 2015 received a median of 16 offers; however, transplant centers declined these kidneys on behalf of their patients due to quality concerns (12). Measuring patient preferences and how they vary in the population can help inform future allocation policies that are more patient centered. Systems may be developed in the future that can dynamically incorporate patient preference information.

This study quantifies patients’ assessment of the trade-off between kidney quality (expected graft survival and level of kidney function) and waiting time using a discrete choice experiment. Discrete choice experiments are a rigorous method for quantifying stated preferences for the outcomes of health interventions or policies (13–15). They incorporate principles from cognitive psychology and economics to elicit respondents’ willingness to accept trade-offs among competing, desirable health-outcome or health-policy features (16). Discrete choice experiments have been used to assess patients’ treatment preferences in nephrology and kidney transplantation (17,18), and patients’ preferences for kidney allocation policies and willingness to accept public health services increased risk kidneys and extended criteria kidneys (19–21). Most patients accepted a marginal kidney when presented with a scenario where they were facing medical urgency or in which they were highly sensitized (21). This study builds on previous work by presenting patients with choices involving trade-offs between a lower-quality kidney offered today and a higher-quality future offer.

Materials and Methods

Discrete Choice Experiment

The research team designed and tested a discrete choice experiment (DCE) following good-practice guidance (22,23). A series of concept-elicitation interviews were conducted with patients who were waitlisted and previously transplanted to understand key deceased donor kidney attributes important to patients (24). The results, along with consultation with leadership and members of the American Association of Kidney Patients (AAKP) and clinical experts, led to the definition of three attributes: (1) time with regular kidney function after the transplant (GFR ≤60 ml/min per 1.73 m2), (2) time with low kidney function after the transplant (GFR 15 to 60 ml/min per 1.73 m2), and (3) waiting time for the kidney. Years of graft survival corresponding to each level of kidney function were set on the basis of the median survival for kidneys with various levels of KDPI. We used post-transplant graft survival rather than providing information on KDPI because findings from the concept-elicitation interviews indicated that many patients do not have a clear understanding of KDPI (24). Waiting time was presented as the additional number of years the patient would have to wait for a future kidney with given attributes (Table 1).

Table 1. - Study attributes and levels
Attribute Levels, yr
Time with regular function after transplant a 8
Time with low function after transplant a 5
Waiting time for kidney 0 a
aStudy design required total time with regular and low function after transplant for kidney today to be either equal to or less than that for future kidney. Also, kidney today required no waiting time, whereas future kidney did.

The choice tasks were introduced to participants as options presented by their physician, given all information available. In each task, participants were asked to choose between a kidney available today and a future kidney, both of which were experimentally constructed deceased donor kidneys described by the attributes and levels in Table 1. The “future kidney” involved additional years of waiting time and always offered longer graft survival (Figure 1).

Figure 1.:
Example choice question.

The introduction to the discrete choice experiment explained the study attributes, the choice task, and decision context in patient-friendly language. Three attribute-comprehension questions were included to test participants’ understanding of the provided information. The survey also included questions about participants’ background and health history. The survey was pilot-tested through cognitive interviews with 14 patients who met the study inclusion criteria to ensure that the information provided was clear and participants were able to answer the choice tasks as expected (25). A copy of the final survey instrument is provided in Supplemental Appendix A.

Experimental Design

To efficiently obtain preference signals from the choices elicited, a D-efficient fractional-factorial design was generated using an algorithm developed in SAS version 9.4 (Cary, NC) (26). The study experimental design included 144 questions, divided into 24 sets of six questions each. Each participant was randomly assigned to only complete one set. The order of the questions within and across sets was randomized to minimize order effects.

Participant Recruitment

Study participants were US residents, English proficient, aged ≥21 years, and were waiting for, or had received, a kidney transplant. Participants were members of the AAKP or were waitlisted or had received a kidney transplant at Northwestern Comprehensive Transplant Center. Potentially eligible individuals were invited to complete an online DCE survey via email. The AAKP distributed the survey link to their listserv. Emails of patients at the Northwestern Comprehensive Transplant Center were obtained through the Northwestern Electronic Data Warehouse. A total of four weekly reminders were sent. All participants provided informed consent. Recruitment and survey implementation protocols were reviewed and approved by the Institutional Review Board at Northwestern University (STU00208614).


The validity of the DCE data was evaluated using commonly followed quality checks, including response nonvariation and attribute-comprehension questions (27). We followed good-practice guidance on logit-based regression analysis to link patient responses to the trade-offs between deceased donor kidneys in the choice questions (16). Results from these models are preference weights (or log-odds) that reflect the average change in preferences (i.e., probability of choice) for changes in kidney attribute levels, all else being equal (28).

Following Vass et al., we used a heteroscedastic conditional-logit model to test for differences between respondents who were recruited through the AAKP and Northwestern Comprehensive Transplant Center (29). We then used a random-parameters logit model to estimate mean preferences and the dispersion (i.e., standard deviation) of preferences across respondents (30). All attribute levels, except waiting time (a continuous linear variable), were included in the model specification as categorical variables and were assumed to have normally distributed values across respondents. An interaction term between time with low kidney function and time with regular kidney function was also considered in the model specification. This interaction allowed preferences for time with each level of kidney function to adjust on the basis of the overall graft survival offered by a kidney.

A latent-class logit model was also used to group respondents on the basis of the patterns of their choices and to obtain preference estimates for each group (or class). Each respondent was assigned to the classes in a probabilistic way, depending on their similarity to the average respondent in each class. The probability of class assignment was then correlated with patient characteristics, including age, race, educational attainment, previous transplantation history, Karnofsky performance status, and whether the respondent was on a waiting list. To make preference estimates among the classes comparable, the difference between no waiting time and 4 years of waiting was set to have a preference weight of 100, and all of the other preference estimates were rescaled accordingly. Missing data in the covariates in the latent class model were replaced by the sample means.

Preference weights obtained from the random-parameters logit and latent-class logit models were used to calculate the reduction in the level of graft survival that respondents would accept for the current kidney. These calculations are called time equivalences because they indicate the reduction in expected graft survival that would leave respondents indifferent between accepting a kidney today or waiting for a future kidney. The 95% confidence intervals (95% CIs) for time equivalences were estimated using the Krinsky–Robb procedure (31).


Participant Characteristics

In total, 605 participants completed the survey; 409 from the AAKP, and 196 from Northwestern Comprehensive Transplant Center. Table 2 provides summary statistics. Comparisons by data source are in Supplemental Appendix B. There were no significant differences between respondents from the two data sources, and the choice data showed no differences in variance (P=0.75). Thus, we focus on the pooled data.

Table 2. - Descriptive statistics of study samplea
Demographic Characteristics All Respondents (n=605)
Mean age (SD), yr 61 (12)
Female, n (%) 335 (56)
Race, n (%)
 White 458 (76)
 Black 84 (14)
 Hispanic, Latino or Spanish 29 (5)
 Others 30 (5)
 Missing 4 (1)
Educational attainment, n (%)
 High school or equivalent (such as GED) 39 (6)
 Some college but no degree 133 (22)
 Technical school, associate’s degree or 2-year college degree 83 (14)
 4-year college degree (such as BA, BS) 206 (34)
 Graduate or professional degree (such as MBA, MS, MA, MD, PhD) 138 (23)
 Missing 6 (10)
Annual household income, n (%)
 Less than $50,000 203 (34)
 $50,000 to $99,999 168 (28)
 $100,000 or more 131 (22)
 Don't know/not sure 5 (1)
 Prefer not to answer 92 (15)
 Missing 6 (1)
Clinical characteristics, n (%)
 Had diabetes 212 (35)
 Currently on dialysis 176 (29)
Self-rating of current health, n (%)
 Excellent 65 (11)
 Good 335 (55)
 Fair 188 (31)
 Poor 17 (3)
Self-reported performance status, n (%)
 My life is normal. 150 (25)
 I have health problems, but I am able to carry on with normal, daily activities. 300 (50)
 I am able to care for myself, but unable to carry on with normal, daily activities or do active work. 95 (16)
 I need occasional assistance, but I am able to care for most of my personal needs. 49 (8)
 I need considerable assistance and frequent medical care. 8 (1)
 I am completely disabled. 3 (1)
Number of kidney transplants, n (%)
 None 157 (26)
 One 364 (60)
 Two or more 84 (14)
Currently waiting for a kidney transplant, n (%) 211 (35)
 Average time spent waiting (SD) in years 4 (3)
On waiting list for first kidney transplant and not yet on dialysis 37 (6)
Performance on survey questions, n (%)
 Attribute dominance
   Always chose “kidney today” with no waiting time 130 (21)
   Always chose “future kidney” 32 (5)
   More time with regular function 63 (10)
   Less time with low function 40 (7)
Number of attribute-comprehension questions correctly answered, n (%)
  0 or 1 56 (9)
  2 97 (16)
  3 452 (75)
GED, graduate equivalency degree; BA, bachelor of arts; BS, bachelor of science; MBA, master of business administration; MS, master of science; MA, master of arts; MD, doctor of medicine; PhD, doctor of philosophy.
aDue to rounding, some numbers may or may not add up to 100%.

The mean age for the full sample was 61 years (SD=12). Other studies have found that patients’ mean age at listing is about 53 years and average waiting time to transplantation is about 4.5 years (4), thus the sample mean age is higher than the population mean age at listing, but close to the expected age of patients near the top of the waitlist. Approximately 56% of respondents were female, 76% were White, 14% were Black, and 5% identified as Latino or Hispanic. In total, 57% reported having ≥4-year college degree, and 34% had an annual household income less than $50,000. Overall, 35% had diabetes and 29% reported being on dialysis. Although 26% of respondents were waiting for their first kidney transplant, the remaining respondents had one or more transplants. Of respondents, 35% were waiting for a kidney transplant. Average waiting time was 4 years.

When evaluating respondents’ understanding, 16% correctly answered two and 75% correctly answered all three attribute-comprehension questions in the survey. A total of 5% always chose the future kidney and 21% always chose the kidney today. Additionally, 10% of respondents always chose the kidney that would provide more time with regular function, whereas 7% always chose the kidney that would have less time with low function. These results are consistent with the distribution of attribute-comprehension and response nonvariation checks in previous DCE work (27).

Participant Preferences for Deceased Donor Kidneys

Figure 2 shows the estimated preference weights for the full sample. Higher weights indicate greater preferences for outcomes. The relative size of the weights and the distance between them indicates relative importance (32). The full set of mean preference and SD estimates from the random-parameters logit are included in Supplemental Appendix B.

Figure 2.:
Preference weights for the full sample ( n =605). Preference weights are from the random-parameters logit model. Error bars indicate 95% confidence intervals.

Respondents preferred kidneys with longer graft survival, especially with longer periods of regular function, and shorter waiting time. Conditional on the attributes and corresponding levels included in the DCE, a kidney with 8 years of regular function (GFR ≥60 ml/min per 1.73 m2) and 2 years of low function (GFR 30 to 60 ml/min per 1.73 m2) was the most preferred option, whereas a kidney with 2 years of regular function (GFR ≥60 ml/min per 1.73 m2) and 2 years of low function (GFR of 30 to 60 ml/min per 1.73 m2) was the least preferred.

Three classes of respondent preferences were identified by the latent-class logit model. Rescaled preference estimates for these classes are presented in Figures 3, 4, and 5. Respondents had a 61%, 25%, and 15% chance of being in class 1, class 2, and class 3, respectively.

Figure 3.:
Rescaled preference weights for class 1, class 2, and class 3. (A) Class 1; (B) class 2; (C) class 3. Preference weights are from the latent-class logit model and rescaled. Error bars indicate 95% confidence intervals.

Class 1 was considered our baseline class due to the higher membership probability. Increases in waiting time had a strong negative effect on preference for a kidney; however, respondents in this class also consistently valued improvements in graft survival.

In class 2, we observed an even stronger aversion to waiting for a kidney than in class 1. Respondents in this class were less willing to accept increased waiting time in exchange for longer graft survival.

Respondents in class 3 were more concerned with graft survival than waiting time. Decreases in preference associated with longer waiting time were small relative to increases in preference associated with longer graft survival.

Relative to being in the baseline class (class 1), Black respondents (P<0.01), respondents above the median sample age of 62 (P<0.001), and those who reported lower educational attainment (P<0.001) and lower performance status (Karnofsky ≤60) (P<0.001) were more likely to belong to class 2. Respondents below the age of 62 and those who were waitlisted before they started dialysis (P<0.001) were more likely to be in class 3. This result was consistent across patients before and after transplant. Specification tests showed that after controlling for all other patient factors in the model, being on dialysis, having a prior transplant, and whether the patient was on a waiting list were not statistically significant determinants of class membership.

The full set of preference estimates by class from the latent-class logit model are included in Supplemental Appendix B.

Time Equivalences

Table 3 presents scenarios where respondents could opt to accept a kidney today or wait 2, 3, or 4 years for kidneys with different expected graft survival levels. The future kidneys all offer 8 years of regular function and either 2, 3, or 5 years of low function. Findings separated out by data source are reported in Supplemental Appendix B. On average, respondents were willing to accept kidneys with shorter graft survival and shorter time with regular function to reduce their waiting time. Acceptability of the kidney today decreased as expected graft survival increased for future kidneys. The average respondent was willing to forgo 4.5 (95% CI, 4.0 to 5.1) years of regular function for a 2-year reduction in waiting time, accepting a kidney that provided only 6.5 (95% CI, 5.9 to 7.0) years of total graft survival (including 3 years with low function) over a kidney with 11 years of expected graft survival (8 years of regular function and 3 years of low function). This means accepting a ≥90% KDPI kidney today rather than waiting 2 additional years to receive a 30% KDPI kidney. To avoid a 4-year wait, respondents were willing to accept a reduction of 7.3 (95% CI, 6.6 to 8.5) years of regular kidney function. The minimum acceptable graft survival in this scenario was 3.7 (95% CI, 2.5 to 4.4) years. This means the average respondent would accept a kidney with 95%–99% KDPI to avoid 4 years of additional waiting time. Table 3 also reports class-specific equivalences.

Table 3. - Estimated time equivalences by model (in years
Expected Graft Survival for Future Kidney Reduction in Total Waiting Time Acceptable Reduction in Time with Regular Kidney Function
All Respondents (n=605) a Class 1 b Class 2 b Class 3 b
From To
10 years including
8 years of regular function and 2 years of low function
2 0 3.7 (3.3 to 4.2) 3.2 (2.5 to 3.6) 7.0 (6.5 to 8.7) 0.08 (0.004 to 0.24)
3 0 5.0 (4.5 to 5.5) 4.1 (3.6 to 4.6) 8.4 (7.6 to 11.3) 0.12 (0.006 to 0.36)
4 0 6.0 (5.5 to 6.7) 5.0 (4.4 to 5.5) 9.8 (8.6 to 13.9) 0.16 (0.008 to 0.48)
11 years including
8 years of regular function and 3 years of low function
2 0 4.5 (4.0 to 5.1) 3.7 (3.2 to 4.3) 5.0 (4.5 to 6.0) 0.10 (0.005 to 0.27)
3 0 5.9 (5.2 to 6.6) 4.7 (4.1 to 5.3) 8.4 (7.3 to 11.8) 0.14 (0.008 to 0.40)
4 0 7.3 (6.6 to 8.5) 5.7 (5.0 to 6.3) 11.9 (9.5 to 19.1) 0.19 (0.011 to 0.54)
13 years including
8 years of regular function and 5 years of low function
2 0 5.2 (4.6 to 6.1) 3.6 (3.2 to 4.0) 10.3 (8.6 to 19.6) 0.11 (0.006 to 0.28)
3 0 7.1 (6.2 to 9.9) 4.3 (3.9 to 4.7) 12.5 (10.0 to 26.0) 0.16 (0.009 to 0.43)
4 0 8.9 (7.5 to 14.4) 5.0 (4.5 to 6.4) 14.6 (11.4 to 32.4) 0.21 (0.012 to 0.57)
Data are presented as estimate and 95% confidence intervals.
aTime equivalences (in years) were estimated using preference weights obtained from the random-parameters logit model.
bTime equivalences were estimated using preference weights obtained from the latent-class logit model. Respondents had a 61%, 25%, and 15% chance of being in class 1, class 2, and class 3, respectively.


The results suggest, on average, respondents were strongly averse to additional years of waiting time for a deceased donor kidney. Respondents adjusted their assessment of acceptable kidneys today as waiting time for the future kidney increased. The diminishing relative importance of graft survival as waiting time for the future kidney increases is likely, because as waiting time increases, the risk of death on dialysis increases. Thus, with increased waiting time to future offer, the offer today becomes more acceptable, whereas the quality of the future offer is less influential. These results suggest marginal-quality kidneys would be more acceptable to patients with longer expected waiting time to receive another kidney offer.

The patient classes identified had distinct views on the acceptability of trade-offs between graft survival and waiting time. Our results suggest patients aged <62 years and those who were waitlisted before starting dialysis would be more willing to wait for future kidneys with longer graft survival. In contrast, respondents aged ≥62 years, respondents with lower performance status, Black respondents, and those with lower education levels were systematically more likely to be concerned primarily about reducing their waiting time.

The prevalence of these three respondent classes in our sample should not be taken as a representative distribution of patient types in the population. However, the identification of these groups provides insights into patients’ views. Offering marginal kidneys to patients who prefer to wait for a higher-quality kidney adds cold ischemia time and can lead to kidney refusals. Meanwhile, not offering the option of accepting marginal-quality kidneys to patients who prioritize limiting their waiting time over kidney quality could reduce their well-being.

The policy and operational implications of our results are worth highlighting. First, systematic variations in patient preferences suggest that transparent communication of patients’ treatment preferences is important in establishing a more patient-centered organ allocation system and aligning patient and provider perspectives. Second, having individual patient preference information at hand when evaluating organ offers could help clinicians identify suitable candidates willing to accept lower-quality kidneys, and avoid delay in placement and discard of these organs. Under the current system, the accrual of cold ischemia time on already marginal kidneys can lead to unnecessary discard. Third, the process of obtaining information from patients could help patients clarify their own preferences and reduce decisional conflict (33). Patient preference elicitation in clinical practice can complement current consenting practices for KDPI >85 kidneys. This information could be integrated into the Organ Procurement and Transplantation Network (OPTN) system to support more efficient, patient-centered organ allocation. An example of such an allocation system could be fast-tracking marginal kidneys to patients who are more willing to accept such kidneys. Recent evidence suggests that fast-tracking could be an acceptable policy for patients and physicians (10).

Our work has some limitations. First, the decisions were hypothetical. They do not carry the same consequences and thus could differ from those made in the real world. To minimize hypothetical bias, we built scenarios that increased the consequentiality of choice through the relevance and plausibility of the questions, and by explicitly inviting respondents to be part of the research team and feel responsible for the success of the study (34,35). The relevance and plausibility of the decision context and trade-offs were tested during pretest interviews. The scenarios assumed that future offers were highly certain to avoid heuristics that could lead to biases associated with choices under uncertainty and affect our measures of preferences. Future research should consider the role of uncertainty in the timing and quality of future kidney offers to better understand how that affects patients’ decisions. Work by Bae et al. predicting who can benefit from a marginal kidney and by Kilambi et al. evaluating acceptance of marginal-quality kidneys using decision trees suggest that our scenarios may have presented a conservative assessment of the benefit of accepting a low-quality kidney today, given that there is no guarantee that future offers will be of significantly higher quality (7,36). Patients might be more risk averse in their decision making and thus more likely to accept the kidney today when the quality of the future kidney is uncertain.

Another potential limitation was the sampling framework. Although respondents recruited through the major transplant center had confirmation of waitlist/recipient status, we relied on self-report from AAKP respondents. Additionally, respondents from the AAKP may be better informed than other patients with kidney failure. However, we found no differences in average preferences between these respondents and those recruited through the transplant center. The recruitment period was extended to obtain a sufficiently large sample size to identify differential effects by race, ethnicity and education level; nonetheless, our sample has over-representation of respondents with a college degree and White, non-Hispanic Americans. Although our sample is not fully representative of the kidney transplant candidate and recipient population, the study objective was not to obtain average population effects, but to better understand different patient perspectives on the trade-offs associated with kidney acceptability. To meet this objective, we sought to obtain a sufficiently large and diverse sample that allowed a robust analysis of preference variation. In addition, we used latent class analysis to evaluate complex patterns in the variations in preferences and to draw more robust conclusions around the potential drivers of preference heterogeneity. We also note that class membership in the latent class analysis is probabilistic. Significant correlations between respondent characteristics and preferences do not mean all patients with given characteristics have the same preferences. It is also possible that other unmeasured patient characteristics were associated with some of the class membership effects we attribute to the variables in our model. Although information from our analysis can be used in the context of policy making, individual-level preference information will be needed to inform decisions in a clinical setting.

To our knowledge, this is the first preference study that has attempted to quantify the trade-offs in expected graft survival that patients are willing accept to reduce waiting time, with a specific focus on kidney offers in the bottom 15% of expected graft survival (KDPI >85). The findings suggest many patients, particularly older patients and patients with lower performance status, would consider high KDPI kidneys to receive a kidney transplant sooner. Patient preferences are heterogeneous, and the reasons behind these variations should be explored in more detail in future research. It is also important to address clinicians’ views on the acceptability of trade-offs associated with marginal kidneys in future research. Future research should also consider how uncertainty in the quality and timing of future kidney offers affects patients’ decisions.


J.J. Friedewald reports having consultancy agreements with Eurofins Transplant Genomics, Inc. and Sanofi; receiving research funding from CSL Behring, Hansa BioPharma, National Institutes of Health, Eurofins Viracor, Inc., and Veloxis; receiving honoraria from Sanofi; having patents or royalties from Northwestern University/Scripps Research Institute; serving in an advisory or leadership role for Eurofins Transplant Genomics, Inc.; and speakers bureau for Sanofi. J.M. Gonzalez Sepulveda reports having consultancy agreements with Biomarin, IQVIA, Merck, and Vertex; receiving research funding from Amgen, Astellas, Grifols, GSK, and Janssen; receiving honoraria from RAND Corporation; and serving in an advisory or leadership role for International Academy of Health Preference Research and ISPOR. R. Knight reports employment with the American Association of Kidney Patients (AAKP); receiving honoraria from American Kidney Fund, Johns Hopkins Center for Health Equity, Labcorp, Northwestern University, Novartis, Otsuka, and Personalized Medicine Coalition; serving in an advisory or leadership role for National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Advisory Council, Scientific Registry of Transplant Recipients (SRTR) Review Committee, and Scientific Advisory Board for the “Rescuing Kidneys at Risk of Discard” project; serving as AAKP President and Quality Insights Patient Advisory Committee Member; and having other interests or relationships with Bowie State University Board of Advisors and SRTR Review Committee. S. Mehrotra reports having consultancy agreements with The Joint Commission, ownership interest in Medecipher, Inc., serving in an advisory or leadership role for Medecipher, Inc., and being the principal investigator for this study. J.-C. Yang reports having employment with and ownership interest in Pacific Economic Research, LLC. The remaining author has nothing to disclose.


This work is supported by the National Institutes of Health 1R01DK118425-01A1.

Published online ahead of print. Publication date available at

See related Patient Voice, “Getting a Kidney: Where Is Patient Choice?” on pages , and editorial, “Improving the Utilization of Deceased Donor Kidneys by Prioritizing Patient Preferences,” on pages .


The authors would like to thank the members of our scientific advisory board and the American Association of Kidney Patients for their support with this study.

Author Contributions

J.M. Gonzalez and S. Mehrotra conceptualized the study; J.M. Gonzalez and J.-C. Yang were responsible for the data curation; J.M. Gonzalez and J.-C. Yang were responsible for the formal analysis; S. Mehrotra was responsible for the funding acquisition; J.M. Gonzalez, S. Mehrotra, K. Schantz, and J.-C. Yang were responsible for the investigation; J.M. Gonzalez and J.-C. Yang were responsible for the methodology; K. Schantz was responsible for the project administration; J.J. Friedewald, R. Knight, and S. Mehrotra were responsible for the resources; J.M. Gonzalez was responsible for the survey programming and hosting; J.M. Gonzalez and S. Mehrotra provided supervision; J.J. Friedewald, J.M. Gonzalez, R. Knight, S. Mehrotra, and K. Schantz were responsible for the validation; J.M. Gonzalez and S. Mehrotra were responsible for the visualization; J.M. Gonzalez and K. Schantz wrote the original draft; and J.J. Friedewald, J.M. Gonzalez, R. Knight, S. Mehrotra, K. Schantz, and J.-C. Yang reviewed and edited the manuscript.

Supplemental Material

This article contains the following supplemental material online at

Supplemental Appendix A. Final survey instrument.

Supplemental Appendix B. Demographic characteristics and time equivalences by data source and random-parameters logit and latent-class logit results.


1. Hart A, Smith JM, Skeans MA, Gustafson SK, Wilk AR, Castro S, Foutz J, Wainright JL, Snyder JJ, Kasiske BL, Israni AK: OPTN/SRTR 2018 annual data report: Kidney. Am J Transplant 20[Suppl s1]: 20–130, 2020.
2. Aubert O, Reese PP, Audry B, Bouatou Y, Raynaud M, Viglietti D, Legendre C, Glotz D, Empana JP, Jouven X, Lefaucheur C, Jacquelinet C, Loupy A: Disparities in acceptance of deceased donor kidneys between the United States and France and estimated effects of increased US acceptance. JAMA Intern Med 179: 1365–1374, 2019
3. Wey A, Salkowski N, Kremers WK, Schaffhausen CR, Kasiske BL, Israni AK, Snyder JJ: A kidney offer acceptance decision tool to inform the decision to accept an offer or wait for a better kidney. Am J Transplant 18: 897–906, 2018
4. Bui K, Kilambi V, Mehrotra S: Functional status-based risk-benefit analyses of high-KDPI kidney transplant versus dialysis. Transpl Int 32: 1297–1312, 2019
5. Mohan S, Chiles MC, Patzer RE, Pastan SO, Husain SA, Carpenter DJ, Dube GK, Crew RJ, Ratner LE, Cohen DJ: Factors leading to the discard of deceased donor kidneys in the United States. Kidney Int 94: 187–198, 2018
6. Massie AB, Luo X, Chow EK, Alejo JL, Desai NM, Segev DL: Survival benefit of primary deceased donor transplantation with high-KDPI kidneys. Am J Transplant 14: 2310–2316, 2014
7. Bae S, Massie AB, Thomas AG, Bahn G, Luo X, Jackson KR, Ottmann SE, Brennan DC, Desai NM, Coresh J, Segev DL, Garonzik Wang JM: Who can tolerate a marginal kidney? Predicting survival after deceased donor kidney transplant by donor-recipient combination. Am J Transplant 19: 425–433, 2019
8. Network OPTN: Removal of DSA and region from kidney allocation policy. Available at: Accessed July 14, 2022
9. Chen SS, Pankratz VS, Litvinovich I, Al-Odat RT, Unruh ML, Ng Y-H: Expanded criteria donor kidneys with kidney donor profile index ≤85%: Are we doing enough to counsel our patients? Transplant Proc 52: 54–60, 2020
10. Mehrotra S, Schantz K, Friedewald JJ, Ladner DP, Becker Y, Formica R, Barah M, Gu J, Gordon EJ: Physician and patient acceptance of policies to reduce kidney discard. Clin Transplant 34: e14054, 2020
11. Huml AM, Albert JM, Thornton JD, Sehgal AR: Outcomes of deceased donor kidney offers to patients at the top of the waiting list. Clin J Am Soc Nephrol 12: 1311–1320, 2017
12. Husain SA, King KL, Pastan S, Patzer RE, Cohen DJ, Radhakrishnan J, Mohan S: Association between declined offers of deceased donor kidney allograft and outcomes in kidney transplant candidates. JAMA Netw Open 2: e1910312, 2019
13. Ho MP, Gonzalez JM, Lerner HP, Neuland CY, Whang JM, McMurry-Heath M, Hauber AB, Irony T: Incorporating patient-preference evidence into regulatory decision making. Surg Endosc 29: 2984–2993, 2015
14. Johnson FR, Zhou M: Patient preferences in regulatory benefit-risk assessments: A US perspective. Value Health 19: 741–745, 2016
15. Johnson FR, Desvousges WH: Estimating stated preferences with rated-pair data: environmental, health, and employment effects of energy programs. J Environ Econ Manage 34: 79–99, 1997
16. McFadden D: Conditional logit analysis of qualitative choice behavior. In: Frontiers in Econometrics, edited by Zarembka P, New York, Academic Press, 1973
17. Clark MD, Determann D, Petrou S, Moro D, de Bekker-Grob EW: Discrete choice experiments in health economics: A review of the literature. PharmacoEconomics 32: 883–902, 2014
18. Clark MD, Szczepura A, Gumber A, Howard K, Moro D, Morton RL: Measuring trade-offs in nephrology: A systematic review of discrete choice experiments and conjoint analysis studies. Nephrol Dial Transplant 33: 348–355, 2018
19. Howard K, Jan S, Rose JM, Wong G, Irving M, Tong A, Craig JC, Chadban S, Allen RD, Cass A: Community preferences for the allocation of donor organs for transplantation: A discrete choice study. Transplantation 99: 560–567, 2015
20. Clark MD, Leech D, Gumber A, Moro D, Szczepura A, West N, Higgins R: Who should be prioritized for renal transplantation? Analysis of key stakeholder preferences using discrete choice experiments. BMC Nephrol 13: 152, 2012
21. Kamran S, Conti F, Pomey M-P, Baron G, Calmus Y, Vidal-Trecan G: Patients’ preferences in transplantation from marginal donors: Results of a discrete choice experiment. Transpl Int 30: 589–602, 2017
22. Bridges JF, Hauber AB, Marshall D, Lloyd A, Prosser LA, Regier DA, Johnson FR, Mauskopf J: Conjoint analysis applications in health--A checklist: A report of the ISPOR Good Research Practices for Conjoint Analysis Task Force. Value Health 14: 403–413, 2011
23. Reed Johnson F, Lancsar E, Marshall D, Kilambi V, Mühlbacher A, Regier DA, Bresnahan BW, Kanninen B, Bridges JF: Constructing experimental designs for discrete-choice experiments: Report of the ISPOR Conjoint Analysis Experimental Design Good Research Practices Task Force. Value Health 16: 3–13, 2013
24. Schantz K, Gordon EJ, Lee U, Rocha M, Friedewald J, Ladner DP, Becker Y, Formica R, Reese PP, Kaufman D, Barah M, Walker M, Mehrotra O, Viveros D, Mehrotra S: Patient and clinician perceptions of informed consent and decision making about accepting KDPI > 85 kidneys. Transplant Direct 8: e1254, 2021
25. Hollin IL, Craig BM, Coast J, Beusterien K, Vass C, DiSantostefano R, Peay H: Reporting formative qualitative research to support the development of quantitative preference study protocols and corresponding survey instruments: Guidelines for authors and reviewers. Patient 13: 121–136, 2020
26. Kuhfeld WF, Tobias RD, Garratt M: Efficient experimental design with marketing research applications. J Mark Res 31: 545–557, 1994
27. Johnson FR, Yang JC, Reed SD: The internal validity of discrete choice experiment data: A testing tool for quantitative assessments. Value Health 22: 157–160, 2019
28. Hauber AB, González JM, Groothuis-Oudshoorn CGM, Prior T, Marshall DA, Cunningham C, Ijzerman MJ, Bridges JFP: Statistical methods for the analysis of discrete choice experiments: A report of the ISPOR Conjoint Analysis Good Research Practices Task Force. J Int Soc Pharmacoecon Outcomes Res 19: 300–315, 2016
29. Vass CM, Wright S, Burton M, Payne K: Scale heterogeneity in healthcare discrete choice experiments: A primer. Patient 11: 167–173, 2018
30. Train KE: Discrete choice methods with simulation, NY, USA, Cambridge University Press, 2009
31. Krinsky I, Robb AL: On approximating the statistical properties of elasticities. Rev Econ Stat 68: 715–719, 1986
32. Bui K, Kilambi V, Mehrotra S: Functional status-based risk-benefit analyses of high-KDPI kidney transplant versus dialysis. Transpl Int 32: 1297–1312, 2019
33. Witteman HO, Ndjaboue R, Vaisson G, Dansokho SC, Arnold B, Bridges JFP, Comeau S, Fagerlin A, Gavaruzzi T, Marcoux M, Pieterse A, Pignone M, Provencher T, Racine C, Regier D, Rochefort-Brihay C, Thokala P, Weernink M, White DB, Wills CE, Jansen J: Clarifying values: An updated and expanded systematic review and meta-analysis. Med Decis Making 41: 801–820, 2021
34. Carson RT, Groves T, List JA: Consequentiality: A theoretical and experimental exploration of a single binary choice. J Assoc Environ Resour Econ 1: 171–207, 2014
35. Özdemir S, Johnson FR, Hauber AB: Hypothetical bias, cheap talk, and stated willingness to pay for health care. J Health Econ 28: 894–901, 2009
36. Kilambi V, Bui K, Hazen GB, Friedewald JJ, Ladner DP, Kaplan B, Mehrotra S: Evaluation of accepting kidneys of varying quality for transplantation or expedited placement with decision trees. Transplantation 103: 980–989, 2019

kidney transplantation; transplant outcomes; waiting lists; patient preference

Copyright © 2022 by the American Society of Nephrology