Editor's note: David Blumenthal, MD, MPP, is Chief Health Information and Innovation Officer, Partners HealthCare System and the Samuel O. Thier Professor of Medicine and Professor of Health Care Policy, Massachusetts General Hospital/Partners HealthCare System and Harvard Medical School. He recently returned to Harvard after a 2-year stint as the National Coordinator for Health Information Technology, where he was responsible for implementing the "Meaningful Use" health care IT incentive system in American hospitals and clinics.

Dr. Robert Wachter, Editor, AHRQ WebM&M: Can you describe what you see as both the clinical and the business case for health care providers and organizations to have information technology (IT) systems in 2012?

Dr. David Blumenthal: I think they need them in 2012 so they can learn to use them, grow with them, and become part of the developing future of information technology. Almost no one who has an electronic health record prefers the paper world. Much as they will often complain about what their electronic records cannot do that they would like them to do or the ways they're not usable—there's virtually universal recognition that care is better on a day-to-day basis using an electronic record. In the present, it's better and almost everything we want to do to improve our health care system depends on good information being present at the right place at the right time—that will be the role of electronic systems. They are the circulatory system of the 21st century when it comes to information.

RW: In terms of the business case, why were federal incentives necessary to stimulate IT adoption in health care where they're not in other industries like shopping or airlines?

DB: Because the incentives in health care are so perverse and misaligned. There's no business case for reducing cost or improving quality, and therefore no business case for the technologies that enable you to do those things.

RW: Do you think we'll reach a point in the next 5 years when people will say that the federal incentives were important as a stimulus, but that health care IT has become absolutely business critical?

DB: I hope so. I cannot be sure. The next election will be important in that regard—our ability to bring some order and rationality to our insurance markets will be really important. The accountable care organization movement has not taken off with the speed that many hoped. That was the next solution to creating a business case for accountability. We still are struggling to find a way to create health care reform on the organization and provider side that will generate the business case that we need. In the meantime, there is a modest opportunity to collect federal funds for becoming a meaningful user of electronic health records. Without that, I'm quite confident that the rate of increase of the use of electronic health records would have continued to lag in this country.

RW: Where do you project it will be in the next 5 or 7 years in terms of adoption rates, both inpatient and outpatient?

DB: I think within 2 to 3 years we'll have 70% to 80% of primary care physicians with electronic health records. Within 10 years, we will have virtually 100% of primary care physicians and probably 60% to 70% of specialists. We already have a substantial number of hospitals, a third with basic electronic health records. They seem to be pushing ahead much faster than I expected. We will probably get to 50% to 60% in 5 years for hospitals; rural facilities and small facilities will lag because of the cost of putting those systems in place. I remain optimistic that for the smaller, less complicated part of the market, relatively cheap, easy-to-implement solutions will arise that are cloud-based and maintained offsite and that are available on a leasing or installment basis, which will be affordable for small places.

RW: Do you see a transformation in vendor-built systems occurring on the horizon? For those of us who are putting IT systems in now, everyone is struck with how much better they are than several years ago, but there are still reports about failed implementations and unintended consequences. Do you think that will turn around soon, that systems will become markedly better?

DB: I think they'll become steadily better. They'll never get better as fast as we want them to because, with the exception of perhaps some systems—like the very tightly purposed Apple products—there are very few things in the IT world that people find really charming. But this is a process that is like being on an escalator. I'm old enough to remember when we debated which calculator to buy: should we buy the $200 one that was about as big as a PC is now, or should we wait until it got a bit simpler? Some of us preferred to stay with the slide rule. But eventually you got off the slide rule and got into the calculator, and the calculators got better and better. Then pretty soon, lo and behold, you had computers. So that's the process we're on in health information technology. It is a feature of technology that it's able to improve, needs to improve, and dissatisfaction with current technology is the motivation to improve.

RW: How important is it for the IT industry to have competition? It does seem like the vendor field is narrowing to a relatively small number of winners, and once you're on a system it's awfully difficult to leave it. What do you think about the marketplace?

DB: The marketplace is more diverse for small- and medium-sized hospitals than it is for large systems. For large systems, there has been a market concentration. That is a relatively small part of the market. Places like UCSF, MGH, Duke, and Hopkins have a relatively small number of options. But physicians in small- and middle-sized practices have huge numbers of options. Dozens of new cloud-based or iPad-based products are coming around. The likelihood is that small hospitals will also see much more diversity. I find it somewhat perplexing, and perhaps discouraging, that a relatively small number of companies are serving the large systems. It may be that some of those groups making their entryway in the 250-bed hospital sector will in 5 to 10 years eat the lunch of those that get too complacent.

RW: The literature in the last 5 to 7 years about unanticipated consequences—does it feel like it's of the right volume or is it in some ways overwhelming? If you woke up 5 years ago and read the literature about IT, you might believe that in some ways there are more harms than good?

DB: If you read the literature 5 years ago, you would have found that there was more good than harm but it was occurring in a small subset of institutions. If you read the literature 2 years ago, you would overwhelmingly find more good than harm; it's a more representative set of institutions and a more representative set of products. The impression that the literature is negative is a misimpression. No review of the literature finds more harm than good. The review done at RAND/UCLA published in the Annals of Internal Medicine in the mid-2000s basically said the literature is mostly positive, but we cannot generalize from it because it's four institutions and they're all homegrown. A study published in Health Affairs 2 or 3 years later more or less said the same thing. Then another study published a year ago in Health Affairs said the literature is overwhelmingly positive, and it now includes commercial systems and a wider range of organizations. So there does seem to be a continued trend in the same direction. What happens is that the press—because their editors find it much more interesting—pays a lot more attention to the negative studies than they do the positive studies. If you talk to reporters, that's what they'll tell you. "We all assume that this stuff works, so it's real news when it doesn't." But in the meantime, there's this steady drip of more positive studies that stay below the radar screen.

RW: Did the unintended consequences literature influence your thinking about it?

DB: No, I think there are always unintended consequences of new technologies and existing technologies. Drugs have side effects, defibrillators fire where they shouldn't, and hips fail—what you have to do is look at the net benefit. I've always tried to keep that in mind. So there will be a new set of safety problems introduced by the widespread use of electronic health records. There will be variability in the functionality, usability, and ultimately safety of different products. But the question you have to ask from a policy standpoint is: Are we better off as a nation and society with information in electronic form than we were with information in paper form, and how do we make these products continually better?

RW: When you assumed the role of leading the Office of the National Coordinator for Health IT, how did you think about the balance between top down and bottom up, and do you think we've gotten it about right?

DB: Well, I don't know if we've got it right yet. I think it's premature to make a judgment. But I knew we couldn't do top down in the United States. It was not possible given our size, political culture, and political institutions. Whether it was wise or not, the way forward had to involve enlisting the support and mobilizing the energy of the provider community and, hopefully, ultimately the consumer community as well. That inevitability enables you to emphasize the positive aspects of a bottom-up system, the most positive being the opportunity for innovation and the opportunity to encourage buy-in. Those are two things that I think we could be much better at than the UK. I think the UK's IT implementation failed not just because it was top down, but also because it was top down with a certain philosophy. A philosophy that treated this as a procurement, like buying a passel of jet fighters or a new aircraft carrier, rather than as a cooperative effort where you had to win the hearts and minds of the users, clinicians primarily. That was a failure of strategy and tactic. It's still possible to do a top-down implementation in a system of modest size; the VA is an example of that. There was much that was bottom up about the VA's implementation, but clearly the impetus was driven from the top. So I do think it will prove to be possible to do top-down implementation. What's not possible is to do it in the United States.

RW: Let's talk about the non-clinical implications of this. We eventually will all be wired; we'll all have IT systems. How do you think that will change the nature of doctoring and nursing, the way we need to train people, and the way health professionals think about their jobs? What are the deeper implications of the wiring of American health care?

DB: It's a great question, and I don't claim to have all the answers. I don't know how we'll be using information and collecting, storing, managing, and sharing it in 15 years. We may have technologies that you cannot even imagine right now. The ease of use may completely short-circuit many of our anxieties about how these technologies are allegedly coming between patients and their caretakers because people are spending so much time keying in information that they cannot even look at their patients. For all I know, we will have such capable natural language processing that people will never look at a keyboard 10 years from now. Clinicians will just be talking to their patients and the whole thing will be recorded, synthesized, and translated into a medical record effortlessly.

One very important thing about this particular sector, which is different perhaps from many other technological areas in medicine, is that it changes so rapidly and up till now it has been, and to some degree will remain, unregulated. Therefore, the rate of change will continue to be very fast. When I say unregulated, I don't mean that there will be no regulation, but what we're talking about here is knowledge management and knowledge stewardship. The electronic mode of information collection and dissemination is comparable in its importance to the creation of the printing press. Could we have imagined Harrison's Textbook of Medicine when Gutenberg printed the Bible? I don't think so. We just have to be careful in trying to predict the future. This is moving so fast. Who would have thought 10 years ago that the iPad would be available to help with electronic health records?

RW: You began your answer saying physicians or patients might have less anxiety because the information is seamlessly going into the IT system without the need to keyboard it. I was thinking that you might go on to ask, why would the patient actually need to go in and see the physician?

DB: I believe there is still a role for physicians in medicine. I know that some people might not agree with that, but I think it's more grounded in human psychology, the way we're wired, than it is the technology. But what we will be doing with and for patients in 20 or 50 years will be different because of the availability of information. It won't be the availability of information alone, it will be the way in which businesses have arisen to help people curate and consume information about themselves. I think there will always be the need for a connection with a caring human being who is wise and provides a kind of synthesis and counsel—and of course people are not going to be able to operate on themselves.

RW: What does a competent and excellent physician look like 5 or 10 years from now when you cannot really predict what the technology will look like? There are some that argue for example: "Why do we need to test knowledge anymore? There's Google and there's UptoDate." Where do you stand on that question?

DB: I think we need to test competencies with diagnosis, physical examination, history, management of illness, and management of populations. One area of competency we need to assure ourselves of is that physicians, nurses, and other caretakers know where to find information and how to mobilize it for their patient's benefit. So what used to be called record keeping is now knowledge management. That skill is inextricably connected now with electronic information sources. Those sources will change over time, and it's the generic skill with knowledge management that we need to continue to test.

RW: People have raised the concern of the "glass cockpit syndrome," where if we go too far in the direction of relying on IT for everything, what happens either when the system is down or the system isn't responding appropriately? Do you worry about that?

DB: I do worry about that. I worry that the human brain doesn't have a backup. If we give it up, it's not clear to me that these systems will ever be nearly as good as we are. I say that with a little bit of trepidation now knowing that IBM and Watson are being used as we speak—people are pouring clinical information into Watson to make it the equivalent of the Jeopardy champion for clinical care. So I can imagine a time when much of routine diagnosis is done with computers. I'm not sure where the physician sits in there except I do think that there will always be a wish for physicians to translate that information for the benefit of their patients.

RW: When we look at the increasing availability of new kinds of decision support tools, every health care organization is going to need to make difficult choices about calibration. Concerns about alert fatigue on the one hand and making the caregivers unhappy, on the other hand not being able to tolerate poor performance because of transparency or pay-for-performance initiatives, or pushback from people about stifling innovation and creating more robotic care. How do you think about getting that balance right?

DB: I think we've got to go a long way before we overwhelm physicians with decision support. There's the example of alert fatigue with respect to drug–drug interaction, that's the most common area where that shows up. But there's so very little decision support that is incorporated into most electronic health records that we have a long way to go before we have to do the careful calibration that you're discussing. We will have to do that. But I think we shouldn't allow that at this point to hold off on trying to create intelligent supports for the routine care of patients.

RW: Can you talk a little bit about your tenure in DC? What surprised you about it? What did you find particularly interesting?

DB: I was pleasantly surprised at how much running room, how much authority, and how many resources I had. I was surprised at the opportunity that was available to me that I didn't anticipate. So that was one important thing. I was surprised at the intensity of the press scrutiny that I had at times. I figured that IT was a pretty benign, white hat kind of thing, but not everyone thought that way. I was also surprised at the diversity of the challenges I faced from organization building to policy development to mass communications to managing staff to just a whole range of things. It was very challenging and instructive.

RW: And the non–white hat aspect of it, where was the sniping coming from?

DB: What I learned was that whenever a lot of money is made available by our government for anything, there will always be a paranoid group that assumes it was done because of corrupt and nefarious purposes. The IT movement went from being a victim, under-resourced savior for everything, almost overnight to being bloated, industry-dominated, unsafe, dangerous, and overly favored. It was very dramatic. It happened just because of the allocation of funds of the magnitude, and with the rapidity, that they became available.

RW: You have had an extraordinarily successful career as an academic policy person. How did your time in Washington inform your day job now when you got back to Harvard?

DB: Well, I am now Chief Health Information and Innovation Officer at Partners HealthCare. My role is taking and implementing a new IT system for a system that has revenues of close to $10 billion, 60,000 employees, and 15 major facilities, a rehab hospital, mental health, community hospitals, teaching hospitals. So it's in some ways a much smaller terrain than I operated on in Washington, but in other ways much more challenging because it's much more detailed and intricate and real in many important ways. I am finding that some of the basic skills and requirements that pertain in public policy also pertain in private management. Communication is still extremely important, and consensus building, governance, and being very strategic about deadlines. So there were a whole series of things that I learned in Washington that are applicable even in this role. Strangely, the Meaningful Use requirements and payments are not particularly important because we can meet them with our old systems. What we are trying to do is build systems that enable us to take better care of patients for decades to come, and that's a totally different challenge than meeting Meaningful Use standards.

RW: Given your life in research, what are biggest research questions in IT that need to be answered?

DB: Well, I don't know if they're research questions. They're big mega questions in which research may play a role. One thing we need help with is figuring out how to create huge functional, flexible nimble information exchange systems. We don't know how to do this, and we don't even have a good theory about how to do it. So we need a lot of work in that area. That's right now mostly conceptual, but there are many model-building activities that could be undertaken.

RW: Is that mostly a technical problem or a political problem?

DB: No, it's a socio-technical problem. It's technical and political and economic and cultural and organizational. It has every dimension. It's one of the truly fascinating health care challenges. Much harder to do than mapping the human genome. And its never been done in the history of human affairs for health information. No models exist. So it's a green field project. So that's number one. Number two is that we have not developed really robust systems for supporting decisions by physicians. I'm not sure why that is. But the question of how to create integrated clinical decision support is a big challenge facing the field. It's a knowledge management question. That's another area that stands at the borderline between research and development.