Perspectives on Safety—New Insights on Safety and Health IT
This month's interview features Robert M. Wachter, Professor and Interim Chairman of the Department of Medicine at UCSF. We talked with him about his new book, The Digital Doctor: Hope, Hype, and Harm at the Dawn of Medicine's Computer Age. (Podcast available.)
In Conversation with…Robert M. Wachter, MD
Editor's note: Dr. Wachter is Professor and the Interim Chairman of the Department of Medicine at UCSF, where he also directs the Division of Hospital Medicine. We talked with him about his new book, The Digital Doctor: Hope, Hype, and Harm at the Dawn of Medicine's Computer Age.
Listen to an audio excerpt of the interview (.MP3 | 11.8 MB | 8 minutes, 38 seconds)
Dr. Niraj Sehgal: You have a new book out and it's already galvanizing a lot of important discussions about the role of health IT in patient safety. What inspired your personal interest in this intersection?
Dr. Robert M. Wachter: It came from my interest in patient safety and the fact that I had been waiting for technology to enter our world for 15 years. I cannot tell you the number of root cause analyses where someone said, "If we just had computers, it would make things better." When computers entered my world, I began to see certain problems that I had not anticipated: changes in the doctor–patient relationship, changes in the way doctors and nurses communicated with each other, and new kinds of medical mistakes. There was a big one here [at UCSF] where we gave a boy a 39-fold overdose of a medicine in a completely wired environment. That case caused me to realize no one has written yet about how it really feels to be a clinician as medicine enters this digital age, and somebody needs to do it—it might as well be me.
NS: In conducting the 90-plus interviews that you did for the book, were there any that stood out or surprised you?
RW: I learned something from each interview, and it made it an incredibly rich experience. Probably the most memorable day was the one I spent at Boeing, seeing how the cockpit engineers and designers thought about their work. It allowed me to compare and contrast that with what I was seeing in health care technology. We often are accused, probably rightly, of overusing the aviation analogy. There are huge differences between the work that Boeing engineers have to do and the work that an Epic or a Cerner engineer has to do. I found that the level of understanding of how technology tools affect the experience of the end user was far more robust and mature at Boeing. That opened my eyes to a fundamentally different paradigm.
There were moments in other interviews that were profound. I spent a day in the living room of a patient dying of cancer in Rockville, Maryland, talking about the value that he got out of his online peer-to-peer community, and how he doesn't like to talk about the fact that he's dying because so many people in the community count on him. His wife told me that the mother of a small child with cancer encouraged him to keep a happy face on the Web site because he's her child's hero. Incredibly profound and important stuff that brought me back sometimes from focusing on the technology to remembering that medicine remains a fundamentally human pursuit.
NS: You mentioned Boeing as one of many examples which you explored outside of health care, and you touched on the challenges in the analogy between aviation and health care. Are you more or less convinced how much health care can and should learn from other industries?
RW: I am more convinced of how much they should learn, yet also how different it is and how hard blind extrapolation is. Seeing and studying other industries that went through their technology learning curve and how, in the early days, their systems were clunky. People had not rethought the work. That was all very helpful because it allowed me to place health care's current experience in context. In most other industries, it wasn't like they turned on the technology and it just fixed everything.
Speaking to Andy McAfee, one of the authors of The Second Machine Age and an MIT professor who has studied the digital transformation of other industries, made clear that what we're going through in health care is a pretty natural, almost expected, set of stages where everyone thinks the technology is going to be magical. We're going to turn it on, and it's going to transform everything. But when you do turn it on, it doesn't make things better, and you don't see the productivity gains you expected. Then 5, 10, 15 years later you see them, not so much because the technology is getting better, but because people rethink the nature of the work. Understanding that deeply, as I came to do, convinced me that health care is going through the same phase.
Those analogies turn out to be extremely useful because they tell you that there is a path by which this could all work out. But when you think about the complexity of health care—the screwy incentives, the remarkable organizational variations, the scientific advances—which creates this moving playing field, it's very hard to find another industry that complex. We've come at digital transformation so late that we have an advantage of seeing how every other industry has done this. I do think there are huge lessons to be drawn from those examples.
NS: Ultimately do you come away from it feeling more optimistic about the role health information technology can play in health care?
RW: Far, far more. It was a fascinating experience for me. The book was almost a byproduct of a period of personal learning and personal growth that I was lucky enough to have the time to do. People were unbelievably open to me, both allowing me to interview them and also being staggeringly honest. Maybe because I'm not a card-carrying journalist, they didn't know what they were getting themselves into, but they really opened up themselves to me. I came into this as not a techie, not really understanding what the journey would be like and where I would end up. But I entered it quite concerned, disappointed, skeptical, and a little bit angry about where we were. Part of my reason for writing it was a reaction to what I saw out there in the literature, which was exuberant, overly hyped, and painting a picture of a world that was not the one I was seeing. So I came into it really not expecting that I would end up in an optimistic place. But I did.
I start one of the final chapters nearly apologizing to the reader. I said: you came into this expecting a realistic appraisal of why things are so bad, and I'm sure you didn't expect to see this kind of an optimistic analysis. I didn't either. But I came to believe that we can and will end up in a pretty terrific place. And that the getting from where we are to there is far less a problem of technology and much more a problem of social change, leadership, governance, government, incentives, money, and training.
NS: We often talk about To Err Is Human putting the field of patient safety on the map. If a second report was released next week, where do you see health IT falling in the pecking order relative to medication errors, wrong-site procedures, and/or communication and teamwork failures as key contributors to patient harm?
RW: It's interesting because it would probably come up very high on the list of solutions and causes, which is a challenging place for it to be. Gordon Schiff's group released a study looking at about 100,000 reported medication errors, and 6% were attributed to technology. That's probably an underestimate. It would certainly be in the top two or three causes of harm these days. Even there, attribution is difficult because when an error makes it through because someone ignored the alerts—because they had 57 other alerts that day—would someone characterize that as a harm due to the technology? They might not think of it that way, but I do. If you narrowly frame the technology as being what's on the screen and what the alert said and what's in the software, I think you miss some of the harms of technology.
If an error happened on the floor in the old days, the nurse would have tapped the doctor on the shoulder and said, "Can you take a look at this? I'm not comfortable with the situation." But now the doctor is two floors away typing on a computer. Nobody would ever categorize that as a harm from IT, but it is. Like many things in medical errors and safety, it's hard to categorize. Certainly technology-related errors are in the top two or three causes of harms, but technology is among the top two or three saviors as well—ways that we catch errors. I constantly have to remind people after having written a book that's critical of the current state of IT that I don't think there's any doubt, nor would the literature say there's any doubt, that having IT is better than not having IT. Personally, I would not see a doctor or go to a hospital that did not have an electronic health record or computerized order entry or barcoding.
NS: How would you approach the tensions from a purely policy and incentive standpoint in terms of using health IT to be more of a solution rather than a contributor to harm?
RW: The role of policy in promoting good IT has to be quite circumspect. Trying to be relatively modest about the role of government and think of policy and government intervention where it is the only possible solution to fix a market failure is, to me, a reasonable way of thinking about this. Interoperability, systems that can talk to each other, I don't think will happen on its own in that the incentives for individual health care organizations and for IT vendors are either not powerful enough to make it happen or in some cases actually detrimental to making it happen. Government has a role to convene and build incentives and maybe requirements to make it happen. The market will not make privacy and security happen well enough by itself. The government has a strong role there. But the most important thing the government did to get IT up and running, and I think this was actually quite wise, was the $30 billion of federal incentives. Because we just were stuck at the tipping point.
There were so many challenges for doctors in their offices and for hospitals to go through the turmoil and spend the huge amounts of money to go digital. But we needed government to get us over that finish line, and I think they did it successfully. My concern now is that may have created a level of prescriptiveness that crosses the line in terms of being unhealthy. So what's the most important thing government could do? Government has a key role in creating an incentive environment where systems that deliver the best care at the lowest cost win. In that environment, the incentive for delivery systems to buy and use health IT in the most positive way flow from their incentives to deliver the best care. That provides them some room for innovation that might not be there if the IT policies are too prescriptive.
Now a problem is that the market in health IT is unlike many other markets. Even if tomorrow our institution was feeling new, increasingly powerful pressures to deliver the best care at the lowest cost and our current IT tools were not the best ones anymore—the switching costs are so unbelievably high that we cannot just change out from our current system. You don't really have a nimble market because it's not like switching from Google to Bing because Bing is better. This creates a situation that sometimes is called vendor lock-in, where once you've made the choice to go with a certain vendor you're stuck for a long period of time. Government has to think hard about that. What can it do to unlock that market? Some of that will happen with the same changes that promote interoperability. Vendor lock-in gets in the way of what you really want to see, which is a set of pressures for delivery systems to deliver the best care at the lowest cost. That pressure pushes them to buy the best IT tools and allows the market to create winners out of new companies that have developed the best IT tools.
NS: Are you really using Bing rather than Google?
RW: No, but the point is I could. If I go back to my computer in 10 seconds and I think Bing is better, it's not that hard to switch. It would probably have to be 5% better or 10% better for me to do that. If it's 1%, it's not worth the hassle. But for me to switch out of my current enterprise IT system it has to be 4000% better. That's too high a bar.
NS: Whether you're a physician, a nurse, a pharmacist, or another clinician, there's a lot of talk currently around burden and burnout that are being attributed to the fact that we've gone digital. How do we resolve that tension? We talked about it from a policy perspective, but how do we address the frontline providers who are being asked to use these tools so it's not adding to their burden, but actually contributing to better care at the bedside?
RW: We need better tools, but I wouldn't completely leave the world of policy. In the book I had a couple of chapters on the history of the note, and if you talk to doctors and nurses now part of what makes them so crazed is the note. The note began at the time of Hippocrates as a narrative description of what the patient was saying. Over the course of thousands of years it took on additional functions. Particularly over the last 20 or 30 years—I liken it to a Christmas tree on which we're hanging so many ornaments that everything's sagging—it is clinical documentation, malpractice prophylaxis, billing, quality measurement, regulatory, and it has all of these functions. Partly because a lot of outside parties care deeply about what happens inside the exam room. The note has become their vehicle to figure that out.
The term is sometimes used, strangers at the bedside. All these strangers are peering in through the window of the note. The computer has a funny role in all of this, it didn't cause that, but it's a massive enabler of it. In the days of pen and paper, there was only so much you could ask the doctor to do. If you were an insurance company and wanted to see what the doctor did, you could look retrospectively at the note and determine the bill. If you were a quality department, you could pull the chart 3 months later and see what the doctor wrote. But you couldn't create an environment where you could mandate that the doctor document all these things, nor influence them through popups and checkboxes.
At some point you wonder, do we have to make people do all this extra work and check all these boxes? We have to rethink the purpose of all of this. It's crept up on us. It wasn't like anybody being venal or too stupid. It was the one-more-thing syndrome. It was your hospital's quality department saying, well now that we have a computer we can just ask about smoking cessation; we're being judged on this. So we can put a checkbox and a forcing function so people will document this, that, and the other thing. Everyone just added one more thing, the consequence of that is about a thousand one-more-things, and it's all being felt on the back of the doctors.
How do you get it right? Well, you have to begin asking, have you created a Frankenstein where you're making the job of the doctors and nurses impossible and compromising the ultimate goals of good quality care and a good patient experience? Once you do that you then turn to the technology and say, can the technology be reconfigured in some way that promotes the simple goal of having the doctor look the patient in the eye while they're having an interaction?
This issue of the clinician experience is rising to the forefront, and people are realizing that there's a huge amount of burnout and that we have to fix that or we're going to thwart all the other things we're trying to do. So we have to look at both the policies and the technology tools. Part of the reason people are so angry about their technology tools is it's easier to be angry at the computer sitting in front of you than at the National Quality Forum or Medicare or somebody sitting far away. The computer has become the messenger that we're unhappy about. You can't just blame the messenger, you have to blame all of these other forces.
NS: So you touched on an element of how the patient relationship with their providers is changing with the computer now involved. How do we balance these elements of becoming increasingly digitalized but also wanting to have a certain patient experience and a certain level of patient-centered care?
RW: The hope is that ultimately the computer frees up time, and it does that in the same way that it frees up time in the rest of our lives. Patients or their families now can do things for themselves through new digital tools. Decisions that are algorithmic can be programmed in the computer. By analogy, I manage most of my finances myself but still see an accountant on April 14, and might see a financial advisor if my finances are complicated. And I manage my travel myself, but if I'm doing a really complicated trip I might still see a travel agent if I can find one. One can envision a world in which patients are managing a lot of their own care, or nonphysicians with the appropriate skills who are less expensive are helping patients manage their care where that's appropriate. In that kind of optimistic view of the world, you've freed up a lot of time for the physician to talk to the patient about the things that the physician uniquely can do—complicated decisions, complex diagnoses, emotionally laden issues—so I could see that happening. That's my hopeful view of the world.
If that happens, I could then see the doctor–patient relationship being reclaimed by the technology because the crux of the matter will be complicated decisions that require a lot of thought and empathy and analysis. My more pessimistic side says, well, in a technologically enabled world where every minute is now being costed out, it's the human dimension that is the thing that goes away. What I worry about is that all this self-care and technologically enabled care sounds great, but you will clearly have times where patients are doing self-care when they really should be talking to a credentialed expert. These are some of the hard choices we will have to make.
NS: If we had you look a bit into your crystal ball, how long it will take to get us to a better or perfect place? What do you think will ultimately be different in 10 years compared to where we are now? Do you think the tensions will be different? Do you think the solutions or even the workforce will be different?
RW: I think everything will be different. Ten years gets far enough out, particularly in technology, that it's almost hubristic to try even to predict where we'll be. About 10 years ago Facebook was just founded. I don't think there was Twitter. I don't think we used the term big data. There certainly weren't wearable patient sensors. There were some clunky experiments with telemedicine, but it wasn't ready for primetime. Importantly, these other policy drivers weren't in place either. The percentage of a hospital's revenue that flowed based on its performance was zero. Today that's 5%, and in a few years it will be 10% or 15% or more. So technology is not happening in a vacuum. The technology is operating in an environment where the pressures to deliver care better and cheaper are growing just as fast as the technology.
A wise person once said: we virtually always overestimate what we can do in a year and underestimate what we can do in a decade. What will it look like? I think that patients will be doing and getting a lot of self-care enabled by technology. The management of your blood pressure, your vaccinations, simple acute problems (like URIs and bronchitis), your cholesterol, maybe even your diabetes, your COPD. You may see a doctor periodically, but most of your care will happen at your home or your workplace, enabled by technology with a physician in the loop only in those times where we add sufficient value to be worth the cost. Visits will probably mostly be through telemedicine rather than in person.
Clinical research will be transformed through big data type approaches. Rather than trying to figure out whether methotrexate or infliximab is better for rheumatoid arthritis through a lengthy complex double blind trial, we'll be looking at what happened to 100,000 patients—who, for whatever reason, were on therapy A versus therapy B. The same technology and sensibility that allows Amazon to say "people who like this book also like that book" will be able to say patients who are on this drug did better than patients on that drug. Or patients who went to UCSF did better than patients who went to another hospital. That sort of thinking flows very quickly once you have a system where data are all in one central place, where all the IT systems speak to each other.
The next 3 to 5 years are really about creating the technological ecosystem that will allow us to fundamentally transform the way we do those things. Years 5 through 10, the tools are in place, now they're talking to each other. Silicon Valley has figured out in the last year or two that health care is the only part of the economy that is unwired and now is getting wired and represents 18% of the GDP. You're going to see solutions coming from companies that we don't think of as health care companies. Some that have not yet been invented and some that began life thinking about how you solve complicated problems in real estate or in finding a restaurant, then begin to turn their attention to health care, applying some of the same approaches.
NS: Finishing perhaps on a more personal note, how has the experience of writing the book and what you learned changed who you are as a provider?
RW: Oh, it has not changed it fundamentally yet. Because it's one of the challenges and miracles of my day-to-day life, which is I get to toggle between running a program and thinking deeply about the world of health care and where we go. Then I get grounded by running out and seeing a patient. That's why I love my job so much. Because I get to do all those things and they cross-fertilize one another. I don't think I could have been nearly as thoughtful about writing the book but for what I do administratively and what I do clinically. In terms of the digital work that I do, it made me probably a little bit more disappointed in today's current state because I have a better sense of where we can ultimately end up and where we can go. As an administrator and a leader, it empowered me to understand that if the technology is not helping us meet our goals, the solution may not live in Verona, Wisconsin, or in Kansas City, the solution may be here.
An example is looking at some of the computerized notes, which are just filled with gibberish because we have this copy-and-paste function. David Blumenthal said to me, "Just because we can copy and paste does not mean we must." He told me that when he was a resident, someone sat down and trained him on how to write a good note. We need to do that again in a digital environment. I don't think we have yet. That's the kind of thinking that I have become much more likely to embrace.
To say the technology is here, there are certain things that we simply cannot fix at UCSF—it has to come from a vendor. But we do have a huge amount of control, and the greatest gains will probably not come from version 9.7 of the technology. The greatest gains will come from us understanding what the technology does and doesn't do, and then reimagining our work. Sometimes that's going to be reimagining some new technology that we have to develop or bolt on. Sometimes it's going to be us saying, "I didn't expect that we would stop going down to radiology now that we no longer have to, but we have. Those interactions we used to have with the radiologist were really wonderful."
How can we reimagine those relationships in a new world without being too nostalgic and trying to recreate radiology rounds the way it looked when I was a medical student 30 years ago? That's probably not the right answer. Rather, it's important to realize there was something really important about these interactions. It might be a doctor–patient interaction, a doctor–nurse interaction. Whatever it is, there was something important about the interpersonal thing, and we've begun to lose it with technology. How could we reimagine it centered around our values, what we want to have, and what we want to accomplish? Then you come to interesting solutions and you ask, can the technology do that? Or you realize that's not a technology problem—that's a people problem, that's a leadership problem. Up until now, we have not asked those questions. We've just been surprised by how the technology has changed our practice and we've kind of lamented it—part of my book was a response to the woe-is-me and doesn't-this-stink attitude that I was beginning to feel from myself and others. I came out with a much more optimistic can-do attitude, that we have a lot of control over our new world. We have to reimagine it and then we have to start building it.