In an interview published online at Health Affairs [2-week free access], David Eddy, founder and medical director of Archimedes Inc. in Aspen, Colorado, discusses evidence-based medicine (EBM) with Sean Tunis, founder and director of the Center for Medical Technology Policy in San Francisco. Archimedes was founded to improve the quality and efficiency of health care by using advanced mathematics and computing methods to build realistic simulation models of physiology, diseases, and health care systems. Eddy has made seminal contributions to evidence-based medicine, coining the term “evidence based” and applying it to guidelines, coverage policies, and performance measures. Tunis was instrumental in reforming the national coverage process at the Centers for Medicare and Medicaid Services (CMS) to explicitly incorporate principles of EBM and was extensively involved in decision making about coverage for ICDs (implantable cardioverter-defibrillators) as the chief medical officer at the CMS.
In the Health Affairs interview, Tunis explains how cost considerations might have made their way into Medicare’s reconsideration of its ICD coverage policy, even though the program does not formally consider costs in such determinations and has stated that it does not do so. In light of a study showing that many additional beneficiaries might benefit from ICDs, the CMS expanded coverage, but not initially by as much as a joint American College of Cardiology/American Heart Association (ACC/AHA) committee had recommended. Here are a few excerpts from the Health Affairs interview.
The ICD Case
Sean Tunis: Just to be clear and perhaps to respond to a lot of guessing and assumptions about the role of economic impact in Medicare’s policies on ICDs, the potential impact on Medicare spending or cost-effectiveness of ICDs was really not discussed much within the agency or within the Department of Health and Human Services when this decision was made. At the same time, it was well understood by me and others at CMS that ICDs were expensive and that there were a lot of additional people who might be eligible for an ICD, and that added up to a large amount of money. So what does that cause us to do differently than for decisions with less potential financial impact? It causes us to look extremely carefully at data on safety and effectiveness. You might think of this as an upside-down or inside-out variation of a cost-effectiveness analysis in which the evidence threshold for coverage is implicitly adjusted based on a qualitative judgment about the economic impact of the decision.
But there is no conscious or explicit policy at CMS to take this approach. Career staffers in the Medicare program aren’t given the assignment of trying to protect the Medicare Trust Fund or to find ways to improve value for money, and it certainly doesn’t affect their take-home pay whether Medicare spending goes up or down. In fact, explicit statements have been repeatedly made by Medicare that cost is not factor in coverage decision making. But my guess is that for anyone who works for a large payer in a policy environment that is increasingly panicked about the cost of health care, it’s easy to imagine how economic impacts could still have subtle and perhaps even unconscious effects on some of the scientific and value judgments that we have been talking about, whether or not these folks are told to ignore costs. Most of the people I know who work for public or private payers share some common views on the need to apply rigorous evidence standards to coverage decisions, and while it’s easiest and safest to justify this as a mechanism to protect patients from harmful or useless technologies, I suspect that the situation is more complex than that.
And I think that all of this is consistent with what you have said about how costs might differentially influence the judgments of different stakeholders. My view is that Medicare decisionmakers didn’t consider costs any more or less than the members of the ACC/AHA guideline committee did, but I think that our implicit level of concern about costs was different and perhaps influenced the analysis of evidence and conclusions drawn by each organization.
David Eddy: I agree with you. The phenomenon you describe of upside-down or inside-out cost-effectiveness analysis occurs in the private sector as well. For example, an insurer or managed care organization might say that their coverage, guidelines, or performance measures are based solely on evidence of effectiveness and are not affected by costs. But at the same time, they can ratchet up on the requirements for the evidence.
The Personal Side Of Cost-Based Decisions
Tunis: I know you’ve given a lot of thought to this topic and written about it. And I’ve thought a fair amount about the practical barriers to having an honest, open, explicit conversation about costs and value, particularly in the context of coverage and payment for new technologies. As I already mentioned, the first dilemma is that nobody in a high-profile public policy job wants to have a conversation about rationing–at least not when that word is defined as potentially restricting access to specific technologies for which the benefits are too small to justify the costs.
When you talk about restricting access to any sort of health care service based on economic considerations, it looks, sounds, and feels like rationing, no matter what other more acceptable terms you might try to use: value, efficiency, high performance, etc. When you weigh costs against benefits in the same decision frame in health care, it could reasonably be called rationing, at least in a technical sense. And while I completely agree with your view that the failure to take this on in health care leads to a huge amount of waste and inefficiency, I get hung up like everyone else as soon as I try to translate that into specific decisions that affect individuals. I, too, struggle to reconcile the need for system efficiency and cost-effectiveness with the “what-if-it-were-your-mother” problem. Imagine a particularly expensive drug for cancer or a costly cardiac device for which the cost-effectiveness models clearly demonstrate that the benefits are small and the costs are really huge. It’s still hard to imagine saying to Mrs. Jones, “There is a reasonable chance that this would help your mom, but we just think it’s too expensive, and there are better ways we could use the money.”
Eddy: That is a difficult issue. But there are several ways to address it. First, we need to recognize that we draw these types of lines all the time in other areas of public policy. There are some people here in Washington, D.C., who have no homes. In terms of spending money to improve someone’s length and quality of life, we would be much better off to give those people a warm place to sleep at night, a good meal, and at least minimal health care. That would be a great way to spend our money to improve quality-adjusted life-years, to put this in the metric that we use in health care. And yet we do not do that. There are lots of other examples of opportunities to improve the quality and length of people’s lives that we pass up because we just don’t think it’s the role of society.
But perhaps more important, saying that we should take costs into account in our medical decisions does not necessarily mean that we cannot pay for lifesaving treatments for your mother. . . . We can decide how we want to spend our money. Elsewhere I have advocated the simple step of asking people how much they value certain types of medical activities. We could, if we wanted to, spend $100,000 on the last year of everyone’s life and make it up by not spending tens of thousands of dollars for some test or procedure that will have a minimal effect on some symptoms that will eventually disappear anyway. What is wrong today is not that we are spending money but that we are spending it with no deliberate analysis, no sensible trade-offs, no controls, in a very wasteful way, and we are not getting the quality we want in return. By refusing to consider costs, we have put ourselves in the worst of both worlds.
Tunis: I can think of a couple of other similar examples. We generally don’t define as rationing things that arbitrarily limit access to potentially effective care–such as applying financial criteria for Medicaid eligibility or setting budget-driven dollar thresholds for the doughnut hole in the Medicare drug benefit. In these examples, no one is trying to claim that providing additional services would not be worth the money. The limitations are not designed to be clinically or economically sensible, and although they might be suboptimal from the perspective of getting maximum health for a given level of spending, they are much simpler to defend politically.
Eddy: Absolutely. It is what I call rationing by meat ax, and it is a good example of the harm caused by our unwillingness to consider costs explicitly. We do ration in health care, but we do so by lopping off people, not by lopping off technologies and services that have low value. What we need to do is ration by scalpel; we need to do a much finer and more precise job of squeezing out the most benefit we can get for the dollar.
Acceptance Of EBM
Tunis: I’m not quite as certain that we are as far along as we need to be on the acceptability of evidence in health care decision making. It seems to me that we are starting to come up the steep part of the acceptance curve, but we still have a long way to go, and there is still a lot of resistance.
Eddy: I agree with you that the application of EBM is not yet firmly in place. But I do believe that the principles have been accepted. Every organization that designs a guideline, coverage decision, or performance measure is talking the talk and claiming that their product is “evidence based.” No one is willing to go back to the old methods and say, “Our decision was not based on evidence; it was based on the opinions of experts.” So in that sense, I am encouraged. But I do agree that the stringency with which EBM is applied varies considerably.
Tunis: Fair enough–perhaps your “glass-half-full” perspective is a better way of recognizing the important progress that has been made. So to continue on that optimistic note, I’m wondering whether you believe that measuring and reporting quality, along with P4P, will save money. My sense is that when you press P4P advocates on that point, they pretty quickly acknowledge that they don’t expect any reduction in spending trends but that we will get more health benefit for the money that we do spend, and that would be a good thing. And yes, it is a good thing, but then it should be clearly understood that it’s not relevant to the problem of unsustainable trends in health care spending. If it’s not going to affect spending trends, then we need something else that will address spending trends, or we need to be able to explain why all the economists that are freaking out over the depletion of the Medicare Trust Fund ought to just relax. Of course, in the meantime, the concerns about spending are prompting changes in benefit structure designed to reduce spending by increasing patients’ financial exposure to the cost of their care. While again I can see the logic of this, it also has the look and feel of another form of rationing, for which the primary appeal is that it avoids the problem of having to make policy judgments about the value of individual technologies and services.
Eddy: On the issue of whether P4P is intended to save money, I daresay that if we took the P4P advocates, gave them sodium pentothal, and asked, “Do you care whether or not this saves money?” the great majority would say “Yes, we want it to save money.” On the issue of cost sharing by patients and rationing, I agree that it is a form of rationing. It has some nice properties in that it lets the patient make the trade-off between benefit and cost. . . . But it is flawed in that the patient being asked to make that choice has little or no information about the magnitude of the benefits or when they will occur. It’s another example of our attempt to control costs without considering costs.
If we really want tools like P4P to save money, then we need to add at least two more things to it. First, we have to make certain that the performance measures and P4P formulas are based on value, not just evidence of benefit. Second, we need to create measures and incentives that turn low-value things off, not just turn high-value things on. That is, we need to add some pay-for-nonperformance to the pay-for-performance, if you will.
Tunis: Which brings me back to one of my general “lessons learned,” which is that to deal with the problem of health care costs, we’ll need to be ready to make some politically difficult decisions that take money out of somebody’s pocket. And I think we need to keep that in mind whether we are talking about EBM, or P4P, or chronic care management, or whatever. They are all good, important, worthwhile things to do. And folks around the country have made tremendous strides in all of these areas, so there is much to celebrate. But we can’t delude ourselves into thinking that they offer a way to deal with the economic pressures in health care without having to make tough choices.