In Conversation With… David Blumenthal, MD, MPP
Editor's note: Dr. Blumenthal is President of the Commonwealth Fund and served as the National Coordinator for Health Information Technology from 2009-2011, during early implementation of the Health Information Technology for Economic and Clinical Health (HITECH) Act and the accompanying Meaningful Use program. We spoke with him about the HITECH Act and lessons learned in health care since it was enacted.
Dr. Robert M. Wachter: Tell us the backstory of your involvement in the HITECH Act. My recollection is that a memo you wrote got people to begin thinking about federal engagement and trying to promote computerization of the health care system.
Dr. David Blumenthal: As part of the campaign for the White House, I worked with several other folks, and as part of the general strategy for health care reform, we included a commitment to digitizing the health care system. We talked about a $50 billion commitment to do so. The Commonwealth Fund had asked me to write a brief essay on the rationale for electronic health records. Then I testified before the Committee on Science and Technology, and I spoke from that Commonwealth Fund essay. That essay may have influenced some of the people who ended up writing the HITECH Act.
RW: As you look back 10 years later, does the fact that many people attribute HITECH and the federal investment in health IT to you give you a warm and fuzzy feeling, or do you brace yourself for critiques?
DB: Both. I understand the failings of the HITECH Act, but I take a long view. Perhaps naively, I think 30 or 40 years from now people will trace an important advance in the way we handle information to that legislation. I see it as imperfect and not entirely successful, but an important first step. One that created what I've taken to calling a natural resource—that is digitized information on a massive scale. That resource will be mined and extracted and refined and distributed in time, once we've figured out how to do that. An important part of the Meaningful Use process was, to the dismay of many physicians, to create a data resource that was a public good that could be used for epidemiological and clinical research over time. We were thinking, for example, about what innovative future NIH [National Institutes of Health] grants would be possible if we included some data collection requirements in Meaningful Use.
RW: You were making the intellectual and policy argument for digitization, but it had to combine with the serendipity that the economy imploded at that moment and there was a stimulus package being discussed. Had that not happened, what do you think the history of that intellectual and policy argument would have been? How would that have played out?
DB: I don't think anything would have happened for many years thereafter. As I look back on the trends toward physician adoption of EHRs, the slope of the curve was somewhat greater. The trend toward hospital adoption was slow. Most of the rest of the world has seen much more rapid physician adoption of electronic health records than hospital adoption, the United Kingdom being a good case in point. But the United States leapfrogged much of the world in hospital adoption because of the HITECH Act. I think we would be waiting and waiting and waiting for hospitals to get to the point where they are right now. I understand that some may not see hospital adoption as an unmitigated positive, but on the whole it will be seen over time as a positive.
RW: Some have made the argument that it might have happened a bit more slowly, but it would have happened because of the value pressures and it might have happened more organically, for better or worse. Do you put any credence in that?
DB: Here we are in 2018, and we're looking at a value payment initiative, which is still in the balance. In general, we're moving in the right direction, but it's slow. Hospitals are still investing heavily in fee-for-service care, still behaving economically a lot like they did before the 2008 crisis. I think the chances are less than 50% that we would have seen near universal hospital adoption of health IT by 2028.
RW: What's your sense of the evidence around EHRs and value in terms of the numerator in quality and safety and also the denominator in terms of efficiency and productivity?
DB: Well, the information is incomplete and contradictory. I don't think we've seen a definitive judgment on the short-term effects of adoption of electronic records on inpatient or outpatient quality and safety yet. We've certainly seen improvements in quality and safety over the last decade. But they could be attributable to lots of things. In the work that we are doing at the Commonwealth Fund on the optimal management of high-cost patients, information technology is a cornerstone. It has upped our game. But implementation of health IT is a haphazard process, and it takes a while for people to take full advantage of the technologies at their disposal.
The other thing is that technologies are a tool, and they don't themselves guarantee a result. It's how they're used that produces a result. If the prevailing incentives in a system are such that the systems are not used in ways that produce value, then I don't think it's an issue for the technology. It's an issue for the way the health system functions. If we had an absolute imperative to improve value and there were burning platforms, you'd see quite dramatic improvements that would be unattainable without IT. Take some of the oncology order entry products that Dana-Farber Cancer Institute developed after a chemotherapy dosing error resulted in the death of Betsy Lehman [a Boston Globe reporter]. They put together an oncology order entry program that was the best of breed, and they did it because they felt they had to. When we adopted Epic at Partners, we had to get the company to incorporate the order entry system that had been developed at the Dana-Farber. Now why did Dana-Farber develop it? Because they had a burning platform.
RW: You told me once that part of the motivation for Partners to go to Epic was the forcing function for standardization. How important is that both within an individual system and maybe even nationally?
DB: I don't think the opportunity is used widely because the incentives are not strong enough. We have 800 plus ACOs [Accountable Care Organizations], but the financial incentives for the MSSP [Medicare Shared Savings Program] are not strong enough to force changes in standard operating procedures. The same is true to a lesser extent for the somewhat more risk-bearing versions of ACOs. I think the Medicare Advantage Program has more potential, and in our high-need, high-cost work, that's where we see IT used most effectively. We're waiting for the always receding horizon where the crunch actually happens and people are forced to change their behavior. Health IT will not be sufficient in its current form because it was not developed with all the functionalities that are most important to improving value. The functionality all certified EHRs do have is order entry, and the literature is pretty clear that order entry improves safety and reduces variation in practice. The full panoply of possible decision support opportunities have not materialized despite the fact that Meaningful Use pushed them.
One area I am most pleased with is the way in which health information technology has begun to incorporate patients into their care. The wide use of patient portals would never have happened without Meaningful Use requirements because doctors and hospitals just haven't been thinking that way. The extent to which those portals in their own primitive way have made it possible for people to learn about the services they receive and, in more advanced uses, to participate in correcting information. Now with the successor to OpenNotes, OurNotes, you can participate in the sharing of writing of notes. The engagement of patients has enormous potential. That was not something that the health system, left to its own devices, would have incorporated as a priority into EHRs.
RW: Do you think the evolution of patient portals and of a more population perspective happens when the value pressure grows? Or does the digital backbone have to be there?
DB: George Halvorson spent how many tens of billions of dollars at Kaiser Permanente several times over to finally get a functioning electronic health care system. I hardly ever saw George when he wasn't on his smartphone tapping into his Kaiser record and showing me the interface. One thing I've concluded from the experience of who naturally and proactively adopted records and who didn't is that risk-based payment—whether it's Geisinger, Group Health Cooperative, or Kaiser—is the natural way in which adoption could have occurred. If risk-based payment had been widespread, the HITECH Act would have been unnecessary. When you are at risk and you do adopt a cost-minimization strategy and you work up the chain of causation back to the patient, that natural evolution is not the result of IT, but IT potentiates it.
RW: The three most common critiques of HITECH 10 years later are the lack of interoperability, physician unhappiness because "I'm an expensive data entry clerk," and complaints about the vendors. That last one can go everywhere from "We promoted a bunch of systems that weren't ready for primetime" to "We created a monopoly in the form of Epic that is now stalling progress and innovation." Do you want to tackle them one after the other?
DB: The last one about Epic as a monopoly has nothing to do with the HITECH Act. It has to do with Epic's competitive advantages, which I saw firsthand when I contracted with them. Epic for a number of years did implementation better than anybody else. So if you were risk averse you would go to Epic, especially if you were big. The best examples were Kaiser, Cleveland Clinic, and Johns Hopkins. If you want to go to your CEO, CFO and say you have a reliable partner, you want to be able to point to other large organizations that have successfully implemented using Epic. Markets sometimes produce dominant vendors because of the success of those vendors. That is a matter of how markets function, and we see it everywhere—take Google and Facebook and Amazon as examples. It did not result from the HITECH legislation. Even when vendors get too big because they're better, they can become an impediment to progress.
The second issue, which is that the process of clinical data entry is causing burnout, probably has more merit. I have three children who are attendings at Massachusetts General Hospital (MGH), all in their first or second year as faculty, and I never hear them complain about data entry burden. It may have to do with the fact that they were medical students and residents in the HITECH era. But I do understand that for many physicians this is burdensome and distracting. There needs to be improvement in the records. I live with an Amazon Alexa and I can pretty much ask Alexa anything. If I were still practicing, I could use Alexa to record my notes. We need a natural language processing alternative to the data entry that clinicians now do. Our billing system may make this work because it rewards exhaustive data entry with higher fees. This may explain why data entry is not apparently a burden in the many European and Australasian countries where there are electronic health records but billing systems are much less complex, if they exist at all.
RW: It might be in E/M [evaluation and management] coding.
DB: It might be that exactly. It might be that there are not the same financial pressures.
RW: Or malpractice as well.
DB: Whatever the reason is, I don't think it's inherent to the record. We can do better. I hope we will do better. I think we will get past this issue of data entry burden. Some of it is transitional. Some of it is familiarity. Some of it is the lack of standards for what constitutes an appropriate professional medical record in the age of electronic data recording.
RW: What about interoperability?
DB: Interoperability is something where we missed the boat. But the problems we've had are deeply systemic and not technical in nature. I do feel I made a technical error as the first HITECH leader: I didn't invest enough of the discretionary funds in standards development. If we had invested more, I'm not sure that would have been successful, but it would have been a wise use of funds. We put a lot of money into state health information exchanges, partly because the law requested that we do that. But also partly because of a concept of how interoperability would evolve from a behavioral standpoint, which suggested that interoperability is a team sport, a collective activity, and that it was much less likely that UCSF and MGH would work hard together to create interoperability than it was that they would work hard within their own community to do it. We failed to understand that there were insufficient incentives to work hard within a community to do it.
RW: And some disincentives given the competitive marketplace.
DB: Right. We misunderstood the forces or underestimated the contravening forces. Having said that, by failing to develop a really effective set of standards, we made it easy to avoid or evade the responsibility to create interoperability. I don't know if it would have made a difference, but I think we underinvested.
RW: I'm going to call out things that did not exist 10 years ago and do now, and I'd love to hear your thoughts on have they surprised you, have they changed things, or how you think they're going to change things in the future. Let's start with something that HITECH did, which was ultimately lead to the entry or reentry of Silicon Valley back into the health care market and the electronic market. Every day we hear an announcement of something Amazon or Google or Apple is doing. Did that surprise you and how do you think that's going to play out over the next 5 years?
DB: It didn't surprise me. It's something we hoped and planned for. It's part of this metaphor that I used about electronic data as a natural resource like oil or coal. If you analogize to the way that natural resources get exploited, it's usually private organizations that drill or transport the resource to a refiner. A private refiner turns it into kerosene or gasoline or heating oil. And then another company prices it and markets it. I foresee that that's what's going to happen in the world of data. The Silicon Valley interest is they are the extractors, the refiners, and the marketers, and they're an essential part of this process in our highly capitalistic free market–oriented system. They are essential to wringing value out of this natural resource. They smell the black gold, if you will. I don't think they quite know how to take advantage of it. But the promise is there, so it's kind of a new gold rush located in California and elsewhere.
RW: But you have the folks who put in the original wells or created the infrastructure, who at some level don't want that to happen or believe that they can control all parts of the distribution channel.
DB: The people who have put sweat equity and real equity into creating the resource—we haven't developed a market for selling it or establishing ownership. There is a gap between the public and private interest here that is a problem. I work with Leavitt Partners and an alliance of stakeholders of various kinds who are looking more and more to consumer-mediated exchange as a way to get past those disincentives, those market failures. The idea being to give consumers their data and then let them sell it or share it with Apple and Google in return for valuable applications that make their lives better. We're at that point we can see the future, but we haven't quite organized in our market to take full advantage of it. The outlines of the solution are it will be private enterprise and entrepreneurship that breaks through. Whether the data is obtained directly in financial deals with the UCSFs and the Partners Healthcares of the world or by consumers using their right of access under HIPAA to that data and then being aggregated into groups large enough to produce value, I'm not sure. And maybe there will be a third way.
RW: How much did you worry about privacy and security when you were head of ONC [Office of the National Coordinator] and how has that played out? You've made an argument for the creation of a frictionless system where data can flow like oil around the system. And now we have this monkey wrench thrown in.
DB: I was very worried about privacy and security from the get-go. More worried than I was about standards, which reflects my personal background as more of a social scientist than a technologist. But I was very worried that the spread of electronic health records would stop in its tracks because of public pressure or public anxiety about privacy and security problems. I did work hard within HHS to get the Office for Civil Rights to start enforcing HIPAA requirements in the paper world in order to create confidence that the government would supervise the privacy of the new electronic records. In the absence of a national identifier and in the absence of the ability to use personally identified information to match records, the question was how many mistaken matches could you tolerate?
This wasn't going to be fixable until we did something that wasn't politically acceptable. So we had to accept an error rate, and the question was how much could you ratchet that error rate down and would that be a low enough rate so that you could make this whole thing fly? The story that you had to tell was that the chance of a mismatch was about the chance of being hit by lightning. If you could tell that story, people would say, "Okay, I might get hit by lightning. That's part of life. I'm not going to stay inside my whole life because I might get hit by lightning." We had to make an argument around the privacy and security thing that acknowledged that we were going to have problems and basically begged people to accept those problems for the benefits that were conferred. I think it's where we're going to end up. But I did worry a lot about it.
RW: One of the most persuasive things I read in research in my book was Erik Brynjolfsson's work about the productivity paradox and that it often takes 10 years before you resolve it. The two keys are that the technology gets better, but the bigger one is that people come in and just say, "Why are we doing it this way," and they reimagine the work. Are you seeing that and do you have any examples that you've seen that give you faith that we're actually beginning to do that?
DB: I'm, to my regret, pretty cut off from the delivery system down here in the Commonwealth Fund. My fantasy always was that, once these records became part of the infrastructure and the fabric of life, instead of fighting them people would turn to improving them—and that that would happen locally, and the same way that you need to have a pharmacy and a biomedical engineering capability and a maintenance capability, everyone would understand that you need to have a fix-it capability for the electronic health record. That meant having a cadre of programmers who could do Epic or Cerner or Allscripts, and it meant all kinds of issues around IP [intellectual property] that you had to work through with the vendor. But I saw Kaiser and Geisinger had basically said we cannot wait for Epic. We have our needs. We'll give them the IP if they want it. We're just going to make our system work. I think that's where we'll head. Frankly, this evolution is not going to get us to Nirvana because most places, at least MGH, never takes full advantage of their pharmacists. They don't have them on rounds the way they should, and they don't consult with them the way they ought to. But I did see really powerful examples at Partners of collaboration between clinicians and developers that gave me hope that this would evolve.
RW: I chaired a commission for the UK a couple of years ago, and one of my biggest takeaways was they did not have a profession of clinician informatics people. I don't know how you get there if you don't have people connecting the frontline docs, nurses, pharmacists, and the people who understand the code.
I'll give you one example that has given me faith, which is actually a low-tech example of what I think this is all about. We have an endocrinologist here named Rob Rushakoff. I was talking to one of the residents a few months ago and the resident says, "I just got Rushakoffed." And I said, what does that mean? Rob Rushakoff gets up at 6:00 in the morning and, sitting in his bathrobe with his coffee, pulls up a screen that we've built for him in Epic that basically shows him every patient with diabetes at UCSF, a 600-bed hospital, and screens them for 2 sugars greater than 220, any sugar less than 60, Type I, or they have an insulin pump. The screen shows him their glucose trends, their insulin doses, their prednisone doses, you name it. It is filtered and shows him maybe 20 patients. He flips through them and for about 8 or 10 of them he says, "They're doing it wrong." And he creates a note in the chart that looks like a consult, an unbidden consult that the resident comes and sees the patient. All of a sudden there's this note from Rushakoff.
I took him out to lunch and I ask, "Who pays you to do this?" He says, "The health system pays 15% of my salary. It takes me 45 minutes a day." Basically the time it would take him to do a single endocrine consult, he is improving the control of diabetics in this entire building. We just wrote up the outcomes in the Annals of Internal Medicine. It was 40% improvement in hyperglycemic episodes, hypoglycemic episodes, and it doesn't come out of the box from Epic. And you have to figure out the workflow and the funding and all that. But here you have one guy at home basically improving the care. It's essentially population health in the hospital.
That to me was a magical example of what this looks like. Somebody smart says why are we doing it this way—you have to call an endocrine consult, which of course you'll never do unless your patient has sugar of 10,000. I think that's the phase that we're beginning to enter. I'm far less grumpy about this than I was a few years ago. I'm beginning to see the tide turn.
DB: The thing that turned me into a supporter of the electronic health record was this one application that was imposed on us. It was probably around 2004. And it was as usual forced upon the system. Partners was negotiating with a payer for rates and the payer was saying, we need third-party authorization for your high-cost imaging, all your CTs with contrast, ultrasounds, stress tests with radionuclide imaging. The representatives of the hospital said, what if we develop a software application that does prior authorization for you. So the radiology group went to work with the IT group and they produced this application called radiology order entry, which asked ordering physicians to put in the computer basically what should be put in on a radiology order slip: 75-year-old patient with chest pain and multiple risk factors needs stress test with imaging. Put that into a field and a template, and push the button, and back would come a signal. It would be a green, yellow, or red ball, with the obvious implications. Then sometimes it might say this patient does not need the imaging part of the study. They just need the stress test. Or Doctor X just ordered a stress test 2 weeks ago for the same patient. Or even more sophisticated, a patient who has a bump in their creatinine or something and you want to image the kidneys to see if they're obstructed so you order an ultrasound of the kidneys, and the software would come back and say these kidneys were imaged on a pulmonary CT exam 2 weeks ago and were normal.
For me, this was like a bolt of lightning. It kept me from making mistakes. It kept me from spending money that wasn't necessary. It made me order the right tests. It kept my patients from going to Chelsea, Massachusetts at 4:00 on Saturday morning to find a spot in the MRI queue, and it saved them from radiation from unnecessary CT scans. It was a win, win, win all around. And it took me 20 seconds. I thought if you could do this around the United States, can you imagine how much unnecessary imaging you would prevent? So that was 14 years ago. I mean that plus interoperability is the value proposition in this whole business. It doesn't scale because of the lack of incentives. Up to now, we've been relying on Rushakoff and his altruism and his heroism, which is no way to run a railroad.