• Adding “value” to Medicare payments

    Continuing his lessons on Medicare payments (worth reading in full) Uwe Reinhardt challenges,

    While no one likes the cost-based Medicare fee schedule, its critics should be challenged to suggest a workable alternative based on “value to patients.” For starters, the critics should explain concretely how they would define and measure value in this context, in monetary terms, keeping in mind the administrative cost of any such system.

    I know what Reinhardt is daring critics to produce, and I agree that it’s a tall order. But I think there’s a first step that admittedly only very roughly gets at “value” in the sense he means it but is worth taking anyway.

    If patients cannot be sufficiently informed to assess and communicate value and/or it would be unwieldy or impractical to design a system to harness just that, then we could at least imagine a system that approximates value in a way that a fully informed patient who is paying his own way might. A fully informed patient would want to know how to rationally choose between therapy A and therapy B based on expected quality and quantity of life each would deliver. In other words, the patient would want to know the comparative effectiveness of the two therapies.

    This is knowable, with research. Medicare could incorporate comparative effectiveness findings into what it pays. David Leonhardt and Peter Orszag described how it could work. In brief, the Independent Payment Advisory Board could propose that the program only pay the lowest price among equivalent therapies and/or only pay for the most cost-effective (in some sense). Beneficiaries wanting more expensive and/or less effective therapies would have to fund the marginal cost themselves. Politics aside, it is not pie-in-the-sky. Why should taxpayers pay more than they have to? Why should they pay for what doesn’t work? Why should they pay whatever physicians have determined their costs to be? Why can’t we harness research to improve efficiency?

    We can get more value out of Medicare. Changing the payment system is precisely how to do so.

    • Of course mixing science and politics goes both ways – the old regulatory capture story. If comparative effectiveness research determines how much hospitals get paid, and how much insurance companies have to pay, you’re going to see lots of interested parties funding skewed research, and lots of research politicking trying to establish a false sense of “conventional wisdom.”

      Medicine has traditionally been pretty bad at discarding outdated theories, because of the idea that research that tests the truth of conventional wisdom is unethical because it’s withholding the standard of care from patients. (Examples: Heliobacter pylori, low-fat diets as a treatment for obesity). If we throw financial incentives into this already very conservative mix of professional ethics and reputational pride, we’re going to see a lot more bad science come out of it.

    • Off topic, but re: Reinhardt’s comments on AMA’s ownership of CPT codes: maybe the federal government should do a property “takings” over the ownership of the CPT codes. Pay fair market value, and free the codes. The benefits to humanity might be great. At worst it shifts the cost of the rents from across the health care industry to a single payor, making society not really any worse off.

    • There should be more clarity in this debate about why reimbursing at cost is a bad idea, and why value should be included. In most markets, it is efficient to price at cost rather than value. Water and bread are very valuable to humans, but they are priced cheaply because they are cheap to produce and competition lowers prices to costs. No one complains about the prices of bread and water not reflecting their high value.

      If doctors actually are looking out for the best clinical interests of patients, then reimbursing at marginal costs will lead to efficient and effective provision of care. Assume the doctors make the decisions about what care is received. If physician utility equals patient benefit minus cost of treatment, then pricing at marginal costs will lead physicians to provide only services whose clinical benefit exceeds cost.

      So, any argument for value-based reimbursement should be accompanied by an argument about why the above model fails (e.g. why doctors don’t care about the best interests of patients). Doctors who say they are acting in the patients’ best interests but are upset about cost-based reimbursement are confused, I think.

      • @Aaron – It’s about what the product is whose cost you’re paying for. Here, the actual product is health, or the remedy of a specific condition. There may be N ways to produce that “product.” Medicare pays for each of those N ways at the (administratively determined) cost of each.

        What if way 1 is just as effective as the other N-1 ways and costs half as much? In a market, the other N-1 guys would be dead. Not in Medicare. They live, and continue to cost us. Why? Because there is nothing in the price setting mechanism that reflects underlying value of the thing we want, health!

        I do not think we can throw this open to the market due to well-known market failures in health. But we can improve the price-setting mechanism in the direction we want, to pay for efficient and effective care, not whatever cost some body (and political process) says we should pay for every conceivable remedy whether effective or not.

    • Austin’s (and David Leonhardt’s and Peter Orszag’s) idea to use comparative effectiveness analysis and reference pricing as a fundation for payment policy makes perfect common sense — in Canada, In Europe, in Asia, in Australia, in Africa and in Oceania. But in the American context it may be viewed as subversive.

      Austin acknowledges it with his aside “Politics aside, this is not pie in the sky.” As a perennialand ever more bewildered student of America, I have learned that, in America, common sense can be very much pie in the sky, especially in health policy.

      As I recall it, the ACA expressly forbids the use of comparative effectiveness research (CER) for coverage decisions. It was a victory for the supply side of the health care system, which is a major shareholder of the US Congress. So the question is whether these major shareholders would allow Congress to use CER for payment policy. And the question is whether the Progressives would accept reference pricing for Medicare.

      So, politics aside, I give you a good grade for owning up to my challenge, Austin, with this advice: when you write stuff like that, stay out of reach of Ms. Sarah Palin’s hunting rifle. She might mistake you for a moose.

      • @Uwe Reinhardt – I’ve been hoisted on my own petard. I’m usually quick to appeal to political feasibility or lack thereof (I devoted a post to that very topic). Here, as you note, I put politics aside in a circumstance where that’s just not realistic. And yet, that’s true of nearly everything about Medicare or health reform. Our only hope is that the range of politically viable options expands to include things that could help. Someday that must happen, but it could be a while. As you recently wrote, we will “experience a few more dire teaching moment before [we] can debate cost-containment responsibly.”

    • But in the meantime we wonks should get the ammunition ready, and that’s what you did, Austin, and that is good.

      My sense is that a combination of CER plus reference pricing is how eventually we’ll go in this country, should sanity make a return visit to our shores. Germany controls drug spending that way. So we have well working models. Drug companies strongly object to it though.


      • @Uwe Reinhardt – “should sanity make a return visit to our shores.” Perhaps it was once here, but not in my lifetime. My reading of Starr’s book suggests that in health care it may never have been, but perhaps there’s another plausible interpretation. 🙂

    • Having worked for many years on the development of new drugs, all I can say is that comparative effectiveness research is a) incredibly difficult and b) extremely expensive. Design of trials to give unambiguous results is extremely challenging, and executing those trials ranges from difficult to impossible.

      Further, by design such trials only tell you average effectiveness, and can’t recognize real patient to patient variation. This is a real phenomenon in medicine. We have a gazillion different hypertension treatments, but typically a new patient needs to try several different drugs and doses before finding an optimum combination of effectiveness and side effects. I worked for a time on NSAIDS, and it is very clear that choosing the best drug for a given patient requires trial and error.

      Because this variation is real, I am reluctant to say “We will only reimburse you for the cost of diuretics and not an ACE inhibitor, even though the diuretics don´t work for you.” This example is becoming dated, because so many antihypertensives are off patent, but you get the point.

    • Very insightful comments from all of you. I think the bigger challenge is the treatment vs. health theories of value. The treatment theory of value is well illustrated by the nature of this discussion on CER as a basis for payment, we want the most effective and least costly treatments, but this presupposes the treatment where the choice between the treatments themselves may be more important (e.g. medical management vs. surgery for coronary artery disease). The “simplicity” problem in building a value system is really the problem. So let’s say we could figure out which expenditures are subject to the treatment theory of value (and maybe we could), others are really about the health theory of value. Most emphatically to me in that camp is the interactions of behavioral health and medical health. We have SF-36 measures of Physical and Mental Components that could be the basis for balancing physical and mental health, but that’s currently the area where I would argue we are performing the worst. How do we treat the health of the whole person?

      I know this is way too dense for a blog post, but that’s my reaction to this exchange….

    • Austin,

      I did not mean to imply that we shouldn´t even try to get good comparative data. I certainly think that this is an area of science, as well as policy, that should be pursued — including where possible subpopulation data that can help with the heterogeneity issue.

      However, I think that policy makers greatly underestimate the difficulty, time, and cost of getting actionable data. Pursuing this will help, but I am skeptical that we will make enough progress, fast enough, to really solve the problem.

      • @David – I am skeptical that we’ll do anything very fast. I think the cost problem will remain with us for at least two or three decades.There may possibly be a pause that will appear as if we’ve solved the problem if HDHPs really take off.

        If I’m right, and there really will be a few decades of time before the country is ready to get serious about costs, that’s plenty of time to do some very good research.

        Moreover, it isn’t just the research that would help, it’s the attitude that we should use evidence to better allocate spending. That would be a sea change. It requires a political atmosphere that doesn’t yet exist. If the politics permits CER to be put to use then imagine what else it would permit.