Health Affairs, 23, no. 2 (2004): 191-198 doi: 10.1377/hlthaff.23.2.191 © 2004 by Project HOPE	New Online   Pay Cuts For Medicare Docs   Access To Care Woes   Public Coverage More Efficient   Empowering Consumers

This Article Abstract Figures Only Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by Scanlon, D. P. Articles by Ubel, P. A. Search for Related Content PubMed PubMed Citation Articles by Scanlon, D. P. Articles by Ubel, P. A. Related Collections Consumer Issues Ethical Issues Legal/Regulatory Issues Research And Technology

Health Tracking

MARKETWATCH

Does Competition For Transplantable Hearts Encourage ‘Gaming’ Of The Waiting List?

Abstract

 
Transplant centers may "game" the severity of listed patients to increase their patients’ likelihood of receiving transplantable organs. Recent lawsuits allege gaming at some centers, and listing policies were modified in 1999 to clarify listing criteria. We tested for gaming and its relationship to heart transplant center competition. We found that increased competition resulted in more patients listed in the most severe illness category (p <.01), consistent with the gaming hypothesis. Gaming was mitigated after the 1999 policy change (p >.05), which suggests that the new rules were effective. Continued monitoring is warranted, given prior gaming and recent accusations.

The United Network for Organ Sharing (UNOS) has developed allocation procedures that give priority to the most urgently ill patients.² Prior to 1999 patient severity was indicated by two categories: Status 1 for patients with a projected life expectancy of less than six months without a transplant; and Status 2 for those with a projected life expectancy of more than six months.³ Generally, patients with total artificial heart, ventricular assist system, intra-aortic balloon pump, or ventilator, or hospitalized in an intensive care unit (ICU) and requiring inotropic agents, were automatically classified as Status 1 with no restrictions on duration. Because the Status 1 category was so broad and patient severity was subject to neither clinical verification nor physician recertification, some experts worried that transplant centers could "game the system" by listing patients as Status 1 who should have been listed as Status 2.⁴

UNOS revised its listing procedures in 1999, dividing Status 1 into Status 1A and Status 1B. Status 1A was reserved for patients with a projected life expectancy of less than one month.⁵ Status 1B was reserved for patients with a projected life expectancy between one and six months.⁶ Physicians of Status 1A patients are required to recertify to UNOS the qualification of the 1A status every seven or fourteen days if their patient does not require mechanical circulatory support for acute hemodynamic decompensation.⁷ Even patients with mechanical circulatory support through a ventricular assist device are only guaranteed thirty days at status 1A before physician recertification becomes necessary.

Concerns remain about the potential to game the waiting list. Gaming could occur by prematurely admitting patients to the ICU or by using intravenous inotropic agents, indwelling hemodynamic monitors, or mechanical assistance earlier than necessary.⁸ These concerns have become more visible recently, particularly in liver transplantation, with federal and state lawsuits against three Chicago-area medical centers, alleging the exaggeration of patients’ illness severity.⁹

We developed a method of testing whether the transplant listing system is being gamed. Our method takes advantage of variation in competition across organ procurement organizations (OPOs). The UNOS allocation system assigns priority to the sickest patients who are listed within the same OPO where the organ was procured, because cadaveric hearts quickly suffer from ischemic damage following procurement. Consequently, each of the fifty-five OPOs represents a unique market for available hearts in which transplant centers compete against each other. OPOs differ in both the number of competing centers and their competitiveness—some have only one transplant center, which is a monopoly by definition, while others have as many as nine centers, many of which have substantial market share. This variation in competitiveness provides an opportunity to explore whether transplant centers game the system. Using data from UNOS, we studied the correlation between the degree of competition faced by a transplant center within an OPO and the proportion of patients listed as Status 1 (prior to 1999) or 1A (after 1999).

Data And Analysis

Top Data And Analysis Results Discussion Editor's Notes NOTES

 
Data for this study were obtained from UNOS’ Organ Procurement Transplant Network database from 1995 to 2000. The unit of analysis is the transplant center. Our six-year panel includes aggregate information about patients on the cardiac transplant waiting list at each center, including demographic characteristics (age, sex, blood type) and listing status (Status 1 or 1A). The data also contain aggregate transplant center-level information on patients who received a heart transplant.

Of the 172 centers performing heart transplants, we eliminated those that exclusively performed pediatric heart transplants. We retained centers that performed at least one adult heart transplant in any given year. The remainder ranged from a low of 114 centers operating in 1998 to a high of 120 in 1995.

The goal of our empirical analysis was to determine whether the proportion of a center’s patients that are listed as Status 1 (pre-1999) or Status 1A (post-1999) was a function of competition. We measured competition in two ways, using the number of transplant centers within an OPO and the Hirschman-Herfindahl Index (HHI), which is based on a transplant center’s share of heart transplants within its OPO. The HHI is a standard measure of competition used by the U.S. Department of Justice (DOJ) and Federal Trade Commission (FTC) in antitrust proceedings.¹⁰ Since our data indicate that the average number of transplants falls with greater numbers of centers in an OPO, we also included the number of centers in an OPO as a measure of competition. Separate models were estimated to compare the fit and robustness of each measure.¹¹

Our dependent variable is the proportion of listed patients in the most severe illness category (Status 1 pre-1999 and Status 1A post-1999) for a given transplant center and OPO. A logit transformation was applied to this proportion and fitted to a linear mixed effects model, where clustering within transplant centers and OPOs was controlled using random effects.¹² In addition to the competition measures, we also include as covariates the average age of a transplant center’s listed patients, the proportion who were female or had type O blood (waiting times are typically longer), year indicator variables to account for time trends, and a measure of how frequently listed patients within an OPO receive a transplant (which we call list turnover).¹³

Results

Top Data And Analysis Results Discussion Editor's Notes NOTES

 
Transplant centers performed an average of seventeen transplants per year (Exhibit 1). The proportion of patients listed as Status 1 increased from 12 percent in 1995 to 18 percent in 1998. There was substantial variation across transplant centers as indicated by the large standard deviations relative to the average values. The proportion of patients listed as Status 1A rose slightly between 1999 and 2000, from 3.4 percent to 4.9 percent (p =.004).

The average OPO contained between two and three transplant centers (Exhibit 2). The most crowded OPOs had nine transplant centers, and there were fourteen to nineteen monopoly OPOs in any given year (about 12–16 percent of centers). The average HHI, based on transplant centers’ shares of OPO transplants performed, rose from 0.65 in 1995 to 0.70 in 2000, which suggests that OPOs became, on average, less competitive over time. OPO waiting list turnover ranged from 0.30 to 0.38, which suggests that about one-third of all listed patients, on average, received a transplant in a given year.

The estimates for the other covariates are fairly stable regardless of which measure of competition is used. The average age of listed patients was not significantly related to the probability of listing patients in the most severe listing status. In both models and both periods, transplant centers with higher proportions of female patients had relatively higher proportions of patients listed as Status 1 or 1A, although these estimates were not significantly different from zero. The effect of blood type was not significantly different from zero for either model in either period. The year indicator variables suggest that the proportion of patients listed as Status 1 was significantly higher in 1998 than in 1995 or 1996, consistent with the trend observed in Exhibit 1.

There were also significantly more patients listed as Status 1A in 2000 than in 1999. Finally, the list turnover variable is positive and significant in the prepolicy period, which suggests that increased turnover of listed patients resulted in more patients’ being listed as Status 1. There was no significant effect of turnover in the postpolicy period.

Exhibit 4 illustrates the relationship between within-OPO competition and listing patterns. Note that higher proportions of patients are listed as Status 1 as competition increases (that is, HHI decreases). Similarly, as the number of transplant centers in the OPO increases, higher proportions of patients are listed as Status 1 (data not shown). The flat relationship between competition and listing practice for 1999–2000 suggests that after UNOS’s listing policy changed, the effect of competition disappeared (Exhibit 4).

View larger version (18K): [in this window] [in a new window]   EXHIBIT 4 Estimated Relationship Between Competition And Patient Listing, 1995–1998 And 1999–2000

Discussion

Top Data And Analysis Results Discussion Editor's Notes NOTES

 
This study is the first to empirically identify gaming of the heart transplant waiting list by centers in competitive OPOs. Prior to UNOS’s policy change in 1999, heart transplant centers in competitive OPOs were more likely than centers in less competitive OPOs were to list their patients as Status 1. After the UNOS policy change, transplant centers in competitive markets failed to exhibit a greater propensity to list patients in the sickest category. This is likely to have occurred because the clear definitions used in the new policy make it harder for transplant centers to upgrade patients to Status 1A. In addition, because Status 1A patients receive priority over Status 1B patients, and since most transplants go to patients listed at Status 1A, centers may have little incentive remaining to move patients from Status 2 to Status 1B.

We have not proved that the association we found between OPOs’ competitiveness and listing behavior is a causal relationship. It is possible that competitive OPOs have sicker patients than other OPOs have. However, a few considerations make us confident that we have found evidence of gaming. First, we adjusted for several other factors, which, if distributed unevenly across OPOs, could have led to disparities in listing decisions. If centers in a specific OPO had a disproportionate number of patients with type O blood, for example, these patients would be expected to wait longer for organs than other patients, and would therefore have had more time for their end-stage heart disease to progress. We adjusted for this factor, as well as for age, sex, and transplant turnover.

The finding that the association between OPO competitiveness and gaming vanished after implementation of the UNOS policy change reinforces our hypothesis of gaming. If an alternative factor accounted for the association between OPO competition and number of sick patients, then the pooled analysis should have yielded results similar to those shown in Exhibit 3. In other words, the association between competitiveness and listing behavior should have persisted beyond 1999. But instead, the association vanished after 1999. Something changed between 1998 and 1999 that reduced the association between OPO competitiveness and listing practice. The most plausible explanation is that the new UNOS rules reduced gaming behavior.

Our study does not conclusively demonstrate that the listing practices occurring through 1998 were deliberate or dishonest. Determining the prognosis of patients with end-stage heart disease is an imprecise science, and some of the influence of competition on listing decisions could have been subconscious. In addition, clinicians may have looked to the behavior of institutions in the same OPO to determine the best and most internally consistent method of interpreting the listing rules. However, even if centers acted subconsciously, the net effect was the same.

We think that gaming is a problem. First, it is impossible to know when the playing field is level; therefore, transplant centers will remain suspicious about the behavior of competitors if they sense that gaming is occurring. Second, clinical decisions to upgrade patients’ status could lead to excessive resource use. Centers may prematurely admit patients to ICUs to elevate them on the waiting list. This not only has economic implications but also could harm patients by subjecting them to unnecessary medical procedures.¹⁴ Third, when centers place less acutely ill patients higher on the waiting list, they potentially transfer organs from more deserving patients. Gaming of the waiting list hinders the list’s ability to allocate organs to the sickest patients.

Study limitations. Our study has several limitations. First, our data represent aggregated data for all listed and transplanted patients within a transplant center. Patient-level data for each center would have been preferable but were unavailable. A second limitation relates to the potential bias due to important omitted explanatory variables. However, these omitted variables would not affect our conclusions about competition unless they were correlated with the HHI or the number of transplant center variables.

Policy implications. Our study has important implications for transplant policy. We have shown that when transplant listing criteria are too broad and not subject to verification, transplant centers facing greater competition for cadaveric organs will game the system. We have also demonstrated that more specific transplant listing policies can effectively reduce gaming. We encourage the transplant community to continue to monitor the system for recurrence of gaming behavior. Because pressure to game the system may increase over time, there is a need to remain vigilant to the possibility that gaming will return.

Editor's Notes

Top Data And Analysis Results Discussion Editor's Notes NOTES

 
The authors thank Bobby Ahluwahlia for preliminary work leading to the analysis presented in this paper and Yulin Cheng from the United Network for Organ Sharing (UNOS) for assistance with data. They also thank Dan Polski; David Meltzer; and seminar participants at the University of Chicago, Johns Hopkins University, University of Michigan, Penn State University, and Department of Veterans Affairs (VA) Newport Health Economics Conference for helpful comments on earlier versions of the paper. Peter Ubel is supported by a Presidential Early Career Award for Scientists and Engineers. The views expressed are those of the authors and not of UNOS.

Dennis Scanlon (dpscanlon{at}psu.edu) is an assistant professor in the Department of Health Policy and Administration, Pennsylvania State University, in University Park. Christopher Hollenbeak is an assistant professor in the Departments of Surgery and Health Evaluation Sciences, Pennsylvania State College of Medicine, in Hershey. Woolton Lee is a doctoral candidate in the Department of Health Policy and Administration, Penn State. Evan Loh is assistant vice president of cardiovascular/infectious disease at the Wyeth Corporation in Philadelphia. Peter Ubel is an associate professor of internal medicine and psychology and directs the Program for Improving Health Care Decisions at the University of Michigan and Ann Arbor Veterans Affairs Medical Center, Ann Arbor, Michigan.

NOTES

Top Data And Analysis Results Discussion Editor's Notes NOTES

 

1. University Renal Research and Education Association, United Network for Organ Sharing, 2000 Annual Report of the U.S. Organ Procurement and Transplantation Network and the Scientific Registry of Transplant Recipients: Transplant Data 1989–1999, 18 February 2000, www.optn.org/data/annualReport.asp (16 December 2003).
2. J.F. Burdick, J. DeMeester, and I. Koyama, "Understanding Organ Procurement and the Transplant Bureaucracy," in Transplantation, ed. L.C. Ginnis, A.B. Cosimi, and P.J. Morris (Malden, Mass.: Blackwell Science, 1999), 875–896.
3. A.M. Harper et al., "The United Network for Organ Sharing/Organ Procurement and Transplantation Network Waiting List and Donor Registry," in Clinical Transplants 1998, ed. J.M. Cecka and P.I. Terasaki (Los Angeles: UCLA Tissue Typing Laboratory, 1998), 73–90.
4. See the discussion in A.C. Anyanwu, C.A. Rogers, and A.J. Murday, "Variations in Cardiac Transplantation: Comparisons between the United Kingdom and the United States," Journal of Heart and Lung Transplantation 18, no. 4 (1999): 297–303; and T.V. Votapka et al., "Heart Transplantation Charges: Status 1 versus Status 2 Patients," Journal of Heart and Lung Transplantation 14, no. 2 (1995): 366–372. A rise of 17.4 percent in the proportion of Status 1 patients and a concurrent decline of 33 percent in Status 2 patients across all heart transplant centers between 1988 and 1995 is cited in D.J. Whellan et al., "Heart Transplant Center Practice Patterns Affect Access to Donors and Survival of Patients Classified as Status 1 by the United Network for Organ Sharing," American Heart Journal 140, no. 3 (2000): 443–450.[Medline]
5. As defined by admission to an ICU with a requirement for inotropic agents or with need for continuous indwelling central hemodynamic monitors or both, or who requires mechanical assistance. See UNOS, "Resources, Policies," www.unos.org/policiesandbylaws/policies.asp?resources=true (10 October 2003).
6. As defined by the need for continuous intravenous inotropic therapy. See Whellan et al., "Heart Transplant Center Practice Patterns."
7. This support must include implantation of left (or right) ventricular assist device, total artificial heart, intra-aortic balloon pump, or extra-corporeal membrane oxygenator (ECMO). Patients at Status 1A on mechanical circulatory support or ventilation need recertification every fourteen days, while those needing infusion of inotopes need recertification every seven days. See UNOS, "Resources, Policies."
8. Votapka et al., "Heart Transplantation Charges"; and Anyanwu et al., "Variations in Cardiac Transplantation"
9. See, for example, T.M. Burton and A. Merrick, "U.S. Alleges Liver-Transplant Fraud," Wall Street Journal, 29 July 2003; S. Warmbir, "UIC Hospital Sued for Medicare Fraud," Chicago Sun-Times, 29 July 2003; and M. O’Connor, "Hospitals Accused of Transplant Fraud," Chicago Tribune, 28 July 2003.
10. The HHI for each OPO is the sum of squares of the market shares of each transplant center. We calculated market share as the percentage of transplants performed by each center in its OPO. In the extreme case, when there is only one transplant center in an OPO, that center will have a market share of 100 percent, and the relevant HHI will take on a value of 10,000 (we rescaled to 1 in our analysis). At the other extreme, there are many transplant centers in an OPO, each with less than 1 percent of the total market share, resulting in an HHI near zero.
11. We first tested whether the presence of any competition had an impact on listing status. This was done using a binary indicator for OPOs with only one transplant center. Results (not shown) suggested that competition was a significant predictor of listing status, so we proceeded to study the importance of the degree of competition. More centers in an OPO may increase the incentive to game if the result is to drive down the average number of transplants per center, making it relatively less likely for a center to receive an organ.
12. Our model equates the proportion (under a logit transformation) of transplant center i’s patients in OPO j at time t that are listed in the most severe illness category (Status 1 prior to 1999 and Status 1A after 1999) to a sum involving an estimated constant and a zero mean, normally distributed error term plus two products involving estimated model parameters and vectors covering transplant center specific characteristics (for example, aggregate patient demographic characteristics) and OPO-specific characteristics (for example, the number of transplant centers and the HHI). To account for the clustering of transplant centers within OPOs, we assumed OPO and transplant center random effects and fitted the data to a linear mixed effects model. See N.M. Laird and J.H. Ware, "Random Effects Models for Longitudinal Data," Biometrics 38, no. 4 (1982): 963–974. A detailed methods appendix is available from Dennis Scanlon, dpscanlon{at}psu.edu.[CrossRef][ISI][Medline]
13. We defined "list turnover" as the total number of transplants performed annually in the OPO divided by the total annual size of the waiting list in the OPO. List turnover measures the frequency by which patients listed in an OPO receive transplants. Low turnover could result in longer waiting times and increased illness severity of listed patients and could raise the number of patients listed as Status 1/1A independent of competition.
14. R.C. Starling, "Economics of Heart Transplantation," in Primer on Transplantation, ed. D.J. Norman and W.N. Suki (Thorofare, N.J.: American Society of Transplant Physicians, 1998), 405–407.

                      What's this?

This article has been cited by other articles:

				S. D. Halpern, A. Shaked, R. D. Hasz, and A. L. Caplan Informing Candidates for Solid-Organ Transplantation about Donor Risk Factors N. Engl. J. Med., June 26, 2008; 358(26): 2832 - 2837. [Full Text] [PDF]

				J E Stahl, A C Tramontano, J S Swan, and B J Cohen Balancing urgency, age and quality of life in organ allocation decisions--what would you do?: a survey J. Med. Ethics, February 1, 2008; 34(2): 109 - 115. [Abstract] [Full Text] [PDF]