Indiana Law Review Getting a Handle on Coverage Decisions: If Not Case Law, Then What? Comments on a Paper by Professor William Sage Maxwell J. Mehlman* Professor Sage has written an important paper on an important issue. The lack of good empirical work on health law in general, and on the activities of health insurers in particular, is a serious impediment to sound policy making. A good example is the controversy over genetic discrimination in access to health insurance. Much has been written lately about the problem,^ but no one has any idea of how widespread or serious the problem may be. Without this information, it is difficult to determine the appropriate public policy response. I am persuaded by Professor Sage's arguments that case law is a poor source ofempirical data on coverage decision making by health care insurers. However, I am less optimistic than he appears to be that any good sources of quantitative data on the subject will be forthcoming in the near future. Let me explain why. First, Professor Sage seems to feel that good empirical data are available in one area of health law, namely, medical malpractice. He states that "medical malpractice data are far more organized and accessible"^ than data on health insurance coverage, declaring for example, that "data on nationwide jury awards are available from Jury Verdict Research . . . ."^ However, the Jury Verdict reporting system, which, as Professor Sage recognizes, is based largely on self- reporting by attorneys,"^ is generally regarded as incomplete and unsuitable for quantitative research purposes.^ Indeed, some of the most basic questions that we might want to ask about the health care system, such as whether malpractice incidence is higher in managed care organizations than in traditional fee-for- service settings, remain unanswerable given the present state of our data. This makes it all the more discouraging that for coverage decisions we lack data even as good as that we have on malpractice. As Professor Sage recognizes, knowledge of coverage decision making is critical to enable us to determine * Arthur E. Petersilge Professor of Law and Director, The Law-Medicine Center, Case Western Reserve University. L See, e.g., Paul R. Billings et al.. Discrimination as a Consequence ofGenetic Testing, 50 Am. J. Hum. Genetics 465 (1992); Richard A. Bornstein, Genetic Discrimination, Insurability and Legislation: A Closing ofthe Legal Loopholes, 4 J.L. & POL'Y 551 (1996); Kathy L. Hudson et al.. Genetic Discrimination and Health Insurance: An Urgent Needfor Reform, 270 SCIENCE 391 (1995). 2. William M. Sage, Judicial Opinions Involving Health Insurance Coverage: Trompe L Veil or Window on the World?, 31 IND. L. REV. 49, 59 (1998). 3. Id 4. Id 5. See, e.g.. Marc Galanter, Real World Torts: An Antidote to Anecdote, 55 Md. L. Rev. 1093, 1113 n.63 (1996) ("Because [Jury Verdict Research] does not provide information about the number of awards in its data base, there is no way to tell how thoroughly it represents the entire universe of awards."). 76 INDIANA LAW REVIEW [Vol. 31:75 whether patients are obtaining access to new or experimental medical technologies; to identify the criteria third-party payors are using to make these coverage decisions; and, ultimately, to discern how private health care rationing schemes are operating in our present healthcare system.^ Clearly we need this information. The question is, where and how can we get it? Professor Sage suggests at the end of his paper that these data will be forthcoming from academic health centers, pharmaceutical companies, government-mandated reports, and the managed care organizations themselves.^ While I agree with Professor Sage that some additional data will be available from these sources in the future, I am pessimistic that we will obtain easy access to substantial amounts of useful information. To understand why, let me list the types of actions we need to know about in order to have a good understanding of what is happening in the realm of coverage decision making. 1. Whether or Not a Person is Insured.—First, a third-party payor could refuse to cover health care services for an individual by declining insurance to that individual. Therefore, to have a complete picture of coverage, we would need information on insurability decisions by insurers. We also would need to know whether employers offer health care plans to their employees, and, if so, what types of plans they offer. Finally, we would need to know what choices enrollees or potential enrollees make when they are offered plan options by employers or within the insurance market in which they reside. In the Cleveland metropolitan area, for example, employees not offered access to Rainbow Babies and Children's Hospital through their health plan would most likely not have coverage for the specialized services that hospital provides. 2. Premium Levels.—^A second way an insurer could decide not to cover services for an individual is to price coverage beyond the individual's ability to pay. Therefore, a complete picture of health care coverage would include knowledge of the premiums associated with different types of plans or coverage options, along with the co-payment and deductible rules for each option. 3. Pre-existing Conditions.—^An insurer may try to avoid covering a service by declaring the condition for which the service is required a "pre-existing condition."^ We would need to know insurers' policies on pre-existing conditions and understand the impact of the Health Insurance Portability and Accountability Act of 1996^ on those policies. 4. Recommendations by Primary Care Physicians.—One of the most difficult types of coverage decisions to discern is the information primary care physicians give to their patients about the suitability of particular services for the 6. Sage, supra note 2, at 50. 7. Id. at 13. 8. A "pre-existing condition," in general, is one that would "otherwise [be] within the coverage of the policy, [but] which existed prior to [the policy's] effective date." George J. Couch, Couch Cyclopedia of Insurance § 41 A: 17 (2d rev. ed. 1982). Specific definitions of what constitute "pre-existing conditions" may vary among insurance companies. 9. Pub. L. No. 104-191, 1 10 Stat. 1936 (1996) (codified in scattered sections of 18 U.S.C, 26 U.S.C, 29 U.S.C, and 42 U.S.C). 1998] COVERAGE DECISIONS 77 patient, the availability ofthose services in the geographic area, and coverage of those services under the patient's specific health care plan. Particularly in managed care plans offering financial incentives to physicians for discouraging patients from obtaining expensive health care services, the physicians may make implicit coverage decisions simply by not informing patients about the benefits of a specific service. For example, Henry J. Aaron and William B. Schwartz have shown that physicians in Britain avoided providing kidney dialysis to patients over the age of fifty-five simply by not informing the patients of the availability of dialysis services. '° 5. Referral Decisions by Gatekeepers.—^Managed care organizations increasingly use a system whereby primary care physicians, placed in the role of gatekeepers, are expected to restrict patient access to costly services. For example, the physician may decline to refer the patient to a specialist within the plan, or to a specialist outside the plan, or to a facility within or outside the plan. All of these decisions are denials of access tantamount to negative coverage decisions. 6. Outright Coverage Policy.—Third-party payors typically include coverage language both in their policies and in that portion of the policies provided to enrollees as their description of covered services. To my knowledge, no one has made a comprehensive compilation of policy language from third- party payors around the country. Moreover, much of this policy language is vague, as Professor Sage recognizes in his description of the role of judicial oversight.'^ Thus, one would need to know how the general policy language is interpreted in practice by the health plan. In trying to determine the coverage policy under various state Medicaid programs, for example, we have had to survey the programs directly to determine whether or not they cover a specific technology for a specific patient population. Even so, some responses indicate that to a large extent Medicaid makes coverage decisions on a case-by-case basis regardless ofthe general statutory language governing Medicaid coverage policy. 7. Utilization Review Decisions.—A primary example of case-by-case interpretations of policy language is utilization review decisions by health plan personnel. These include pre-certification decisions authorizing hospital admissions, extensions of lengths-of-stay, and authorization for specific procedures. 8. Decisions by Plan Medical Directors.—^Decisions interpreting general policy language and utilization review decisions by administrative personnel may be reviewed by medical directors when challenged by enrollees or their providers. We need to know the fate of these challenges. 9. Claims Determinations.—^Another way in which coverage decision making takes place is through retrospective utilization review. This includes decisions by third-party payors not to reimburse providers and/or enrollees for services already supplied. These refusals color future recommendations by the 1 0. Henry J. Aaron & William B. Schwartz, The Painful Prescription: Rationing Hospital Care 101 (1984). 1 1 . Sage, supra note 2, at 58. 78 INDIANA LAW REVIEW [Vol. 3 1 :75 providers to patients and influence future efforts by the same enrollees to obtain similar services. Over time, reimbursement decisions may shape general plan coverage policies. 10. Judicial Review.—Finally, we come to the focus of Professor Sage's paper: cases in which courts review coverage decisions by third-party payors. Given our need for much or all of the foregoing data to give us an accurate picture of coverage decision making and its effect on enrollees in our health care system, how available is this information at this time? We have some information on who is insured and who is not and some information on employer insurance practices. Although enrollees should have policy language for the plans in which they are enrolled, no one as yet has attempted to gather this policy language in one place, and any effort to do so would be hampered by changing circumstances, such as plans coming into and going out of business, and changes in policy language. We do have judicial opinions, but Professor Sage has done a superb job of persuading us how limited that database is for our purposes.'^ Given that so much of this information is not available at this time, is Professor Sage correct in predicting that it will become available in the near future?'^ Professor Sage seems to think that we will get a good deal of it from the managed care organizations themselves. But why should they gather this information? They are primarily interested in data that would affect their profitability, and therefore in information about the behavior of providers within their network or providers being considered for membership. Therefore, we might expect them to gather information on provider practice patterns and referral decisions, mentioned in items four and five above. But it is unlikely that health plans would have an interest in gathering much ofthe other data we desire. Moreover, even if they collected the data on practice patterns and referrals, or indeed any of the other data on our list, why would they make this information public? Assuming they did make it public, how could we verify the accuracy and completeness of the information they disclosed? Even if an isolated managed care organization gave us access to this information, how could we be sure that what we learned from it reflected the practice in other managed care organizations? The primary force driving the behavior of health care organizations is increasing competition. Should we expect competition to lead third-party payors to publicize their coverage decisions in order to gain a competitive advantage? This is hard to imagine. What health care plan, for example, would want to advertise that it covers an expensive service that other plans do not? Would this not lead to adverse selection by enrollees? Professor Sage does not rely exclusively on the private sector to provide access to coverage decisions. In addition, he states that "government will mandate reporting by the full range of regulated entities, and will make that information available to researchers."''* Again, I fear this is too optimistic. The 12. Mat 6 1-68. 13. Id. at73. 14. Id 1998] COVERAGE DECISIONS 79 one potential source of good malpractice data is the National Practitioner Data Bank and it is true that this database has been created by the federal government. But the data are not accessible to researchers, let alone to the public.'^ Nor does it appear that the government is making a systematic effort to ensure that reporting to the National Practitioners Data Bank is accurate and complete. Even if the government wqyq to require similar reporting for coverage decisions, we would be hard pressed to ensure that they too were accurate and complete. Furthermore, would the government take this step when doing so would increase the administrative burden on third-party payors? If the answer is that medical records will be computerized and therefore much easier to retrieve and analyze, does this not raise privacy questions that might block government data collection efforts? Finally, how could the government obtain information on referral practices and recommendations by primary care physicians? The only way would seem to observe physician office practice directly, a solution either impractical, unethical, or both. So far I have only been talking about getting information from third-party payors themselves. Another potential source of information might be enrollees. Some studies of discrimination and access to health care services, for example, have relied on self-reporting by enrollees or potential enrollees.^^ But it is well recognized that patient self-reporting is inaccurate, biased and incomplete. An interesting possibility is to identify and contact enrollees to get their permission to access their medical records, and then to analyze the records to ascertain what coverage decisions may have been made in their cases. Gerald Hickson and his associates have used a similar technique to determine why patients who have arguably been the victims of medical malpractice file claims.^^ But this promises to be an extremely expensive and laborious process, and it is unclear whether sufficient information could be ascertained by this method. In conclusion, I want to stress two points. First, the difficulty of obtaining data in no way diminishes our need for it. Second, I think we are going to be limited to more theoretical, extrapolative efforts to examine coverage decisions for the foreseeable future. 15. 42U.S.C.§ 11137(1994). 16. See E. Virginia Lapham et al.. Genetic Discrimination: Perspectives ofConsumers, 274 Science 621 (1996). 17. Gerald B. Hickson et al.. Factors that Prompted Families to File Medical Malpractice Claims Following Perinatal Injuries, 267 JAMA 1359 (1992).