On Monday, February 11, at the AMDIS HIMSS Physicians’ Executive Symposium, one of numerous pre-conference symposia prefacing HIMSS19, the annual conference of the Chicago-based Health Information & Management Systems Society, being held at the Orange County Convention Center, David Brailer, M.D., Ph.D., offered a highly critical, yet also nuanced, view of the current landscape around federal healthcare IT policy.
Dr. Brailer, who served as the first National Coordinator for Health IT during 2004-2006 and who is currently chairman of Health Evolution, an organization that “engages influential leaders from all corners of health to catalyze improvement in health care, sat down for a fireside chat with Richard Gibson, M.D., Ph.D., president of the Health Record Banking Alliance, and a well-known healthcare IT leader, during the AMDIS HIMSS Symposium, co-sponsored by AMDIS (the Association of Medical Directors of Information Systems), the nationwide CMIO association, and HIMSS.
Below are excerpts from Dr. Gibson’s fireside-chat discussion with Dr. Brailer.
Richard Gibson, M.D., Ph.D.: Can you give us the long view around federal healthcare IT policy?
David Brailer, M.D., Ph.D.: I think what people should remember is, when we started in 2004, with the health IT initiative in the federal government, there was no game plan. And we created one with four basic elements: put EHRs in docs’ hands; develop consumer-centric, consumer-facing records; move towards interoperability; and support public health and healthcare research efforts. And those goals came from what we had tried to do in Santa Barbara County; and by the way, that project was a miserable failure. We couldn’t get the EHRs [electronic health records] to work well, there was no interoperability, nobody wanted to share records. So we really are on the verge of a 25-year breakthrough. We’ve done poorly in getting a consumer-facing record; we’ve done somewhat well in interoperability; we’ve done nothing in the public health and discovery environment. So our scorecard for the original program is at best 50 percent. But we’re getting towards the end of those goals. And there’s a bit of inertia—but I think we’re on the verge of a major step in trying to re-articulate what health IT means in this environment today, and that’s the critical piece.
What part will physicians play in that? How will they interact with the technology landscape in the next few years?
Obviously, I’m a doc, so I’m pro-doc. And if you think about it at the federal level, when we articulated it in 2004 as a set of goals, doctors had really played a bad hand; hospitals and health insurers were rising, but doctors had done so poorly politically that they weren’t even in the mix. If you look at it today, doctors rule—they repositioned themselves in the Obamacare years, and positioned themselves as advocates for patients. Back then [in the early 2000s], doctors weren’t listened to. And doctors were in the middle of it, but weren’t really welcome. So politically, whatever happens, doctors will be in the middle of it all. And now there isn’t a patient care organization that doesn’t have a clinician-driven group, trying to help lead organizational success. So doctors are correcting some of the really tragic mistakes we [collectively] made earlier.
Is it felt that doctors can influence the course of healthcare policy, with value-based contracting, and others?
I think doctors are back to being politically influential. I remember being the first medical student on the board of the AMA [American Medical Association], when I was a med student [in the late 1980s]. And back then, the only thing we talked about was how much doctors got paid. The idea of being advocates for patients or public health was unheard of. I think doctors blew it. And how Washington works is that doctors, health plans, hospitals, and pharma, are the great powers in terms of healthcare. And the thing is, if you keep them apart, they can’t gain power. But whenever their interests coincide, [they can be very influential]. What happened was doctors formed a coalition during Obamacare with hospitals and health plans and became a key coalition that’s still in effect today. Doctors got carved out because they weren’t part of a coalition; but they are now. Value, quality, patient safety, patient advocacy—doctors are in it, and we have a full, robust agenda that people actually care about, and therefore, people listen. I think it’s just being politically sophisticated.
When you look at the issue of cost, compared to the European countries, The U.S. still paying twice as much for about half of the outcomes quality. What are people in Washington saying about the cost of healthcare and how physicians, hospitals, and pharma can work together?
There is honestly not a constituent group in healthcare that has credibility on healthcare costs on the Hill. Every single group has a plan, but not one of them walks the talk. You look at the readmissions program, and the death rate went up substantially, and costs went up, too.
So, I have a degree of cynicism: we have this machine [the healthcare delivery system] that’s perfectly designed to consume money. It is a voracious, insatiable machine, and nobody knows how to turn it off. So I think all the leaders of healthcare are genuinely scared to death that this machine will eat us all, and maybe they would turn it off if they could, but they don’t know how. So all of this talk about value-based care—the [Medicare Shared Savings Program] shows that we can create teeny, teeny progress around costs, but nobody knows how to scale that. We’re trying to reposition the Moon from earth by blowing on it from the earth.
That’s why Berkshire Hathaway, Amazon and JP Morgan are trying to do what they’re trying to do.
If you look at people my age, I’ve been committed to health cost management since I was a PhD student—as have many others—and we have nothing to show for it.
Do you have any intel on what Amazon/Berkshire Hathaway/JP Morgan Chase are going to do?
No, I don’t. My view is that I would be shocked if they were successful. And I say that only because I think the problem is bigger than what any commercial entity might do, no matter how powerful, how hyped, how committed. If they use their purchasing across their employees, how many markets could that influence? Bob Galvin [Robert Galvin, M.D.] used to be the most enlightened, forceful purchaser, while at GE. And he went into private equity, and they have five times the number of employees, and they can get a percentage better of overall HC cost reduction, but that’s it. It’s not a problem you can solve just by throwing good intentions at it.
Do we have an opportunity with patient-generated data, AI [artificial intelligence], big data, and the Internet of Things, to figure out what works—do we have the chance to show what really works in healthcare for future spending?
I think we’re closer to it than we ever have been, because of the passage of 21st Century Cures. For the first time, the FDA [Food and Drug Administration] is required to take into account real-world evidence. How that’s actually used and filtered against scientific evidence—can only be imagined. But imagine there’s a drug that’s on the market; we know that doctors can use drugs for off-label use. Suddenly, you can see how a drug works in off-label use. Doing research on that could be useful. And what about consumer behavioral modification techniques? We already do quite a bit of that, and it is real-world evidence. And frankly, real-world evidence is often not that convincing. But we will see a world of real-world evidence, but it will remain highly controlled when it comes to (intense) interventions.
What about HIPAA, right now?
HIPAA [the Health Insurance Portability and Accountability Act of 1996] is tragically broken, in two ways. One, it does not clearly say that I own my data. It says almost the opposite, in real life—that if you produce data, you control it, except you have to share the data with the patient. I have to applaud the Obama administration—they moved as closely as they could, to give patients some control. But you cannot cross into that world without statutory change. The other thing is that it’s created a world of regulated, covered entities that have regulated data like a lab test, and this whole world of unregulated data, from apps, things I put on my skin to measure things, etc. So let’s imagine I own my data. How many of your organizations will take in unclean data that’s patient-generated. Will anyone do it? I’m co-chairman of the CARIN Alliance with Anish Chopra, David Blumenthal, and Mike Leavitt to make up for the massive hole around consumer data-sharing. But in the end, nobody wants to take that data in. So here we have data that’s regulated and clean and unregulated, unclear data, and nobody wants to bring all that together. So we don’t even an entity that exists that wants to manage that data—an entity regulated for the purpose of amalgamating my data in one place. And until we have those things happening—what will it take to recognize that HIPAA is tragically broken? And you look at this Congress and their inability to get anything done, and frankly, a huge lack of expertise—there used to be a number of members of Congress with real health tech expertise—but I don’t think the conditions exist for a HIPAA redo right now.
What are some of the barriers to improving data-sharing under HIPAA?
There are three challenges. As everyone knows, under HIPAA, I have quite a long period of time from when a patient makes a request until when I have to fulfill that request. I basically have 90 days. Do most people wait that long? No. But you can’t create a data marketplace with that latency. Meanwhile, format is an issue. Nobody does things in standard formats. Everybody can, but few people do. Thirdly, while I can do it myself, I can’t designate a third party to facilitate data-sharing. So you have time, place, and entity barriers. And in CARIN, we’ve learned that Entity A might take my request, but Entity B will say, you have to sign a release for me to do this. So there’s all this legal infrastructure on top of it. And nobody has the concept of a persistent token. So there’s a very deep chasm in terms of policy and the expression of policy, as opposed to one-time, difficult, manual data collection. The issue is a legal issue driven by so many regulatory and policy issues that the legal officer says no. So unless someone shows some leadership, I don’t see us getting anywhere.
What should doctors know about AI?
There are huge variations in what we call AI and machine learning. Huge variations in the degree to which it can establish causality; and second, the ability of it to continue to learn, without being continually retrained. We’re all invested in statistics. We know all about how to establish correlations, including complex, multi-variate correlations, such factors such as for sex and age. But to establish causality—think about the complexity and what you do with AI is to try to take dirty, messy, noisy, human-intervened data, and try to establish with firm conclusions that X caused Y. it is doable, but it requires a certain set of requirements around the math, but also around the data. The question ultimate is, what’s the training data set? IDX-DR, an autonomous system for diabetic retinopathy evaluation—a doctor’s determination is not needed for diagnosis. You do a retinal scan and run it through an algorithm. It’s magical, it’s autonomous, and it’s software that operates without the need for a doctor. And what’s magical is that they got to autonomy. Why did that happen? For years, the American Academy of Ophthalmology created one of the best sets of scans of retina that have ever been created. It doesn’t happen in dermatology or mostly in radiology. Some uses have been found in cardiology. But why do we have an autonomous system in diabetic retinopathy? Because we have a fantastic training data set. Where you see a fantastic data training set, you have fantastic AI. In the end, if it’s well-trained and well-established in its causality, it will be adopted. And I think most of the uses will be in radiology. I think we’re in the mode of developing really well-curated, well-trained data sets in AI.