Where's the evidence?

In relation to the single-payer discussion, Tom asks where is the evidence that Medicare overpays for services.

Actually I was coming from a different angle that does not have much to do with benefit design or program operation. It has everything to do with the way the medical system actually works. My experience has been that government agencies assume that uniform policies produce uniform results and they are ill prepared and have no incentive to deal with variation in results, especially when that variation itself is the result of a wide variety of factors that require careful localized analysis, patient targeted interventions and are the result of factors that are not within governmental control.

Or put another way, I was focused on use of medical services. That's where the action is. Single-payer advocates focus on administrative overhead. That's simple and easy to understand. Regulatory types tend to focus on prices. That's simple at first blush and it can work in the short term, but once the regulatory process is captured by those regulated, it's just a lot of politics that produces no financial benefit to anyone but lobbyists and it may produce higher prices. More subtle, more complex, and harder to explain are the bell curves of medical service use. At the high end, they produce nothing but expense and the potential for patient harm. At the low end, they indicate inadequate care. In the equation of total costs ((service use X price) + overhead) = total cost, the longest lever is service use. And no, this does not lead me to having clerks deny care over the phone.

But since you asked Tom, here's a long one and since you asked you have to promise to read it all and think about it and its relationship to what we've been discussing.

Here goes ...

When Medicare and Medicaid were established in 1965, based on what most everyone involved thought they knew about mainstream, privately insured, medicine, they took for granted three key assumptions. These assumptions were so deeply embedded, that they weren't even discussed. This was just the way the world was:

1. That medical care was uniformly provided
2. That when provided, with the possible exception of some oddities, medical care was consistently of good quality. And good quality produced uniformly good results. A related, implicit assumption was more is better.
3. That, where quality was not good, it would be extremely bad and could be weeded out through a regulatory survey process and that aside from that there was not much that could be done.

All three assumptions have turned out to be incorrect with significant implications for for patient care, clinical quality, and spending. What we've learned in the 40 years since is profound, but we still have a very long ways to go in taking advantage of it. This is relevant to the single-payer discussion because while a single-payer system might be centralized, medical care is not. I'll discuss that tie-in further below.

Medical care is not uniformly provided

Where's the evidence?

Read anything by John Wennberg, MD. (If you do a search at the New England Journal alone, you'll find well over 100 citations.) Here's a recent one in the British Medical Journal that focuses on end-of-life care. In the late 1970s, Wennberg began to apply epidemiological methods to medical care. The first study I read was in Scientific American in 1979 as I recall, but that's not online. What Wennberg and his colleagues found was that there is substantial variation in medical care from one community to the next. They have also found what they characterize as a "practice pattern" effect, that physicians in one community develop a rough consensus on the best way to handle particular types of cases, but that their consensus differs from the next community and will produce care patterns that are stunningly different from those in other communities, even within the same state or region.

For a quick introduction, start with the Dartmouth Medical Atlas, published by the Center for Clinical Evaluative Studies, which is one of the offshoots of Wennberg's work.

These variations also exist within similar populations and in some cases very homogeneous populations. For example, a colleague studied variation in hospitalization patterns among Nebraska Blue Cross subscribers for conditions that can be managed on an outpatient basis (diabetes, asthma, CHF, COPD, etc.) As another example, in the late 1980s and early 1990s, my own work focused on variation in hospitalization patterns of Medicaid clients in New York. Here's a quick and dirty update that I did in 2003. The earlier work was much more rigorous and I'll be happy to send a copy to anyone who asks.

Wennberg and his followers also learned that the variation itself varies from one condition to another. For example, appendectomies are always performed in hospitals and there is no known relationship of appendicitis to age, gender, race, socio-economic status. So there is very little variation and, from what I've seen, it's just random. Other conditions, however, are quite different. For just a tiny sample take a look at these:

• Variations in hospitalizations of children in three urban communities, NEJM, May 4, 1989.
• Regional variation in medical care, NEJM, August 31, 1995 where Detsky noted variation in clinical outcomes such as mortality after cardiac surgery. "Once again we have learned that there is substantial variation among regions, hospitals, and even individual physicians."
• Among New York's Medicaid clients the most variable conditions as causes of hospitalization were psychiatric and substance abuse, blood related conditions, and respiratory conditions.

There is much, much more.

This leads to "what difference to these differences make?"

Medical care does not provide uniform results and spending more does not necessarily produce better results

Where's the evidence?

Start in the Annals of Internal Medicine with two papers by Elliott S. Fisher, David E. Wennberg, Thérèse A. Stukel, Daniel J. Gottlieb, F. L. Lucas, and Étoile L. Pinder. The first is "The Implications of Regional Variations in Medicare Spending. Part 1: The Content, Quality, and Accessibility of Care" where they conclude that Medicare patients who live in areas that spend more on Medicare do not necessarily get better care than those who live in regions that spend less. The second, "The Implications of Regional Variations in Medicare Spending. Part 2: Health Outcomes and Satisfaction with Care," where they conclude that Medicare enrollees in higher spending regions receive more care than those in lower-spending regions but do not have better health outcomes or satisfaction with care.

Or even more recently, take a look at Baicker and Chandra who "find that states with higher Medicare spending have lower-quality care. This negative relationship may be driven by the use of intensive, costly care that crowds out the use of more effective care."

Fisher wrote an op-ed for the New York Times, More Medicine is Not Better Medicine (fee required for full article) December 1, 2003. He pointed out that additional service provided in higher spending regions are largely discretionary. He estimated that as much as 30 percent of Medicare's service spending was unnecessary. For you New Yorker's he also pointed to spending in NYC being double a community in the Pacific Northwest with no additional benefit. Here's Fisher's slide show, "Is More Better?" (Quicktime required.)

You can also look at the numerous studies that have come out of RAND on excessive and inadequate care. For example, take a look at Assessing the Appropriateness of Care, How Much is Too Much?.

If care is not of uniformly high quality or does not produce uniformly good results, there's nothing that can be done about it

Where's the evidence?

Start here at the Institute for Healthcare Improvement, founded by Don Berwick. In January of 1989, Berwick wrote "Continuous improvement as an ideal in health care," the seminal paper in the New England Journal of Medicine on the use of industrial quality control methods, statistical process control and other similar methods in medical care. (That paper is not online, but it's worth tracking down if you're interested in the history.) The Institute was founded two years later (with a grant from a health plan by the way) and it has expanded each year. Their annual meetings now get thousands attending.

IHI is just an organizational manifestation of many efforts to measure performance, understand clinical processes, and improve them. You can go into many hospitals today and see graphs of their performance on a wide array of issues hanging near the offices of department chairs and sometimes even in public places. That simply wasn't true 20 years ago. And in many healthcare organizations, it's still not true.

What's necessary to deal with all this variation?

Here's a starter list:

• Data to identify community-by-community, condition-by-condition, and service-by-service where the high and low rates are and, where at all possible, outcome data
• Peer consulting and other knowledge transfer
• Time and support for dealing with it locally
• The organizational infrastructure to support local efforts
• Financial and organizational incentives for local medical organizations and professionals to participate
• Organizational frameworks for evaluating outcomes when, as it increasingly does, care crosses organizational or practice boundaries.
• Patience

Comparative data must be collected centrally or at least in accordance with a standardized design. That would be simpler with a single-payer plan, but certainly feasible in a multi-option universal system.

Beyond that, it seems to me that most of the advantages are with alternatives to a single centralized system. There's economic incentive and local involvement on scales that don't exist for Medicare, much less the entire US population.

And what does this have to do with single-payer?

The most politically compelling rationale for single-payer is that it would instantly provide a large savings in administrative expense. It quite likely would. But where would we go from there? Nothing I have personally seen or read gives me confidence that a large centralized system would be capable of effectively dealing with variations in care. In my experience with New York's Medicaid program and with the exception of some legislators and local officials, there has been nothing but resistance from government officials. Perhaps that's biased my view of the potential, but let's remember that New York is one of the states that is most intrusive into medical care and certainly, given the cost of its Medicaid program, it would seem that there should be an adequate incentive. For government agencies this requires too much subtlety and long-term commitment.

What's required is a system that includes a significant incentive such as economic competition for example. And it requires lots of people working on it in different organizations with different perspectives. That's where innovation comes from. Innovation slows when you have to ask for governmental permission.

It may also require the ability to be selective in ways that government agencies just can't be. I'm reminded of the multi-decade stalemate over Medicaid physician fees in New York. The state had stopped updating the fees in the same manner that it updated institutional rates. So here's what happened. First, physicians dropped out of the Medicaid program at increasingly rapid rates until there only a few left and many of them were practicing in the so-called Medicaid mills. Second, the patients recognizing reality rather than what the law said about what Medicaid covered moved to emergency rooms and hospital outpatient departments for care because there was no alternative. Third, the hospitals paid political attention to the regulators and the physicians largely did not.  When the state finally recognized the imbalance it had created, it was unable to fix it because to raise physician fees would have been to reward the Medicaid "mills." The log jam was only broken with the large scale use of Medicaid managed care which brought physicians back into contact with Medicaid clients by way of intermediary health plans. As a result primary care access has significantly improved for Medicaid clients. Docs may hate the health plans, but at least there's a working relationship. That was long past for Medicaid.

You need an ongoing, permanent economic incentive, local working relationships (which admittedly some of the national health plans may be fairly weak on) and the ability to work with medical professionals targeting local patterns everywhere. I just don't see the Federal government doing that. Certainly if it attempted to, much of the administrative savings would be lost.

If, as Fisher argues, 30 percent of Medicare expenditures for service today produce no medical benefit, then it is likely to become worse as medicine becomes more complex and involves more with each patient. Ignoring that for the cheap win of killing off the health insurers would leave us with no where to go and relying on the Federal government to find a new direction.

 

 

Comments

[Abby]

I am in no way an expert on these issues, but I did notice that you had much better data on the Medicare and Medicaid populations than you do for private plans.

I think that I would be better served by a good single payer system. I have a chronic condition that was mis-diagnosed when I was in college. In college I was restricted to using a university health service (similar to student health service--only larger). I was not happy with the quality of care offered.

I would have liked to be able to go to proper specialists, but I couldn't. I make sure to follow the medical literature on my condition, and I'd like to be able to fire any doctor who doesn't. With single payer I would have a greater choice of doctors. (The only doctor I've met that I thought was really competent didn't take any insurance. So, I prefer a PPO model where I can get 70% coverage to an HMO where I'm stuck with the crappy Kaiser doctors.)

If I got sick outside of my regional coverage area, I wouldn't have to worry about whether it was covered. And I wouldn't have to worry about doctors leaving my plan or a particular plan being dropped by my employer.

[TomH]

OK, I've read John's post. Here are my thoughts.

I think John is exploring a new angle of the universal care debate, one that I really haven't seen debated before. Usually, when an analysis drills down to the delivery level, it's to examine how a particular model will save money. That's critically important, but frustrating. The rejection of the Clinton plan, in part, because of the canard that it would have limited consumer choice, marked the beginning of a consumer revolution that rejected all limits on choice, thus virtually all cost-control mechanisms. So discussions of how we'll reshape the delivery system under a particular universal health care system to control costs probably miss the point, which is that the American public may not allow costs to be controlled under any model. (which is why I think it makes sense to target administration and marketing costs, since those may be the only excess costs truly on the table.)

So perhaps it makes more sense to create a universal health care model that does what John suggests, which is to 1) reduce overuse of health care services; 2) reduce misuse; 3) increase utilization of underused services. The public might respond less suspiciously to a proposal that made this a high priority. And it would ultimately save money. The overused services are very expensive. The underused services are cheap and needed to control chronic conditions, which are much more expensive at the back end.

I'm intrigued by the questions John raises, but not so sure about his answers. In part, that's because I don't know what plan he supports, just that it involves competition between health plans. But if we look at our choice as an election, in which we're choosing between some form of managed competition and some form of single-payer based on which would do a better job of improving quality of care, it's not at all clear who I should vote for.

The starting point is our current system, which is highly decentralized - that's what created the preconditions for geographical disparities in the first place. Have private insurers incorporated Wennberg's findings into their rates and benefit structures? Have they demanded that providers create shared decision-making structures so that patients can exercise more control over the decision whether to undergo surgery or to stick with a more conservative plan of treatment? They have not. In fact, it's not clear what private insurers have done to improve quality of care in the United States. That doesn't mean that health plans are bad and should be abolished. Hey, maybe it's the employers' fault for not demanding high-quality care. But it means that John is pinning a lot of hopes on a market sector that has no track record for doing what he wants. Well, except for Kaiser Permanente. If John's proposal involves forcing all insurers to turn into integrated staff-model group practices, count me in.

One last thought on Jack Wennberg. Like John (and Governor Pataki), he's concerned about "finding mechanisms to clear regional markets of excess capacity." Here's his plan: "If Medicare was willing and able to take steps to select providers on the basis of quality and efficiency - and other payers were willing to play by similar rules - this would serve as a life or death wakeup call to the provider community, one which would presumably result in accelerated change throughout U.S. health care markets."

Got that? Wennberg wants the Federal government to decisively intervene in our high-functioning market economy to solve an intractable quality of care problem. And he wants this intervention to be carried out through a single-payer system, which is to say, Medicare.