Why Should Doctors Care About Big Data
When he speaks at Healthcare IT News' Big Data & Healthcare Analytics Forum next month in Boston, Robert Wachter, MD, will have some provocative things to say about quality and safety – and the responsibility physicians have to embrace the promise of business and clinical intelligence.
[See also: Big data doesn't have to be 'Star Wars']
Associate Chair of the Department of Medicine at University of California San Francisco, Wachter is a pioneer of the hospitalist field (he coined the term), a medical specialty that focuses on primary care in inpatient settings. He's a past-president of the Society of Hospital Medicine and has helped the specialty become the fastest growing in healthcare.
[See also: Analytics and the future of healthcare]
Wachter is also author of a forthcoming book, due out in 2015, titled, The Digital Doctor: Hope, Hype, and Harm at the Dawn of Medicine's Computer Age. In it, he explores the consequences, both hoped for and unintended, of the great health IT revolution.
On Thursday, Nov. 20, at the Big Data & Healthcare Analytics Forum, Wachter will deliver a talk titled, "The Value Agenda: Why Quality, Safety, and Patient Satisfaction are No Longer Elective." In it, he'll discuss how physicians must make smart use of data and analytics – not just to heal their own patients but also to help improve the prognosis of the healthcare industry itself.
We spoke to Wachter recently from Boston, where he's taking a sabbatical from UCSF to finish up his book.
Q: Tell us a bit about yourself and about how you helped create the hospitalist specialty.
A: I'm a general internist and an academic physician. My career can best be explained as "What happens when a political science major becomes an academic physician." I got very interested in how the system works, or doesn't.
In the mid-90s I was given a new job: to run the inpatient medical service at UCSF Medical Center, a big academic hospital. I had a very smart boss, and he said, "The service looks like it's organized when I was a resident here 20 years ago. That can't be right. Come up with some new model."
I started sniffing around to see who was doing innovative things in inpatient care. And I started to hear tales of different ways of doing it. The overarching theme was moving from the old model of your primary care doctor taking care of you in the hospital to a new model where a separate doctor did that. You can argue there are a lot of reasons why that might be a bad idea, in terms of discontinuity. But it struck me as probably being a good idea.
And being in keeping with what happened with the rest of medicine – if you think about it, in the old days there were no separate intensive care unit doctors, no separate emergency room doctors. You expected your own doctor to go there and take care of you. And then over time people said, "That doesn't work. These places are really complicated. We need someone who's there all the time. Who lives there. Who understands the system well."
And so I organized a model of doing that in my own hospital and wrote an article in the New England Journal of Medicine that coined the term "hospitalist." I started getting calls from hospitals all over the country saying, "This is exactly what we need. Come out here, and tell us how to do it."
More interestingly, I began getting calls from doctors who said, "I've been doing this for five years here in Springfield, Mass., or Gainesville, Fla., and I thought I was the only one in the country." I thought it was local and idiosyncratic or had to do with the nuances of payment or physician structure and preferences.
That's when I realized something interesting and organic was going on. And to make a long story shorter, that was 1996, and we're now almost 20 years into it. It's the fastest growing specialty in the history of medicine. Over 40,000 doctors now, a thriving professional society, all the attributes of a specialty: textbooks, meeting, board certification.
The evidence, by and large, supports the premise I had in the beginning. Concerns remain, though, which is why I got interested in technology. We do need to figure out now how to move information effectively from one doctor to another doctor, from one setting to another setting. But we have to do that all over the place in medicine.
By and large, I think it works better: The idea of having your own primary care doctor take care of you in the hospital sounds romantic and sounds terrific. When Marcus Welby did it, it was great.
But it just doesn't work. Patients are too sick in the hospital. They really need someone to be there, and primary care doctors are too busy in the office juggling the balls they're juggling. I think this is a better mousetrap.
Q: It's clearly an idea that's found a need. That's got to be gratifying.
A: At the Society of Hospital Medicine meeting, I get chosen each year to give the closing address: speaking to 2,500 or 3,000 people who really want to do the right thing. My group at UCSF is now 60 physicians. They're amazing people, and they do great things every day.
I'd say the most gratifying thing was a very lucky break. As the field began to grow, we came of age at precisely the same time as the healthcare industry was being pushed to transform itself in ways that were very different than what I grew up with. We were being pressured quite vigorously to figure out how to provide high quality, safe, satisfying care at a cost that won't bankrupt the country.
A lot of other physician groups said, "Leave us alone; we're too busy." For our field, since we're a brand new field, and a generalist field, we don't have a procedure, per se. We said, "Terrific. That's exactly correct. That's exactly what we need to be doing."
So we jumped in with both feet to this idea of value improvement, the idea of making the system we work in work better.
And what's unbelievably gratifying now is to see leaders in hospitals all over the country emerge from this field. The top physician in Medicare is a hospitalist. The surgeon general nominee is a hospitalist. They are really emerging as national and local leaders. Many CMIOs are hospitalists.
That's not surprising to me. It's born of the field's driving and founding philosophy, which is: We are here not just to take care of the individual patients but to take care of this other really sick patient which is the healthcare system.
Each one deserves a lot of attention, and each one is hard, and each one requires special training. The same rigor you put into being a really good doctor – to diagnose people and know how to treat them – we also need to get trained to diagnose the system, when it's screwed up, and make it work better. I'd say that's the most gratifying thing of all, seeing that all come together.
Q: It's clear why a hospital C-suite would pay close attention to analytics and business and clinical intelligence – they want to reduce readmissions, avoid penalties, increase patient satisfaction scores and improve their bottom lines. Why should physicians, with so many sick patients and so much else on their plates, care about big data?
A: A quick story: I was speaking to the medical students at my school about a year ago. I said to them, "You people are entering this field where you're going to be under intense pressure your entire career – that was very different than what I was under – to deliver high-quality, safe, satisfying care at the lowest possible cost."
I was trying to shake them up. One of them raised his hand, and he said, "What exactly were you people trying to do?"
I wake up in the morning and say, as my defining mantra, that the system has every right to say to us as a profession, "You are here to serve us." You think you've been doing it your whole career, but that's not the game here.
The game here is to deliver incredibly terrific, evidence-based safe care – and to do it at a cost that's survivable. I think physicians believed for a long time that we're not part of that. That we sort of operated above, or independent of those imperatives that somehow our ethical duty was to focus like a laser on what the patient needed, and damn the costs.
I think we're waking up now and realizing that that's not right. In a no-money, no-mission way. As in: If it's damn the costs, we're going to go out of business and not be able to do the things we need to do.
And it's right for our individual patients as well. When we are profligate in our spending we don't take advantage of the data we have to figure out the best way to treat patients, the best way to prevent bad things from happening, the cheapest way – we often use mealy-mouthed words, but the correct word is cheap – the cheapest way to safely and effectively take care of a patient. Should that be in the hospital, should that be at home, should that be in a clinic?
When we're not doing that, I think we're not following our Hippocratic Oath.
Now, does every doctor need to be an expert on analytics and big data? I don't think so. In the same way that not every doctor needs to be an expert in surgery or radiology.
We all need to know how to use it; we all need to know what to make of it; we all need to be good consumers of it. That's a new set of competencies that's extraordinarily important. And bring the doctors and the systems they work in closer together. Good systems that survive and thrive in the future are going to be ones where that way of thinking – that the C-suite thinks this way, but I'm a doctor and I think this other way, and somehow I'm on a little higher horse, a little bit more morally pure.
That's all going to go away. We're all going to think the same way. The job here is to produce the best care at the lowest cost. And there's a set of structures and culture and data and analytics that allows us to do that. A good doctor will say, "Terrific. I need to be a part of that. I need at the very least to understand how to do that."
And some of us need to be experts in that because if it's just non-clinicians that do that they won't be asking the right questions, and they won't be able to communicate as effectively with their brethren as some people who are physicians.
It's quite parallel to the emergence of CMIOs. I spent the other day at Epic. You could argue: "Do IT vendors need to have physicians and nurses on staff?" You could argue: "Why, this is all about technology, this is all about code." But they've all come to realize, not really!
You need to understand the workflow; you need to understand how these people think. And you're going to have to have other people who cross over between these two worlds. I think the same is true with analytics and data.
Q: Talk a bit about your equation for determining "value" in healthcare: quality plus safety plus patient satisfaction, divided by cost.
A: It's not rocket science. Every industry in a capitalist economy is driven toward producing the best product at the lowest cost. The code word for that is value.
What's funky about medicine, as that medical student asked me, is that we have really not been. There are multiple reasons. One is that the insurance system insulates everyone, to a large extent, from cost. The costs get hidden and moved around in funny ways that make them not obvious.
This morning I had to run out and get my large mocha because otherwise I would not be functioning at 9 a.m. I have to decide whether it is worth $3.65 to give to Starbucks every morning for this cup of coffee. And I do that based on my consideration of value: quality, satisfaction, whatever pleasure it gives me, divided by the cost. I decide whether it's worth it. That cost comes completely out of my pocket, and if I decide one day that it's not worth it, and the $1.80 I can get from Dunkin' Donuts down the street is good enough, and I'm gonna save that money and do it another way, that's what I'll do.
We have an abiding belief in America, and I think it's been largely borne out, that that sort of pressure – producing the best thing, however you define the best, divided by the lowest cost, or the survivable, acceptable cost, is how we ended up with Google and how we ended up with Apple and how we ended up with Amazon. These companies that seem to work and seem to thrive.
Medicine has been insulated from that. And I believe that the abiding philosophy of modern American health policy is that insulation has led us down a dark alley – to care that is too expensive, that is unsatisfying for patients, to access that's not good enough, to quality that's spotty and care that is often unsafe.
The question that health policy gurus grapple with is how do you create an ecosystem that drives doctors, hospitals, vendors and new entrants to the field to come in and deliver the best product at the lowest cost.
That's very tricky because the insurance is there, and needs to be there – if we take insurance away and treat it like Starbucks, a lot of people are going to go broke very quickly.
And it's difficult because when I judge whether my tall mocha is worth $3.65, I'm a pretty good judge of does it taste good and make me feel better in the morning – whereas patients have a very hard time judging whether their doctor or hospital is any good.
So you have all these nuances that make healthcare purchasing a little different. But the push to deliver value is, I think, the defining moment of where we are in health policy. We're trying to figure out how to create a system that delivers appropriate incentives to everyone, including patients, to increase the numerator – the good stuff: better-quality, safer, more reliable – and does that at, if possible, a lower cost.
Q: So as doctors try to do these two jobs – healing their patients and trying to improve the healthcare system – what are some ways of getting them on board with all this data we have? Are there ways to improve the technology and the user experience? What role should that play?
A: I think that's a real part of it. I'd start even before that – which is to say that the way to get doctors, who are largely both Type A and used to getting As on tests, to care about data is to show it to them and illustrate how it can be useful to them in ways that are in sync with how they already think:
"You, doctor, want to do the right things by your patients. You care deeply about your patients. Here are data that demonstrate that when we look at this guideline, which you believe in and the data is incontrovertible, when we look at that, you're not doing as well as a.) you think you are and b.) your colleagues are."
Then you have to be ready for the absolute inkblot response, which is: "The data are wrong," or, "My patients are older or sicker."
You have to be able to demonstrate that, "Yes, I knew you were going to say that. And here are analytic tools" – you have to explain it in English – "that adjust for the fact that your patients are older or sicker. We've only compared apples to apples."
To make that meaningful, sometimes it's going to be appealing to their sense of professionalism: "I know you are a good doctor, here is a tool to help you be a better doctor, in the same way a stethoscope does. There's now a whole new set of tools you need to learn in the service of your patients."
Then you get to more of a sledgehammer approach: "Whether you like it or not, you are going to be measured in these ways. You are measured in these ways. You're still going to get paid by Medicare, even if you are not very good." But look at what's beginning to happen: Those measurements are being publicly reported; they're beginning to influence your payment. They're going to begin influencing whether or not the insurance network wants you in or wants you out.
Maybe you'll be able to survive doing not so well for a while because you're close to retirement. OK. But, by and large, for the physician of the future, these data are really going to be meaningful.
It's showing people the data, showing them it's meaningful, getting their competitive juices flowing. And then once you have all that established, it's showing it to them in ways that work. I think we've come to realize that all the physicians maybe have better numeracy than the average person, in some ways, but not really – and if you show them a raw spreadsheet of data or a bland curve or a bar graph, that may be in some ways less impressive than showing them in some new data visualization.
So, it's how do you get it, how do you make sure it's correct, how do you ensure it's presented in the least painful way, and how do you use it in the service of what you're trying to do?
That becomes tricky because all physicians are different; all people are different. Somebody may like seeing their data presented in one way; somebody may like it another way. I might like seeing a spider graph, and my colleague might like seeing a column or a heat map. It's going to be important to figure that out and give people data in the ways that are most meaningful to them.
Q: I hear you're working on a book, The Digital Doctor: Hope, Hype, and Harm at the Dawn of Medicine's Computer Age.
A: A year ago it dawned on me that the world of healthcare was changing very, very rapidly. This may seem odd to you if you've been working in IT, but for folks like me, who are not IT gurus, it hasn't been at the center of our existence.
That's one of the funky things about healthcare: Until 2010, maybe 10 or 15 percent of doctors offices were truly computerized and the same percentage of hospitals. Because of the HITECH Act and the sprinkling around of $30 billion of federal money, that initiative has been wildly successfully if the goal was to get our system wired. In a very short period of time, we went from an analogue system to a digital system.
And so in all of my roles, as a practicing clinician, as a group administrator, as someone who thinks and writes and speaks about improving quality and safety, I've been waiting for this forever.
All of a sudden we got computerized, and I started looking around, and I saw huge amounts of unhappiness. I saw doctors and patients not looking each other in the eye. I saw changes in workflow that completely flabbergasted me – we used to go down to radiology and have this wonderful conversation with the radiologist, and we each learned something. I didn't realize it, but the reason we went there fundamentally was because that's where our film lived. Now there's no film; there's a digital image I could see anywhere, so I stopped going to radiology.
And then the final straw was at my institution a few years ago; we gave a kid about a 40-fold overdose of a common antibiotic. And as I sat at the cause analysis meeting to find out how that happened, my jaw dropped because it didn't happen despite our computer system, it happened because of it. Subtle little problems in the display of the CPOE, alarms firing and being ignored because there are thousands of alarms a day.
In the old days, an order for 40 antibiotic pills to be taken at one time would have, at some point, gone to a pharmacist who would have looked at it and said, "What the hell?" as he or she was pouring out a bottle of these pills. Now it goes to a robot that says, "You want 40 pills? No problem."
Then it went to a nurse who said, "That's really screwy. But I have my barcode, and it will tell me if I have a problem." But at that point in the medication process, the barcode is essentially set to defend the order. It said, "No, that's the right dose." So just imagine, she gave this kid 39 pills instead of one.
It was the combination of hearing all of that, how technology is changing medicine, and it hit me like a lightning bolt one day: Wow, this is a really unbelievable moment in healthcare, and no one has written the book on this. There have obviously been a lot of informatics books, and I read Eric Topol's book, which I liked but was sort of very futuristic and about how wonderful all this is and is going to be, and that's actually not what I was experiencing: It's harder than it looks.
The book will come out in March. I'm 98 percent done with it. What's been fabulous about writing it – it's been one of the most fun experiences I've had – is that my wife, who's a journalist, said the only way you can make this come to life, this otherwise technical story, is by talking to people and getting perspective. And so I've taken that to heart and have interviewed probably more than 100 people.
It's been just unbelievably interesting. I spent a day at Epic. I spent a day with IBM's Watson. I've interviewed four of the five ONC directors. I spent a day at Boeing seeing how they design cockpits to be safe. I've watched frontline doctors and patients. Because I live in San Francisco and am on sabbatical in Boston, I've had the chance to be in both places where there are a lot of very smart people. I've spoken to a lot of cognitive psychologists and artificial intelligence experts.
It's been extraordinarily interesting. And it's confirmed my feeling: I'm the farthest thing in the world from a Luddite. I think we have to do this. There's no doubt in my mind that care is better now with computers than without them. But it is clear to me that it's not as good as it should be. It's harder than it looks.
This is an unbelievable adaptive change. The (IT) system comes in, but then you have to think deeply about everything around it: What's the sociology of the workplace and how do people relate to each other? Have we screwed that up with computers? And if so, how do we mitigate that and bring people back together in a way that is helpful?
These are things that people in IT have sort of known for years, but I think because we've flipped a switch and all of a sudden healthcare has gone from analog to digital, from 10 percent adoption to 70 percent in a few short years, it's sort of hit us like a tsunami. And I think we're all kind of staggering, like, "Wow, that's not what I expected."
Part of what's fun for me is the story. One of the things I worried about is that I'm not a technical person. But I've realized this is not a technical story. This is a clinical, sociological, political story of the transformation of a very important industry.