From 3M Health Information Systems
Podcast Episode Transcript: Good intentions and unintended consequences: A look at healthcare policy
Gordon Moore: Welcome to the 3M Inside Angle podcast. This is Gordon Moore, your host, and today I am speaking with Harold D. Miller. He’s the president and CEO of the Center for Healthcare Quality and Payment Reform, which is a national policy center that facilitates improvements in health care payment and delivery systems. He’s also an adjunct professor of public policy and management at Carnegie Mellon University. Welcome, Harold Miller.
Harold Miller: Well, thank you, Gordon. It’s nice to be here.
Gordon: You think, write and talk a lot about payment and policy, and that’s interesting to me because it’s one of the deep policies that tends to drive a lot of behavior, a lot of work in health care. And the intent may be good in some of the stuff, but sometimes the follow-through may not be good. And I was intrigued by an email blast you had sent that talked about saving primary care before it’s too late. I wanted to ask you about that, what you think the problems are, why it is we need to save primary care, and how you think we ought to go about doing that.
Harold: Well, I think there’s been recognition for quite a while now that primary care is really core to the health care system, that you have to support it so that people have someone who can diagnose things whenever something appears, help the patient be able to make decisions about whether they may need to go somewhere else for care or what care they may need to get, et cetera.
One of the challenges, though, is that there has been fairly weak support for primary care. There’s lots of recognition about how important it is, and what seemed to happen for a while was people were starting to load up the expectations onto primary care about what it could do without actually providing the kinds of resources that primary care would need to be able to carry that out. It’s a little bit like the, “We’re going to give you a promotion by giving you a better title but no better pay and more responsibility for that. “ And so there were all these expectations about primary care being able to manage total cost of care, improve quality, et cetera, but without the resources to be able to do that.
And it’s pretty clear, when you talk to primary care practices, that they struggle just to be able to do the basic things they have been doing with the kind of payments that they get from Medicare and most commercial insurance plans as well as Medicaid plans. And to be able to do more than that with the same amount of money is really unrealistic. And what has become more troublesome now, I think, is that many primary care physicians and other primary care providers are really feeling crushed and depressed by all of that responsibility and expectation and, in many cases, administrative burden without the resources to be able to manage it.
Gordon: Isn’t that administrative burden driven by rational and appropriate needs for certain kinds of information?
Harold: Well, I think that there has been this attitude for a while that all you have to do is measure things and that magic will happen, and they will get better. And it’s pretty clear that, in most spheres, if you don’t measure things, you can’t manage them. But simply measuring something doesn’t mean you can manage it, and there have been a variety of measures developed that require collection of data that is difficult to get.
The simplistic versions designed to make the administrative burden lighter end up coming up with bad measures that don’t really measure the genuine clinical differences in patients that would be relevant to outcomes. And the measures that are more clinically nuanced are very difficult to track, particularly with the current state of data collection and electronic health records that exist today.
So I think a lot of primary care physicians—and it’s not just primary care physicians. I think this is true of specialists in hospitals as well. We get caught in this unfortunate zone between if it’s easy, it doesn’t work well, and if it works well, it’s hard. And so which of those bad outcomes do you want?
Gordon: Give me an example on both sides, if you can, about a measure that’s easy but not necessarily all that useful and something that would be really interesting and useful but hard to get at.
Harold: Well, I think one simplistic but common example of this is diabetic measurement. For years, the measure was simply, did you test somebody’s HbA1c? Well, okay, you clearly have to test it to be able to know what it is, but merely testing it doesn’t say that you’ve actually done anything about it. But the other side of that, then, goes to we then have HbA1c measures that say, “How many patients are above nine? Or how many patients are above seven or below eight?”
If you look at the guidelines for diabetic management, there’s nothing that says there is some magic for every patient that occurs at 8.95 and disaster that occurs 9.05, yet the measure could penalize a physician significantly simply for being a little bit above that line or reward them. In many cases, there isn’t much award but theoretically would reward them or at least avoid a penalty if they’re slightly below that line without having any kind of nuance as to where the patient started out. Can the patient afford the medications that they need? Do they have any kinds of other issues going on that may make it more complicated for them to be able to achieve that? Which would then say, “Well, we need to have a much more nuanced measure,” but getting a much more nuanced measure would mean collecting and recording a variety of data about the individual patient.. To me, the issue would be, whatever the guidelines are, is the physician following them, yes or no, as opposed to trying to track inordinate amounts of data so that somebody else can try to figure out whether the physician did the right thing or not.
Gordon: The challenging thing for me when I think about yes/no guideline adherence as a marker for good practice is the work of Barbara Starfield at Hopkins when she was describing that diseases are constructs that we use in medicine to make sense of complex physiology and that the boundaries around diseases are a little bit fuzzy, much fuzzier than you would think and that, therefore, a disease orientation may optimize disease outcomes but may actually sub-optimize person-level outcomes, which tend to be the outcomes that most people are after—help me feel good, live long, not feel bad from side effects, and things like that. And so there’s a tension between managing a condition and managing a person, and so I get nervous about that. I wonder how do we, then, think about measuring person-level outcomes? Is that something that you’ve thought about?
Harold: Sure. When I use the term condition-based, I generally mean all of the related conditions together, not just disease by disease. There are clearly some patients who have one disease. That’s the thing that they have, the thing that is dominating their life. And you can say, “Are they being treated appropriately for that? Are they achieving the kind of outcomes they can possibly achieve?”
There are also patients who have many, many interacting conditions and comorbidities. They need to be managed together. And I think one of the challenges we have in general is that we don’t make good distinctions between those. We want to force patients into large integrated health systems whether they have one disease or whether they have 20 diseases, which is limiting their choices if they simply need to have one thing taken care of and isn’t necessarily the best solution in many cases, even if they do have multiple conditions.
When I talk about having some kind of guideline, I think the issue would be for clinicians to, in fact, make those distinctions. What is the guideline, or what’s the appropriate way to treat a patient who has stage four breast cancer? What is the right way to treat a patient who has three different chronic conditions at the same time?
And it’s pretty clear that we won’t necessarily have the best evidence about all of those. Much of the evidence that does exist is all very disease specific and is even very patient specific—homogenous patients ideal for clinical trials as opposed to the ones that are more complex. But I think that’s a matter of recognizing the limitations of medicine and having some regard for the professional capabilities of the physicians. That’s why we send them to medical school and force them to go through lots of training so that they can, in fact, be good at making those judgements. And to try to squeeze all that back into narrowly defined quality measures, I think, both diminishes their professional capabilities and is really harmful to patients.
Gordon: And so that has occurred to me as the tension between detailed metric tracking, which may be very helpful for me if I’m facing a person with a condition. I want to see things like: Are you up to date on your vaccines? We checked your blood pressure, and you’ll have a blood test. That kind of thing is important and helpful, but the aggregate of that day-to-day, minute-to-minute work doesn’t seem to add up to an understanding of this—am I a good clinician or not? So I wonder about looking at rates of hospitalization for all cause and ED visits and things like that. Is that a more rational approach to measurement?
Harold: I think all of the all cause measures are very problematic because in order to have something that is all cause, you have to have all risk adjustment built into that too. And one of the things that we are poorest at today, I think, in health care and in measurement and in payment is the notion of risk adjustment.
There seems to be this assumption today—everybody, I think, acknowledges that you need to risk-adjust measures because there are differences in patients that could affect the outcome, but the standard response is if somebody says, “Oh, yes, we’re risk adjusting.” Okay. Fine. You’re risk adjusting, not asking, “Well, are you risk adjusting the right way?
And CMS, Medicare, for example, has been fairly consistent in all of its programs using hierarchal condition category risk adjustment, which was not designed at all to be able to measure the specific things that happen to patients or how their primary care physician is treating them. It was designed simply to be an approximate way of predicting future expenditures for a Medicare Advantage Plan. But to refine that risk adjustment method, you would have to have additional information. You would have to know things not just about the patient’s diseases but about the severity of those diseases, and that, the relevant information, isn’t collected today.
For example, we may know whether you have cancer, but there is not an ICD code for the stage of cancer, yet anybody who knows anything about cancer will say, “That’s a pretty big, important difference between whether you have stage four cancer or whether you have stage one cancer.” The kinds of treatments you’re going to get are going to be different. The kinds of outcomes you’re going to expect are going to be different, et cetera. So to say somehow that we have a risk adjustment that simply incorporates the diagnosis of cancer in it can’t possibly work.
And personally, I think we have to recognize the need for that, and we need to work on trying to improve it for long-run measurement management and not simply use the bad system that we happen to have today because everybody says you have to have a value-based payment system in place this month.
Gordon: That’s an interesting segue for me, I think, going back to something that you wrote about in your “Saving Primary Care” article when you talked about value-based systems, because my impression is that value-based payment is driving some of this, “Please track lots of administrative stuff.” And from your article, you’re saying that the work burden of that is disproportionate to the payment. Is that a consistent problem?
Harold: Well, I think there are two issues. One is I think that the work burden today is disproportionate to the actual benefit that the work achieves. Tracking lots and lots of measures that are poorly defined, not well risk-adjusted, and don’t really measure what matters is really a genuine waste of time because it’s not accomplishing anything for anybody. I think if we get to some better measures, they may require more time, but then it will be worthwhile to actually pay for that time.
And we have to think in general about the idea of if it’s important to do, then you have to compensate for it. But what we’re trying to, in some ways, is to get all of this on the cheap. I think one of the problems with value-based payment is that people believe that all it means is simply throwing a couple of quality measures on top of the current payment system, adding a penalty for that, and then somehow expecting that everything will transform.
But the problem is if it takes more time to do care in a better way, and if it takes time to be able to track something because it’s important to track it, then it’s important to pay for that because if you believe that tracking will achieve a better outcome, then it’s a high-value thing to say, “Let’s pay for doing that because we expect to get a better result.”
If someone is not willing to pay for that, then you have to ask, “How much value really is that? How much value is it to track something that we never tracked before if we’re not actually willing to pay for it?” And if, in fact, it’s taking a lot of time that we’re not paying for, and that’s detracting from actually spending time with patients and being able to do good diagnosis, et cetera, it is entirely possible that we’re making value worse by this so-called value-based payment structure that we’ve created.
Gordon: I’m thinking about the budget pressures and the premium pressures and the hue and cry over you must reduce these trajectories. How do we pay more for stuff that we value in the face of those pressures?
Harold: Well, I think that there are many opportunities to be able to save money in health care now. It depends on the circumstance and the patients and the community, but the Institute of Medicine did a big report several years ago that said that they estimated that 30 percent of the spending in health care was unnecessary. I would note in passing that about a quarter of that unnecessary, avoidable spending was administrative cost with people who are spending a lot of time watching each other rather than trying to actually improve care.
But there are clearly patients who are being hospitalized unnecessarily. There are patients who are getting unnecessarily expensive drugs. I don’t mean that the prices are too high. I simply mean that they’re getting a drug that’s an expensive drug that they don’t really need. There are patients who are getting tests they don’t need, and all of that could save an extraordinary amount of money and potentially much of it in the short run. It’s not that it takes years to be able to get to that point.
However, to get those savings, in many cases you have to pay more for something else. It’s not a matter of simply taking away something from the patient. There are some cases where clearly patients are getting things that are completely unnecessary, and you can simply say, “We’re not going to do that thing.” But in many other cases, the issue is that there’s something better to do or equally good that costs less but isn’t available or isn’t being paid enough to be able to support it.
For example, there are many patients who are getting hip or knee surgery, a very expensive operation, or back surgery because they have not been able to get any kind of appropriate relief any other way. But we don’t pay very much at all, in many cases, for the kind of care for good physical therapy, for careful management by patients to be able to do that.
There was a project that was done in Michigan where patients who needed to have back surgery or thought they needed to have back surgery simply went to a physiatrist first. And the rate of surgery went down dramatically because the physiatrist knew what he was doing and knew how to help the patient be able to address the problem that they had, but it takes time. It takes extra time. It takes more money to pay for that time. Not very much. Not very much more money, but the notion that we should pay physicians slightly more to be able to save an enormous amount somewhere else simply doesn’t seem to exist in health care anywhere today. All we want to do is reduce spending on things without paying any more for the things that patients may need in return for that reduction.
So I think there are many, many opportunities where we could say that we could improve primary care by being able to reduce spending on specialty care, but it’s not just simply taking patients away from the specialists. It’s giving the specialists an opportunity to do what they do in a different way that would be able to save money and then have the primary care physicians be able to direct the patients appropriately to those specialists.
Gordon: How would you go about shifting payment or designing a payment model that would make that work well?
Harold: Well, I think we’re doing it today in exactly the wrong way, in exactly the wrong order. People are running around trying to say, “Okay. Somehow it’s bundled payment, so let’s see what we can bundle.” And there are many cases of bundling the wrong things together.
To me, the right way starts with asking the question, where is there an opportunity to reduce spending by improving patient care? And for years, I’ve asked physicians that in a variety of settings, and they all have answers. You don’t have to drag it out of them. They can give you a long list of things that is happening to patients that are expensive and potentially avoidable.
The second question, then, is, what would you have to do differently to be able to avoid that spending? In most cases, the answer is not just stop doing something. It’s to do something else instead, like my example of the physical therapy or the work through the physiatrist or whatever, or it’s hiring a nurse care manager to be able to work with patients with chronic disease to be able to figure out why they are being hospitalized so frequently, why they’re having trouble managing their COPD, their heart failure.
The third step, then, is to say, “What’s the barrier that exists in the current payment system?” And, in many cases, we simply don’t pay for those things. So still, to this day, in most primary care practices around the country, a primary care practice cannot be paid to hire a nurse to work with their chronic disease patients.
Medicare still doesn’t do that for many kinds of patients. They have a chronic care management code now that is only available for patients who have two chronic diseases, and it has to be done for 20 minutes a month, et cetera, which in many cases isn’t all the right patients and isn’t the right amount of money and doesn’t work for small practices. So the practices can’t deliver that service because they’re not paid to deliver that service, and the patients end up in the hospital more frequently than is necessary.
The fourth step, then, is to say, “So I know there’s an opportunity. I know how I would change care delivery . Here’s the barrier in the payment system.” The fourth step is to fix that barrier, to fill that gap. But then the final and the critical step —and this is what physician practices have to do—is to say, “Well, if I think that opportunity exists and if think this approach to care delivery will be able to capitalize on that opportunity, then I should take accountability for that.” If I think that hiring the nurse is going to help patients stay out of the hospital, then I should take accountability for keeping my patients with chronic disease out of the hospital. But then you have to have the measure—the accountability measure has to be focused on that specific thing.
And so what a lot of payers try to do is to say, “Well, we’re going to make the primary care practice or whomever accountable for total cost of care.” They may be extraordinarily successful at doing the thing they said they were going to do, keeping the patient out of the hospital, but all of the other things that are changing that they can’t control makes total cost of care change, and you can’t even see the change that they made in there.
Now, if you would take all those individual things, if you gave every physician the opportunity to take advantage of the thing that they could do, you could get up to that 30 percent savings. But you don’t get it by any one physician doing something. You get it by every physician and every specialty doing the piece that they can do and then having it add up. But instead, too many payers want some simplistic quick fix where they try to shift full financial risk to a little physician practice somehow when the physician practice can do one thing but not everything.
Gordon: It’s interesting. I had not thought about it in that level of the step off from risk adjustment and total cost of care to how do clinicians begin to work on something, because I always thought the work on that thing should be able to show if we have a good enough patient population and intervention. But you’re right. The washout effect is significant because of all the other things that are happening in health care.
Harold: Well, what you’re seeing today is people think that somehow an ACO is the answer to everything. And the problem is that, yes, you might want to have coordinated care in the ACO, but the question is, what are all the people in the ACO actually going to do? And if you simply say to the ACO, which is what Medicare says today, “We’re not going to pay you differently for anything you do. We’re not going to pay you anymore for any additional services. We’re simply going to say if you can somehow figure out how to save money, we’ll give you some of it back maybe someday.”
Just to take primary care for example. If you’re the primary care practice, and you say, “Well, I think if I could hire a nurse I could actually reduce hospitalizations,” the Medicare model says, “Well, go take money out of your bank account and hire that nurse, and if it works, maybe we’ll give you some of that money back later.” Now, from the primary care practice’s perspective, they don’t have any money in the bank account to go and hire the nurse. And, second of all, they have no assurance that if their nurse actually kept their patients out of the hospital, that they will be able to get any of that because it’s all going to be in some big, complicated calculation at the ACO level that gives no assurance. No ACO has any way of assuring any individual physician that if the thing that that physician does saves money for that physician’s patients, that that physician will get the relevant amount of savings back.
So you end up with what is a very large prisoner’s dilemma problem, which is that if any individual physician works harder, spends more money that they don’t have to accomplish something, they might contribute to the overall savings of the ACO. But they may or may not get anything back, whereas another physician who does nothing at all could sit back and potentially wait for all the other physicians to see if they save money and then potentially get a share of that. So, as a result, what you have is a system where you basically just created another health plan where the ACO tries to hire people itself, care managers itself. And no one, no physicians in the system, no hospitals really do anything differently than they would’ve done otherwise.
I think if you want to have a good ACO, you need to build it from the bottom up, not from the top down. Every physician who’s in the ACO should get into the ACO by saying, “Here’s the thing that I can do differently. Here’s the way I need to be paid differently to able to achieve that,” and then join with other like-minded primary care and specialty physicians to build that ACO from the bottom up. That ACO will reduce the total cost of care, but it’s not because some payer said you’re at risk for the total cost of care. It’s because all the physicians were doing the things that they could do and were taking accountability for doing the things that they could do to be able to reduce spending and improve care for their patients. We have nothing like that in America today.
Gordon: You’ve touched on, in your writing and in some of your discussion today, the concept of independent, small versus large and organized. And I sense that there’s a bias towards large and organized because large entities are able to absorb some loss leader costs and think about going out on a limb, but that’s—
Harold: You mean I’m biased that way? I’m biased in exactly the opposite way.
Gordon: No, no.
Harold: Oh, there is a national bias, absolutely, towards that.
Gordon: Yes. Would you to weigh in on that?
Harold: I think the assumption nationally that started a few years ago was that somehow the answer to everything was big integrated delivery systems, mostly, I believe, because big integrated delivery systems had people who were on salary and who could go to Washington and tell stories about what good things that they were doing, whereas all the independent primary care practices were out there and didn’t have anybody to send to Washington.
I speak at lots of meetings, and you can usually count on the fingers of one hand—and may not even move any fingers at all—the number of independent physicians who are there. And it’s not because they don’t care. It’s because they can’t afford to leave their practice because we pay on a fee-for-service basis. And so the people who actually do the care have to be delivering care in order to get paid. They don’t get a salary to be able to go to the meeting.
Everybody heard about all the wonderful things that a few integrated delivery systems were doing, which, in fact, were good things. People like Geisinger, for example, were doing very good things. But that doesn’t necessarily mean that they were doing it because they were an integrated delivery system or that the logical connection is if you’re integrated delivery system, then thereby you will be able to do that. And it doesn’t mean that somebody who is not an integrated delivery system can’t do many of those same things.
Geisinger got a lot of attention, for example, for giving a warranty on surgeries. Well, the first documented example of anybody giving a warranty in health care was a single doc, an orthopedic surgeon in Michigan 30, 40 years ago who basically said, “I’m going to give a warranty on my surgeries. Anything goes wrong in two years, I’ll fix it free of charge.” And it’s in the literature. The doctor made more money, the hospital made more money, and the health plans spent dramatically less because all of a sudden they didn’t do any of the unnecessary stuff anymore, and there was a real reason to be able to focus on making sure that the outcomes were good.
Yet here we are today, and we still don’t do that in most health care in America. Every other product or service that you buy in America, you expect a warranty. If there’s a defect, it’s going to be fixed free of charge to you. There’s a money-back guarantee. If it doesn’t work, you don’t pay for it. Yet in health care we do exactly the opposite. We’re going to do this thing, and whether it works or not, you have to pay. And if we screw it up, you’ll have to pay more to fix the mistakes that we made. And yet we accept that, and we shouldn’t have to accept that.
And physicians, I think, can and will do better than that, but all warranties are limited warranties. They’re warranties for the things that, in fact, can be prevented, and they’re the things that are due to a manufacturing defect, not due to abuse by the user. I think we could move to a system where people are getting much higher quality care if we allow it, and yetwe’re doing everything else but.
Gordon: I wonder then —there may be independent practices that are doing interesting and valuable things, and our ability to recognize that and figure out how to protect and support them when they don’t have the time to send somebody to Washington on their behalf because they’re all independent, so it’s not a burden. They’re all individuals. So it sounds like we have a need to do something at a national level to say, “There’s good there. We don’t want to ignore it just because of a disproportionate representation of voice in DC.” Is that rational?
Harold: Well, I would extend the bottom-up notion to say that I think this should start at the local level, not at the national level. The national attitude should be, “We’re going to allow local innovation.” There are a variety of communities around the country that actually have, through what are generally called regional health improvement collaboratives, all of the stakeholders at the table and particularly when you think about it, health care is fundamentally a local enterprise.
Most health care isn’t delivered at the national level. It’s delivered locally, and most health care’s actually paid for locally. It comes from employers in the community with employer-sponsored health insurance or people buying their own health insurance, et cetera. So there’s communities that have all of those actors around the table.
And the biggest problem that they generally have is that—they come up with very innovative ways to do things and then they can’t get the payers to participate. The big national payer, Medicare, won’t do anything in individual communities. They only want to do things that they can do nationally. The big national health plans don’t want to do anything differently in one community. No matter how innovative it might be, no matter how much it might be the thing that could be replicated later on elsewhere, they only want national approaches to things. And guess what. That hasn’t gotten us anywhere.
So I think not only do we want a bottom-up approach where we say, “Rather than big integrated delivery systems, let’s let every individual physician or hospital, for that matter, develop their own approach.” We need to say, “If people in one community want to tackle this thing, let’s let them. If there are clinicians there—if there’s a rheumatologist in this community who wants to do something different for rheumatoid arthritis, we should support because it may turn out to be valuable to rheumatologists elsewhere in the country.” And if in a different community an oncologist wants to do something differently, we should allow them to do that and then think about how to take those innovations and spread them rather than waiting for years to come up with a perfect model and trying to launch it nationally because it probably won’t be perfect.
Gordon: I am quite attracted to this concept of bottom up. It makes a lot of sense to me, and I love the idea of trying stuff out to see how it works because we’re dealing with very complex systems, and a big D-Day launch is probably not going to be the solution. We certainly have had lots of those coming out of DC, and that’s been kind of frustrating.
And it sounds like we have a pretty good call to action. I like this let there be local experimentation against outcomes that matter. Let this be driven by clinicians who start the conversation by saying, “Where do I see opportunity to improve? How would I avoid doing those unnecessary things? How could I identify the barriers in the current system and design a system that would support me in doing that work? And then let’s see how far we can go with that.”
Harold: And then how can I take accountability for the thing that I said that I could do, right? It’s not what somebody else thinks I should be accountable for. It’s the thing that I think I can do. And if there’s no value proposition there, then it shouldn’t proceed. But if the doctor says, “I think I can reduce the rate of hospitalizations for my patients by hiring a nurse,” then we should say, “Sure. Yes. Absolutely. We should help you hire the nurse to be able to do that,” right? But show us that, in fact, you are doing that.
I ran a project some years ago to try to reduce hospital readmissions. In fact, it was before all the national attention came on hospital readmissions. And what we did was we hired a couple of nurses to make home visits. We couldn’t get any health plans to pay for it, so we got a foundation grant to pay for it. And my instructions to the nurses that we hired was, “Your job is to keep 13 people out of the hospital next year because 13 people out of the hospital would pay for your salary.”
And it was a frightening and liberating thing for those nurses because they weren’t being told, “You have to follow this follow-up protocol. You have to call each of the patients five times and fill out these forms.” It enabled them to use their clinical judgement about what to do for those patients, but it also meant that they had the flexibility to figure out what to do for those patients. And we found all sorts of amazing things about why those patients were having problems and why they were ending up in the hospital that we were able to solve because we were letting those clinicians, those nurses, be able to use their professional judgement about how to solve that problem.
And that’s what I think we need to get to more in health care, is not this top-down approach where somebody says, “We want every single person to be doing X because we somehow in Washington decided that that’s the right approach,” but a much more patient-centered approach that recognizes that every patient is different and enables the people who are trained to be able to help patients stay healthy and then allow physicians to take accountability for actually accomplishing something.
Gordon: Well, Harold Miller, I want to thank you for the wonderful conversation and your time today.
Harold: Well, I appreciate the opportunity, Gordon, and I hope that other people who listen to this might see the wisdom in that and help support these kinds of approaches.
Gordon: Me too.