Author + information
- Received April 7, 2015
- Revision received July 6, 2015
- Accepted July 9, 2015
- Published online December 1, 2015.
- Raymond C. Givens, MD, PhD∗∗ (, )
- Todd Dardas, MD, MS†,
- Kevin J. Clerkin, MD∗,
- Susan Restaino, MD∗,
- P. Christian Schulze, MD, PhD∗ and
- Donna M. Mancini, MD∗
- ∗Division of Cardiology, Department of Medicine, Columbia University Medical Center, New York, New York
- †Division of Cardiology, University of Washington, Seattle, Washington
- ↵∗Reprint requests and correspondence:
Dr. Raymond C. Givens, Department of Medicine, Division of Cardiology, Columbia University Medical Center, 622 West 168th Street, New York, New York 10032.
Objectives This study sought to assess the association of multiple listing with waitlist outcomes and post–heart transplant (HT) survival.
Background HT candidates in the United States may register at multiple centers. Not all candidates have the resources and mobility needed for multiple listing; thus this policy may advantage wealthier and less sick patients.
Methods We identified 33,928 adult candidates for a first single-organ HT between January 1, 2000 and December 31, 2013 in the Organ Procurement and Transplantation Network database.
Results We identified 679 multiple-listed (ML) candidates (2.0%) who were younger (median age, 53 years [interquartile range (IQR): 43 to 60 years] vs. 55 years [IQR: 45 to 61 years]; p < 0.0001), more often white (76.4% vs. 70.7%; p = 0.0010) and privately insured (65.5% vs. 56.3%; p < 0.0001), and lived in zip codes with higher median incomes (US$90,153 [IQR: US$25,471 to US$253,831] vs. US$68,986 [IQR: US$19,471 to US$219,702]; p = 0.0015). Likelihood of ML increased with the primary center’s median waiting time. ML candidates had lower initial priority (39.0% 1A or 1B vs. 55.1%; p < 0.0001) and predicted 90-day waitlist mortality (2.9% [IQR: 2.3% to 4.7%] vs. 3.6% [IQR: 2.3% to 6.0]%; p < 0.0001), but were frequently upgraded at secondary centers (58.2% 1A/1B; p < 0.0001 vs. ML primary listing). ML candidates had a higher HT rate (74.4% vs. 70.2%; p = 0.0196) and lower waitlist mortality (8.1% vs. 12.2%; p = 0.0011). Compared with a propensity-matched cohort, the relative ML HT rate was 3.02 (95% confidence interval: 2.59 to 3.52; p < 0.0001). There were no post-HT survival differences.
Conclusions Multiple listing is a rational response to organ shortage but may advantage patients with the means to participate rather than the most medically needy. The multiple-listing policy should be overturned.
Severe heart failure is an increasingly prevalent public health problem (1). Although heart transplantation (HT) remains the definitive therapy for severe heart failure, there is a worsening mismatch between donor supply and the demand of an increasing number of HT candidates. Over the past decade, the increase in size of the adult HT waitlist has outpaced the increase in annual transplants (2).
The United Network for Organ Sharing (UNOS) allows transplant candidates to pursue multiple listing (i.e., simultaneous transplant registration at multiple centers) (3). Previous studies indicate enhanced transplant rates and reduced waiting times for multiple-listed (ML) adult kidney and liver candidates (4). An analysis of pediatric HT candidates from 1995 to 2009 revealed a low ML prevalence (0.4%) but a trend toward higher HT rates (5). The nonmedical costs of multiple listing, which may include transportation and temporary housing, are not covered by insurance. Additionally, patients with state-run Medicaid do not have the option to be listed at out-of-state centers. Critics have therefore argued that multiple listing advantages wealthier patients, possibly at the expense of those with fewer resources (3,6). Studies of kidney and liver transplant candidates suggest ML patients live in zip codes with higher median incomes than do single-listed (SL) patients (7,8).
In this study, we investigated demographic and clinical characteristics and waitlist outcomes of adult ML and SL HT candidates listed from 2000 to 2013. We hypothesized that ML patients would have higher income and socioeconomic status, a higher prevalence of private insurance, and higher HT frequency. We conclude that although multiple listing is a rational response to prolonged HT waiting times, it advantages candidates who have more financial resources but are objectively less sick than their SL counterparts. We argue the need to reconsider the multiple-listing policy.
We identified patients aged ≥18 years in the Organ Procurement and Transplantation Network (OPTN) database listed for their first, single-organ HT in the United States between January 1, 2000, and December 31, 2013. The OPTN database includes demographic and clinical information at listing for all HT candidates and at transplantation for all recipients in the United States. These data are supplemented with Social Security Death Master File data, including for patients removed from the waitlist without undergoing HT, and are provided to investigators as deidentified data. Review by the Columbia University Institutional Review Board was thus unnecessary. The Health Resources and Services Administration of the U.S. Department of Health and Human Services provides oversight of the activities of the OPTN contractor, UNOS.
Study Design and Definitions
We defined ML candidates as those having ≥2 overlapping registrations at different centers. Patients with multiple registrations who did not meet the ML definition were defined as SL. We compared demographic and clinical characteristics between SL and ML patients. We analyzed predictors of ML status and HT by univariable and multivariable regression. The median adjusted gross incomes (AGIs) of patients’ home zip codes were obtained from 2013 Internal Revenue Service Statistics of Income (9). A socioeconomic status index on a scale of 0 to 100 was calculated from U.S. Census variables, as described previously (10). Linkage between candidates’ home zip codes and Census Bureau zip code tabulation areas was performed with Housing and Urban Development U.S. Postal Service crosswalk files (11).
Summary statistics are presented as median (interquartile range [IQR]) or as number (%). Baseline ML and SL characteristics were compared with the Wilcoxon test for continuous variables and the chi-square test for categorical variables. To address variables that confound the relationship between multiple listing and waitlist outcomes, we performed propensity score matching (PSM) with the MatchIt package for R to identify a cohort of SL patients to compare with ML patients through a genetic algorithm (12). Our multivariable regression model included 14 baseline variables: age, race, sex, ABO blood type, cardiac diagnosis, initial primary listing status, UNOS region, year of initial listing, initial urgency status of primary listing, education level, insurance status, socioeconomic status index, crossmatch requirement, and ventricular assist device. Competing outcome analyses were performed with the cmprsk package for R to compare waitlist outcomes of ML and SL patients (13). Post-HT survival was compared with Kaplan-Meier analysis and the log-rank test. Paired analyses of continuous variables of ML patients before and after multiple listing were performed with the Wilcoxon signed rank test.
Data were analyzed using JMP version 11.0.0 (SAS Institute Inc., Cary, North Carolina) and R version 3.1.1 (R Foundation for Statistical Computing, Vienna, Austria). All statistical tests were 2-sided and a p value <0.05 defined statistical significance.
We identified 33,928 patients, responsible for 36,040 listings, who were ≥18 years of age and who were listed for a first-time, single-organ HT during the study period. These patients formed the study cohort. We classified as ML 679 patients, who comprised 2.0% of all patients and accounted for 1,381 listings (3.8%).
Among ML patients, 656 (96.6%) had 2 listings and 23 (3.4%) had 3 listings. Among SL patients, 31,947 (96.1%) had 1 registration, 1,203 (3.6%) had 2 registrations, 90 (0.3%) had 3 registrations, and 9 (0.0%) had 4 registrations. Of the 1,302 SL patients with >1 registration, 652 (50.0%) received HT during their first listing and thus the subsequent listing was for retransplantation. For the remaining 650 SL patients with >1 registration, frequent reasons for removal from the first waitlist were clinical improvement obviating HT (n = 238; 36.6%), clinical deterioration (n = 110; 16.9%), removal in error (n = 52; 8.0%), transfer to another center (n = 51; 7.8%), and refusal of HT (n = 31; 4.8%).
Table 1 summarizes baseline characteristics of the study population. In comparison with SL patients, ML patients were younger and more likely to be white, male, to have some college education and private insurance, and had higher zip code–level socioeconomic status index. Median zip code–level annual AGI was higher for ML patients than for SL patients.
ML candidates more often had clinical characteristics associated with prolonged waitlist time among the SL candidates (i.e., blood type O, history of congenital heart disease, larger body size, and lower UNOS priority status at the start of their primary listing; all p < 0.0001). ML patients less often received therapy with intra-aortic balloon pumps, ventricular assist devices, or inotropic medications at initial registration and were less often hospitalized or treated in an intensive care unit, each of which also predicted longer wait among SL patients (all p < 0.0001). There was no difference in need for prospective crossmatch and no difference in prevalence of hypertension or diabetes mellitus. Median glomerular filtration rate, calculated with the Modification of Diet in Renal Disease study equation (14), was nominally lower among ML patients but there was no difference in need for hemodialysis. On the basis of a validated risk model derived by Singh et al. (15) that incorporates listing status, age, and mechanical circulatory support among other variables, ML patients were predicted to have lower 90-day waitlist mortality than the SL cohort at the start of the primary listing.
There was a nonsignificant difference between the groups in the year of initial listing. Figure 1 shows variation in number and proportion of multiple listings by year. There were regional differences among ML and SL patients (Table 1). With UNOS region 1 as reference, there was a significantly higher proportion of ML patients originating from region 9 (odds ratio: 1.78 [95% confidence interval: 1.16 to 2.73]; p = 0.0072), with intermediate p values for regions 7 and 8. Median waiting times for ML patients’ initial regions were higher than for SL patients (125 days [IQR: 115 to 135] vs. 123 [IQR: 99 to 128]; p = 0.0002), as were waiting times of primary listing centers (151 [IQR: 114 to 201] vs. 134 [IQR: 73 to 183]; p < 0.0001) and donor service areas supplying these centers (136 [IQR: 111 to 180] vs. 125 [IQR: 87 to 166]; p < 0.0001). The probability of multiple listing increased progressively with each 100 days of initial waiting, from 0.9% at <100 days to 4.5% at 801 to 900 days (p < 0.0001) and thereafter declined.
Priority Status Changes by Listing
Table 2 and Figure 2 summarize priority status changes of ML and SL patients. By the end of the primary listing the proportion of ML patients listed as 1A had declined from 14.4% to 7.2%, the 1B fraction had increased from 24.6% to 27.4%, and status 7 (inactive) patients had increased to 39.0% (n = 265) from 1.9%. As shown in Table 3, primary reasons given for inactivation were candidate choice (n = 42 [15.8%]), insurance issues (n = 25 [9.4%]), and being too sick for HT (n = 34 [12.8%]). The rate of HT at the secondary center, waitlist death, and post-HT outcomes were not different for patients temporarily inactivated for being too sick compared with all other ML patients. Among SL patients, there was an increase in status 1A candidates from 20.7% to 35.1% by the end of listing and a decrease in 1B patients from 34.4% to 31.0%; status 7 patients increased from 2.4% to 18.9%.
On secondary listing, 19.7% of ML patients were initially listed as status 1A, 41.0% as 1B, and 8.8% as status 7 (p < 0.0001 for comparison with primary listing). Of ML patients who were status 7 at the end of the primary listing, 62 (23.4%), 99 (37.4%), 67 (25.3%), and 37 (14%) were status 1A, 1B, 2, and 7, respectively, at the start of the secondary listing. On secondary listing, the median predicted 90-day mortality of ML candidates for whom all relevant data for both listings were available had increased to 3.6% (2.3% to 6.0%; p < 0.0001 vs. primary listing) and was similar to that for SL patients (p = 0.0394). This increase was largely driven by upgrades in ML patient priority at the secondary center. At time of HT, priority at the transplanting center did not differ significantly between the ML and SL groups.
ML patients had higher median waiting times during the primary listing that contributed to higher total median waiting times (Table 2) (both p < 0.0001). There was a median time of 253 days (IQR: 89 to 549 days) between the initial dates of primary and secondary listings. In paired analyses, median waiting times for all patients at ML candidates’ secondary centers were lower than those at their primary centers (105 days [IQR: 53 to 153 days] vs. 151 days [IQR: 114 to 201 days]; p < 0.0001); 70.3% of ML patients registered at a secondary center with a lower median waiting time than their primary center. ML patients were much less likely than SL patients to receive HT at their primary center (4.9% vs. 70.0%; p < 0.0001) but had a higher HT rate (74.4%; p = 0.0196) across all listings in unadjusted analyses, and lower crude waitlist mortality (8.1% vs. 12.2%; p = 0.0011). zip code–level AGI was positively associated with likelihood of HT in the SL cohort (p = 0.0006).
Figure 3A shows results of competing risk analyses of ML and SL waitlist outcomes. After an initial lag of more than 1,300 days, the HT rate for ML patients exceeded that of the SL cohort (p < 0.0001). The ML waitlist death rate was lower at all times throughout listing (p = 0.0035). As shown in Table 4, the PSM cohorts were well balanced: the p value for each comparison was >0.05. Figure 3B shows competing risk analyses of the matched cohorts: the ML cohort had significantly higher rates of HT (relative risk: 3.02 [95% confidence interval: 2.59 to 3.52]), and significantly lower rates of death (relative risk: 0.25 [95% confidence interval: 0.18 to 0.34]) and delisting (all p values <0.0001).
Because multiple listing requires migration between centers to hold simultaneous listings, we questioned whether the competitive advantage reflects this migration alone. We compared outcomes of SL patients with 2 nonoverlapping registrations who did not receive HT during the first listing and changed centers (n = 161) with those of patients who relisted at the same center (n = 415). After removing those who were delisted in error, rates of HT (65.2% vs. 61.7%; p = 0.4318) and death (8.7% vs. 12.3%; p = 0.2213) during the second listing did not differ significantly. There were no differences in AGI or insurance distribution between these groups (p = 0.6184 and 0.6190, respectively). Compared with ML patients, SL patients who changed centers had lower rates of private insurance (57.9% vs. 65.1%) and higher rates of Medicaid (15.2% vs. 18.2%; Fisher exact test, p = 0.0137 for all insurance types).
At first listing, 9.4% of SL and 19.7% of ML candidates were registered outside of the UNOS region in which they resided (p < 0.0001). Among ML candidates, 264 patients (39.1%) switched regions to obtain a secondary listing. Of these crossings, 195 (73.9%) were migrations to a bordering region and 157 (59.5%) to a region with a lower median waiting time than their primary region. Primary regional median waiting times for interregional candidates were higher than for those who remained within their initial region (128 days [IQR: 121 to 135 days] vs. 123 days [IQR: 99 to 128 days]; p < 0.0001). As shown in Figure 4, some regions (6, 7, and 9) were consistent “exporters” of ML candidates and some (2, 5, and 10) were consistent “importers.” Each importer region is bordered by an exporter region with a higher median waiting time and from which it generally received most of its in-migrating patients. The relative HT rate for interregional candidates was 1.16 (95% confidence interval: 1.06 to 1.27; p = 0.0011) and the relative death rate was 0.44 (95% confidence interval: 0.23 to 0.84; p = 0.0099). There was no significant association between AGI and the distance traveled between primary and secondary listing centers (p = 0.2361).
The present study indicates that: 1) adult patients who pursue multiple listing for HT have clinical and geographic characteristics that predict longer waiting time; 2) multiple listing is associated with enhanced HT access and lower waitlist mortality; 3) ML candidates seem to have higher income and are more likely to have private insurance; and 4) ML patients seem to have less medical need than SL patients throughout most of their waiting time.
Our findings agree with studies suggesting higher income among ML liver candidates (8,16) and higher transplant rates among ML liver and kidney candidates (4,17). Our study aligns with literature revealing that income and other socioeconomic factors determine outcomes of heart failure patients (18,19). Singh et al. (20) found household income was associated with waitlist mortality among pediatric HT candidates. Schwartz et al. (8) demonstrated higher waitlist mortality among lower-income adult liver candidates.
Among ML pediatric HT candidates, Feingold et al. (5) found a higher proportion of male and privately insured patients and trends toward less use of extracorporeal membrane oxygenation and greater need for prospective crossmatch in unadjusted analyses. There were no differences in race, income, blood type, diagnosis, ventilator dependence, or inotrope use among the underpowered sample. Multiple listing was more prevalent among our sample of adult patients (3.8% of listings vs. 0.4%). Although our definition of multiple listing was more relaxed, requiring overlap time between simultaneous listings as long as 2 weeks did not significantly change our multiple-listing prevalence.
The National Organ Transplant Act of 1984 established OPTN to ensure equitable organ distribution (21). OPTN is administered under contract by UNOS, which approved multiple listing in 1987 to grant patients increased control over their access to care (22). In the multiple-listing debate, patient autonomy has been weighed against concerns for justice. In consideration of a multiple-listing ban in 1988, the UNOS Board of Directors declared that the practice favored wealthier patients over others, creating “inequality of opportunity to receive a donated organ” (6). The proposed ban was tabled after response from patient advocacy groups. In 1990, New York banned statewide multiple listing for cadaveric renal transplantation (7). The ban had little apparent effect on equity, likely because of patients listing at out-of-state centers, and was overturned. Another consideration of ending multiple listing in 1995 was abandoned for concerns about patients with geographic and biologic impediments to timely transplantation (22). Disagreement continues. The American Medical Association states in its Code of Ethics that “patients should not be placed on the waiting lists of multiple local transplant centers, but rather on a single waiting list for each type of organ” (23).
In 2000, the Department of Health and Human Services issued the Final Rule, which states that organ allocation “shall be based on sound medical judgment” and calls for “setting priority rankings expressed… through objective and measurable medical criteria [that] shall be ordered from most to least medically urgent” (24,25). The Final Rule is intended to prioritize patients’ clinical need over geographic determinants of transplant access. Although the multiple-listing allowance is similarly intended, it allows socioeconomic status and physical mobility to prevail over medical acuity, thus standing at odds with the Final Rule’s mandate and undermining the medical community’s duty to reduce the impact of socioeconomic status on medical access.
The choice to engage in multiple listing seems rational in the economic sense. ML candidates have clinical and geographic traits, such as type-O blood, larger body sizes, congenital heart disease, and residency in UNOS region 9, that predict longer wait times in this and prior studies (26). But our analyses suggest that multiple listing increases HT access for patients who are objectively less sick at primary listing and have lower predicted waitlist mortality; competing risk analyses of the raw cohorts reveals lower ML mortality even though ML patients lag behind SL patients in HT rate for years after initial listing. Our study adds measures of utility pertaining to the multiple-listing debate. According to OPTN “the principle of utility … specifies that an allocation should maximize the expected net amount of overall good while minimizing harm” (27). A critical issue that we are studying is the impact of multiple listing on SL patients. Although our analyses suggest that ML candidates are a small percentage of patients, a larger proportion of lower-status SL patients at secondary centers may be affected. As we have shown, ML patients primarily seek secondary listings at centers with lower waiting times than their primary centers and are overwhelmingly more likely to be transplanted at the secondary versus the primary center. We therefore expect the possible detriment to SL patients of an ML candidate entering the secondary list would outweigh the benefit of the ML candidate opening a space on the primary list at time of HT.
We contend that the multiple listing allowance should be overturned, perhaps with an exception for candidates inactivated because of disease severity who might be granted an active secondary listing. A ban on multiple listing would not prevent patients from registering initially at centers distant from their primary residence. The potential impact of residency requirements may merit further study. A multiple-listing ban also would not prevent delisting at 1 center and then relisting at a second. But the multiple-listing allowance has not adequately remedied geographic inequities in donor heart access, and instead perpetuates existing socioeconomic disparities, and a ban alone would not fix these problems. The results of this study ultimately highlight the need to redesign allocation protocols to ensure more equitable organ distribution.
Important limitations of the current study include the inherent problems of analyzing registry data. Also, we assigned patients the median AGI of their home zip codes, which reduced our ability to detect an income effect given the internal heterogeneity of zip codes that have an average population of 30,000 persons (28). To be defined as an ML candidate, a patient must have enough mobility and survive long enough to register at a second center. For candidates who are too sick to pursue multiple listing or who die or receive HT before a secondary listing can be obtained, waitlist outcomes are assigned to the SL group, thus potentially imparting bias to our findings. In addition, the ability to draw causal inference between multiple listing and HT is limited by the tremendous demographic and clinical imbalance between the ML and SL cohorts. Some of these concerns are addressed by PSM on covariates that predict multiple listing. Although the PSM procedure produced cohorts that were well matched on observed covariates, unobserved confounding may have persisted (29). Lastly, our testing of a large number of hypotheses increases the possibility of individual type I errors but does not significantly diminish the overall robustness of our results.
Multiple listing for adult HT is a rational and effective response to organ shortage. But participation in multiple listing, and thus the benefit from it, is determined more by socioeconomic status than by medical need. We contend that the continued allowance of multiple listing stands at odds with the ethical principles that govern human organ allocation and that it should be reconsidered at the national level. The effects of multiple listing on the outcomes of SL and ML candidates for other solid organs deserve continued study.
COMPETENCE IN MEDICAL KNOWLEDGE: There is an insufficient supply of donor hearts to meet the needs of transplant candidates. Economic disparities contribute to inequities in appropriate access to heart transplantation.
TRANSLATIONAL OUTLOOK: The allowance of multiple simultaneous transplant listings should be overturned. Additional study is needed to quantify the impact of multiple listing on single-listed candidates.
Supported by the Health Resources and Services Administration (contract 234-2005-370011C); research grants from the International Society for Heart and Lung Transplantation, and the Heart Failure Society of America (to Dr. Givens); and by grants from the National Heart Lung Blood Institute (K23 HL095742-01 and P30 HL101272-01 to Dr. Schulze). Dr. Dardas has received a research grant paid to his institution from the American College of Cardiology/Daiichi Sankyo Corp.; and a travel education grant paid to a third party from Heartware and Thoratec. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- adjusted gross income
- heart transplantation
- interquartile range
- Organ Procurement and Transplantation Network
- propensity score matching
- United Network for Organ Sharing
- Received April 7, 2015.
- Revision received July 6, 2015.
- Accepted July 9, 2015.
- American College of Cardiology Foundation
- ↵Internal Revenue Service. SOI Tax Stats- Individual Income Tax Statistics- ZIP Code Data (SOI). Available at: http://www.irs.gov/uac/SOI-Tax-Stats-Individual-Income-Tax-Statistics-ZIP-Code-Data-%28SOI%29. Accessed February 28, 2015.
- ↵Agency for Healthcare Research and Quality. Creating and validating an index of socioeconomic status. Available at: http://archive.ahrq.gov/research/findings/final-reports/medicareindicators/medicareindicators3.html. Accessed March 15, 2015.
- ↵U.S. Department of Housing and Urban Development. HUD USPS ZIP Code Crosswalk Files. Available at: http://www.huduser.org/portal/datasets/usps_crosswalk.html. Accessed February 16, 2014.
- ↵Ho DE, Imai K, King G, Stuart EA. MatchIt: nonparametric preprocessing for parametric causal inference. Available at: http://www.jstatsoft.org/article/view/v042i08. Accessed October 27, 2015.
- Stevens L.A.,
- Coresh J.,
- Feldman H.I.,
- et al.
- Singh T.P.,
- Milliren C.E.,
- Almond C.S.,
- Graham D.
- Foraker R.E.,
- Rose K.M.,
- Suchindran C.M.,
- Chang P.P.,
- McNeill A.M.,
- Rosamond W.D.
- ↵National Transplant Act (NOTA; P.L. 98–507), 1984 Available at: http://www.gpo.gov/fdsys/pkg/STATUTE-98/pdf/STATUTE-98-Pg2339.pdf. Accessed October 8, 2015.
- ↵American Medical Association. Opinion 2.16. Organ Transplantation Guidelines. Code of Medical Ethics. American Medical Association, 2014. Available at: http://www.ama-assn.org/ama/pub/physician-resources/medical-ethics/code-medical-ethics/opinion216.page?. Accessed October 8, 2015.
- ↵Duda L. National organ allocation policy: the final rule. Available at: http://journalofethics.ama-assn.org/2005/09/hlaw1-0509.html. Accessed November 18, 2015.
- U.S. Government Publishing Office. Code of Federal Regulations. Title 42. Allocation of Organs. 1999. Available at: http://www.ecfr.gov/cgi-bin/text-idx?tpl=/ecfrbrowse/Title42/42cfr121_main_02.tpl. Accessed October 8, 2015.
- Yang J.A.,
- Naka Y.,
- Ota T.,
- et al.
- ↵Organ Procurement and Transplantation Network. Ethical principles in the allocation of human organs. Available at: http://optn.transplant.hrsa.gov/resources/ethics/ethical-principles-in-the-allocation-of-human-organs/. Accessed March 2, 2015.
- Krieger N.,
- Chen J.T.,
- Waterman P.D.,
- Soobader M.J.,
- Subramanian S.V.,
- Carson R.