Joseph P. Newhouse

1. The Larger Issue: Financing Medical Research Out of the Trust Fund
   1. Graduate medical education
   2. Disproportionate share payments
   3. Financing research from the trust fund
2. Costs of Clinical Research
3. Summary
4. References

The instructions I received for this paper were to prepare some remarks on funding medical research through Medicare. The rationale for such funding appears to be implicit in the second paragraph of the statement on the Workshop's Goals and Objectives:

"Despite the United States' recognition of medical research as a public good and the sustained support it has enjoyed, this research still relies on the annual decisions of congressional authorizers and appropriators. Unfortunately, the scientific community cannot always rely on federal support for medical research to consistently parallel or provide for future scientific opportunities or, more importantly, public health needs."

As I translate this paragraph, it says the needs of the research community are not well served by the vagaries of the appropriations process; as a result, let those of us who are concerned about support for medical research look to what are elsewhere in the meeting materials termed "unconventional" sources for support. Following Willie Sutton's principle of robbing those institutions where the money was, namely banks, the organizers ask whether Medicare should pay for medical research.

I expect I will disappoint several of you by saying that I personally do not agree with the general principle that Medicare should finance medical research. I say this even if health services research were included in the definition of health care research, as I believe it should be, and thereby received increased funding. In the interest of full disclosure, any increased funding for health services research could benefit me personally. I do, however, favor a limited change in policy that would permit Medicare to pay for experimental procedures that are being tested in clinical research. I take these two issues in turn.

The Larger Issue: Financing Medical Research Out of the Trust Fund
It would be tempting but wrong to say that the Medicare trust fund only finances medical services for its beneficiaries and so it should remain pure and not finance research. In other words, there is precedent for using Medicare monies for purposes other than simply paying for medical services. In particular, the Congress has authorized payments from the Medicare Part A trust fund for graduate medical education and for disproportionate share hospitals. Some of the graduate medical education monies may well be used to cross-subsidize clinical research, but I am excluding those monies from this discussion.

Graduate medical education Since its inception Medicare has paid a share of salaries of residents and certain other costs of graduate medical education such as the training director and overhead costs. Because these costs were part of teaching hospital budgets and because Medicare originally reimbursed hospitals for its share of costs, it was natural for the original program to finance such costs, which came to be called direct medical education costs.

In 1984 Medicare changed its method of reimbursing hospitals to prospective payment, or a lump sum for each admission. Because teaching hospitals had higher costs, they successfully argued that over and above the payment of resident salaries they needed higher reimbursement to be able to compete with non-teaching hospitals. As a result, the Congress authorized additional payments for each Medicare case in a teaching hospital, with the amount dependent on the hospital's resident-to-bed ratio. This is known as payment for indirect medical education costs.

Exactly what accounts for the higher patient care costs at teaching hospitals is not clear, but at least some of the cost may be attributable to the cost of clinical research. Thus, as pointed out above, Medicare could well be indirectly paying for some clinical research.

The amounts paid for graduate medical education are non-trivial; the direct and indirect payments together were $6.8 billion in 1997. There is currently considerable debate about the structure of these payments, and both the Bipartisan Commission on the Future of Medicare and the Medicare Payment Advisory Commission are to report to the Congress in 1999 regarding these payments.(2)

Disproportionate share payments Medicare also pays additional amounts to certain hospitals under the Disproportionate Share program. These monies are intended to preserve and enhance access to services for poor Medicare beneficiaries and for low-income individuals more generally by increasing the Medicare payment per case at hospitals treating substantial numbers of Medicaid and dual eligible patients. The original justification for this program was some evidence that low-income Medicare beneficiaries were more costly to treat, but subsequent research has suggested this is not the case, and the monies are now seen as more generally promoting access to care for low-income populations (Kominski and Long, 1997). These monies are also non-trivial, amounting to $4.5 billion in 1997, and they go disproportionately to teaching hospitals (Medicare Payment Advisory Commission, March 1998).

In short, because of graduate medical education and disproportionate share payments, the Medicare program does more than simply finance medical care for its elderly and disabled beneficiaries. Should it therefore also finance medical research in an explicit fashion?

Financing research from the trust fund Although as a researcher I can certainly understand and sympathize with a desire to avoid the vagaries of the annual appropriation process, and although I am sympathetic to the goal of increasing monies at NIH and AHCPR for research, as well as at HCFA, I do not favor Medicare's financing research (other than the HCFA supported research targeted at the Medicare program itself) for one general and two more specific reasons:

1. Most economists, including myself, believe that public goods should be financed through general revenues rather than through earmarked taxes unless those taxes are targeted narrowly to the beneficiaries of the program. This latter rationale, for example, would apply to the taxes on gasoline that are earmarked to finance highway trust funds; as a result, those using the highways will pay the costs of the highways in some rough approximation to their use of them.
   
   For broader public programs, however, earmarks work against sensible tradeoffs among public purposes. If we had, for example, earmarked a trust fund for national defense, it would presumably have required adjustment with the demise of the Soviet Union or with other changes in the external threat. But if there is to be an annual review of how much defense we should buy from the earmarked fund, that begins to look like an appropriations process. Similarly, scientific opportunities may appear that warrant additional funding through general revenues, as seems to be happening now. But if there were an annual review of scientific opportunities to see whether trust fund monies should be adjusted, that too would begin to look like an appropriations process - and one subject to annual vagaries.
   
   The foregoing argument comes from what my colleague David Cutler calls the "good government view;" public processes will rationally allocate limited resources among competing ends in a way to maximize public well being. Perhaps, however, those wishing to tap Medicare funds to support research have a "bad government view." That is, in the hurly-burly of the annual appropriations process, where all sorts of mischief may slip in during the small hours of the morning, Congress will not get research appropriations right - or more specifically they will get them too low. The size of campaign contributions might affect decisions, and senior legislators might procure a disproportionate share of public funds for projects in their districts or states. Indeed, legislators might even earmark research funds in ways that do not correspond with scientific merit. Therefore, something is needed to protect certain favored activities such as research.
   
   The problem, of course, is determining which activities should be so favored. Every advocacy group would, no doubt, want its program(s) in the favored category, and it is not obvious how to select. To be sure, certain categories of spending are somewhat removed from the appropriations process, for example interest on the debt and the entitlement programs. Even these, however, are not on automatic pilot, witness legislation to raise debt ceilings and, more to the point of this discussion, the legislated annual increase in Medicare hospital payments (subject to a default value), and the Balanced Budget Act's reduction in payments for graduate medical education. In short, even if there were a general consensus that medical research should be treated like interest on the debt -- and despite its current high reputation I do not think it yet ranks with interest on the debt -- there would still be an issue of setting the annual amount to be spent. Multi-year appropriations could address the problem of year-to-year variability in appropriations, but such a reform is getting well past the idea of whether Medicare funds should be used to support research.
   
   
2. In the specific case of using Medicare to support research, financing could come from the Part A trust fund or from Part B. Part B, however, is financed 75 percent from general revenues, with the remainder from beneficiary premiums. Thus, I assume the proposal to tap Medicare funds for medical research envisions using the Part A trust fund in a fashion similar to the Indirect Medical Education or Disproportionate Share payments.(3)
   
   The Part A trust fund, however, is already in danger of becoming the engine that couldn't. Indeed, many would say it has already achieved that status. Prior to the Balanced Budget Act of 1997 the Part A trust fund was scheduled to have a zero balance (or, as the press puts it, "go broke") by 2001; by saving $115 billion, the Congress in 1997 bought the fund another 7 years or so before the day of reckoning arrives. For details see the report of the Board of Trustees of the Federal Hospital Insurance Fund (1998). The low rate of increase in Medicare spending in 1998 will likely buy a few more years when CBO and OMB next update their estimates.
   
   Any projections of future fund balances necessarily rely on projections of how much medical costs will increase. As we all know, future projections can be notoriously inaccurate; recall the 1970s predictions of future energy costs. Although granting that future medical costs are highly uncertain, in my view the Trustees have in anything been overly optimistic in their projections (see Wilensky and Newhouse 1999 for a discussion). Moreover, there is much less uncertainty about the demographics; the Trustees project the ratio of workers to beneficiaries will fall by almost half over the next three decades. As a result, even if medical costs increases stay down at their present low rate for several more years, there is likely to be a problem keeping the trust fund solvent.(4) In short, given reasonable projections of medical care costs and of beneficiaries, the Part A Trust Fund already has more on its plate than it can digest without adding still more obligations.
   
   
3. The Part A trust fund is financed from payroll taxes of 2.9 percent on all earnings. The tax is nominally split between the employee and employer, but virtually all economists believe the employer's share is mostly shifted to wages; i.e., cash wages are lower by most of the amount of the employer payment (Fuchs, Krueger, and Poterba, 1998). As a result, who pays for medical research is different if payments are made from the Part A trust fund rather than general revenues. Specifically, whereas general revenue financing is progressive, financing from the trust fund is proportional to earned income. Although who should pay how much for medical research is clearly a value judgment, a broader based tax that accounts for non-earned income would seem preferable to payroll tax financing.

Costs of Clinical Research (5)
The Medicare program is by statute restricted to paying for services that are reasonable and necessary for treating a clinical condition. The Health Care Financing Administration interprets this to mean that Medicare should cover only services that are "…safe and effective, not experimental or investigational…" (Health Care Financing Administration, 1989).(6)

In the case of new services, a decision must be made as to whether the existing evidence warrants a judgment that the treatment is safe and effective. There are few bright lines here, although in the cases of drugs and devices the FDA must have approved the product or cleared it for marketing or Medicare will not cover the service.(7)

When services are part of clinical trials, there is by definition uncertainty about the benefits of the treatment and so Medicare often does not cover the cost of treatment in the experimental group. Typically a drug manufacturer will cover the cost if a drug is involved, expecting to recover the costs from profits made during the time the drug is on patent.

But of course there are many instances of surgical and medical procedures for which there is no patented product, and hence no manufacturer to pay the costs of treatment in the experimental arm of a clinical trial. Because the trial is generating common knowledge, a public good argument suggests that public funds should pay. It seems to me, however, that these funds could be Medicare funds rather than research funds.

Specifically, I believe Medicare should pay an amount for the experimental group in an approved clinical trial in which no patentable product was foreseen that would equal what it paid on average in the control group. The argument for Medicare support is that its failure to pay raises the price of clinical research and hence causes less such research to be done. In those cases in which the research is not done, not only do we lose the knowledge that would be generated by the trial, but Medicare presumably pays the cost of the conventional treatment in any event. Hence, Medicare is no worse off under this new rule than it would be if the trial were not undertaken but meanwhile knowledge has been gained from undertaking the trial.(8)

I have not made a cost estimate for this proposal, and costs would clearly depend upon how many trials there were. If payment was restricted to NIH financed trials, however, it would seem as if the costs could not be very large relative to Medicare program costs.

Summary
I have made two arguments:

1. The Medicare trust fund should not generally pay for medical research; as an earmarked trust fund, it should be reserved for its original purposes of financing the delivery of services to its beneficiaries, albeit that principle has already been breached in the case of graduate medical education and disproportionate share payments. Research financed by public funds should be financed by general revenue. Moreover, looking forward a few years the trust fund will be running large deficits; this would not seem to be the time to speak of additional taps on the trust fund.
   
   
2. Medicare should pay an amount equal to what it otherwise would have paid for patients in the experimental arm of an approved trial where no patentable product is expected to result.

(1) Prepared for "How to Fund Science: The Future of Medical Research," AAAS Workshop at Wye River Conference Center, February 14-16, 1999.

(2) Elementary economics would suggest that subsidies to hospitals for additional residents would induce increases in the number of residents or their salaries or both, there has in fact been a marked increase in numbers (Medicare Payment Advisory Commission, July 1998, p. 148). The Balanced Budget Act capped the number of residents eligible for payment at 1996 levels.

(3) There is a small Part B trust fund but it is only for covering working balances.

(4) In fact, the trustees project the rate of medical cost increase per beneficiary falling to approximately the rate of growth in the economy. Even with that assumption, which I regard as highly optimistic, the Medicare program (Parts A and B) is projected to grow from its current 2.65 percent of GDP to 5.85 percent in 2030, and those costs are being shared among a relatively much smaller workforce.

(5) This section draws heavily on Physician Payment Review Commissionk, 1995, chapter 6.

(6) It is not the case, however, that simply including a service in a research study means lack of coverage, because effective services may be used in studies.

(7) The drugs in question are mostly inpatient, since Medicare covers few outpatient drugs.

(8) In those cases in which the research would have been done in any event, the issue is which public source pays, payroll taxes (insofar as there are Part A services) or general revenues. The consequences for economic efficiency, however, seem second order compared with the discouragement to research that the present non-neutral treatment of costs in the experimental and control group poses.

References
Board of Trustees of the Federal Hospital Insurance Trust Fund, "1998 Annual Report," available on the Internet at www.hcfa.gov/pubforms/tr/default.htm

Fuchs, Victor R., Alan B. Krueger, and James M. Poterba, "Economists' Views about Parameters: Survey Results in Labor and Public Economics," Journal of Economic Literature, September 1998, 36(3):1387-1425.

Health Care Financing Administration, "Criteria and Procedures for Making Medical Coverage Decisions that Relate to Health Care Technology," Federal Register, January 30, 1989, 54(18):4302-4318.

Kominski, Gerald, and Stephen H. Long, "Medicare's Disproportionate Share Adjustment and the Cost of Low-Income Patients," Journal of Health Economics, April 1997, 16(2):177-190.

Medicare Payment Advisory Commission, Report to the Congress on Medicare Policy, volume 1, chapter 6 and volume 2, chapter 5; March 1998.

Medicare Payment Advisory Commission, Chartbook, July 1998.

Physician Payment Review Commission, 1995 Annual Report to Congress; Washington: The Commission, 1995.

Wilensky, Gail R., and Joseph P. Newhouse, "Medicare: What's Right? What's Wrong? What's Next?" Health Affairs, January/February 1999, 18(1):92-106.

© 1999 American Association for the Advancement of Science