Skip to contentSkip to navigation

A doctor’s prescription: Data may finally be good for your health

Robert Wachter, M.D., author of The Digital Doctor, is increasingly optimistic that technology can improve healthcare. Better strategic management of people, systems, and data is making a difference.

In 2015, Robert Wachter published The Digital Doctor: Hope, Hype, and Harm at the Dawn of Medicine’s Computer Age, a skeptical account of digitization in hospitals. Despite the promise offered by the digital transformation of healthcare, electronic health records had not delivered better care and greater efficiency. The cumbersome design, legacy procedures, and resistance from staff were frustrating everyone — administrators, nurses, consultants, and patients. Costs continued to rise, and preventable medical mistakes were not spotted. One patient at Wachter’s own hospital, one of the nation’s finest, was given 39 times the correct dose of antibiotics by an automated system that nobody questioned. The teenager survived, but it was clear that there needed to be a new approach to the management and use of data.

Wachter has for decades considered the delivery of healthcare through a lens focused on patient safety and quality. In 1996, he coauthored a paper in the New England Journal of Medicine that coined the term hospitalist in describing and promoting a new way of managing patients in hospitals: having one doctor — the hospitalist — “own” the patient journey from admission to discharge. The primary goal was to improve outcomes and save lives. Wachter argued it would also reduce costs and increase efficiency, making the business case for better healthcare. And he was right. Today there are more than 50,000 hospitalists, and it took just two years from the article’s publication to have the first data proving his point. In 2016, Wachter was named chair of the Department of Medicine at the University of California, San Francisco (UCSF), where he has worked since 1990.

Get the strategy+business newsletter delivered to your inbox



Today, Wachter is, to paraphrase the title of a recent talk, less grumpy than he used to be about health tech. The hope part of his book’s title has materialized in some areas faster than he predicted. AI’s advances in imaging are already helping the detection of cancers become more accurate. As data collection has become better systematized, big technology firms such as Google, Amazon, and Apple are entering (in Google’s case, reentering) the field and having more success focusing their problem-solving skills on healthcare issues. In his San Francisco office, Wachter sat down with strategy+business to discuss why the healthcare system may finally be about to change.

S+B: Three years ago, you were a skeptic about technology in healthcare. What’s your view now?
I did not want to come off as a Luddite in The Digital Doctor, because that’s wrong — we are desperately in need of technological solutions. But I didn’t want to come off as utopian, either. I always try to be realistic. Today, things have improved more rapidly than I thought they would. For example, I underestimated the speed with which AI was going to get better, and I underestimated the degree to which the digitization of the healthcare system and data would lead to the entry of Amazon, Apple, Microsoft, and Google — along with billions of dollars of startup venture capital money — into the field. It was obvious that healthcare, which accounts for 18 percent of the economy, was ready for disruption. But when Google tried it in about 2005, and when it launched Google Health in 2008, it just couldn’t do it, because there was little the company could do when most healthcare data was stored on paper in dusty charts. Google pulled the plug in 2011 because [the undertaking] was too hard.

Today, Google and Apple don’t have to collect the data. Healthcare data is now almost completely digital. [The companies] just have to figure out a way of getting it from the hospital onto your smartphone, and then create an ecosystem where a smart kid in a garage can create an app that helps you manage your diabetes better.

When I talk about digital healthcare now, I talk about why the entry of the big players is beginning to change the landscape, in a way that is not entirely predictable. I don’t think Amazon actually knows what it wants to do; it just knows it wants to be in this space. The same is true to some extent with Google and Apple. I don’t think they fully have sorted out a path to monetization. They just know that it’s a big space, it’s wildly inefficient, there’s a huge amount of money at stake, people care about it deeply, and there’s massive opportunity for improvement.

I am beginning to see more examples of what [MIT professor and author] Erik Brynjolfsson describes as the reimagining of the work. Say a patient is admitted to the hospital with diabetes now. I’ll go and see the patient at UCSF Medical Center, and as often as not, if their blood sugar is bouncing around, there will be a note from our glucose management service with recommendations about how to manage the diabetes better.

Scaling Specialist Expertise

S+B: What’s different about that?
It’s automatic. Previously, if I was a generalist taking care of a bunch of patients in a hospital, I would have to call a specialty consult and say, “I need your help.” The diabetes management service doesn’t work that way. Now there’s a diabetes expert who, at 6:00 in the morning, in a bathrobe, drinking a latte, opens his computer. The computer system at the hospital has filtered every hospitalized patient, looking for the 20 or 25 diabetic patients whose sugars are out of whack or who are at risk of bad outcomes. It triages them, then sends the information via a customized template so that the diabetes expert sees each patient’s sugar trend and insulin dose on one computer screen. He toggles through them and says, “The care is right,” or, “No, they’re screwing up.” For the latter, he recommends that the team change the dose of insulin, so that when I see my patient, there’s a note from this Diabetes Wizard of Oz. And we published evidence last year that showed the rate of both high sugars and low sugars went down by about 40 percent with this system. It’s a massive improvement in the way we take care of diabetes.

“I’ll often see a patient who I know would have died of [his or her] condition a decade ago, but is alive because of some medical innovation. And that feels good and is a driving force.”

The reason I like this story is that it’s not all about the tech. There’s no new app. It doesn’t involve AI. Eventually the diabetes expert may well be replaced by AI, because you can only have so many answers to so many scenarios. But it’s not an informatics play; it’s a workflow innovation — reimagining the work. The most powerful element is the IT platform that allows the system to look at all 800 patients in the hospital, figure out which ones are highest risk, and connect the brilliance of the specialist to the people taking care of the patient. That’s a wonderful example of where we need to go. And we’re starting to see more and more of [such platforms].

S+B: How did this platform come about?
UCSF uses the Epic health records system, and it doesn’t have that feature. But, to some degree, it’s customizable. So our internal IT people, sitting down with the specialist, said, “What would that screen look like?” It took a handful of hours of programmer time to pick the specialist’s brain to find out what he needs shown graphically: all the sugars over the last week, all the insulin doses, any other medications that affect sugars, and results of five to 10 laboratory tests. Constructing that screen wasn’t all that hard. Ditto for constructing the triage rules to sort out the 20 diabetes patients that he should look at.

One of the fundamental questions of medicine is “How do you scale the expertise of specialists to make it available for generalists?” And that’s probably a key question in business as well.

S+B: It is in every field.
I helped start this role called the hospitalist, which became the fastest-growing specialty in medical history. [Hospitalists are] generalists who serve as the orchestra conductors for hospitalized patients. In the old days, you came in the hospital sick, and one of two things happened. Either your primary care doctor took care of you in the hospital, or a specialist saw you. Today, the first option can’t work, because it violates every principle of physics. Your primary care doctor is in the office seeing patients every 12 minutes for nine hours of the day; how can he or she be available to you when you’re desperately ill in the hospital? In the second option, in academic hospitals like mine, you were typically seen by a hospital-based specialist, perhaps a GI or lung doctor rotating on the wards for a few weeks each year. If you were lucky enough to have only one thing wrong — if you had a kidney problem and the nephrologist was on service — it worked out OK. But if, God forbid, you had a kidney, heart, and lung problem, it didn’t work. And most patients who are sick enough to be in the hospital come in with several problems, not just one.

So we realized there needed to be a generalist project manager — an orchestra conductor to oversee your care — and then the specialists would come in and make recommendations when the problems in their areas were particularly knotty.

Systems for Fresh Thinking

S+B: The changes you appreciate seem to have less to do with technological design and more to do with people getting used to the new systems, building their own variations, and making them work.
The original electronic health record was just a platform play to get the data in digital form. It didn’t do anything particularly helpful in terms of helping the physicians make better decisions or helping to connect one kind of doctor with another kind of doctor. But it was a start.

I remember that when we were starting to develop our electronic health record at UCSF, 12 or 13 years ago, I hired a physician who is now in charge of our health computer system. I said to him, “We don’t have our electronic health record in yet, but I’m pretty sure we will in seven or eight years. What will your job be when that’s done?” I actually thought once the system was fully implemented, we’d be done with the need to innovate and evolve in health IT. That, of course, was asinine.

S+B: That’s like saying to an auto mechanic, “What will your job be when we have automatic transmissions?”
Right, but even more so, because many of us saw electronic health records as the be-all and end-all of digitally facilitated medicine. But putting in the electronic health record is just step one of 10. Then you need to start connecting all the pieces, and then you add analytics that make sense of the data and make predictions. Then you build tools and apps to fit into the workflow and change the way you work.

One of my biggest epiphanies was this: When you digitize, in any industry, nobody is clever enough to actually change anything. All they know how to do is digitize the old practice. You only start seeing real progress when smart people come in, begin using the new system, and say, “Why the hell do we do it that way?” And then you start thinking freshly about the work. That’s when you have a chance to reimagine the work in a digital environment.

The healthcare ecosystem is particularly interesting because of its pressure to deliver better care at a lower cost. That pressure is natural in every other walk of life because that’s how business works, but in healthcare, we have been insulated from core business pressures. Now we’re increasingly un-insulated from them, and now we have digital as a potential enabler. That’s why it’s a very exciting time.

S+B: Do other hospitals have the same types of systems in place?
Big academic hospitals, yes. Virtually all the major academic health systems in the U.S. use the same IT platform, Epic. We’ve all recognized simultaneously that it solves a few basic problems, but doesn’t get you where you need to go. So we’re all now trying to go further. Some of it involves internal innovation, like our diabetes screen. But then I get emails from people who have developed some useful new app. Sometimes it’s from a big, established digital company, sometimes from a startup that’s entering the space, and sometimes from another hospital like the Mayo Clinic, Mass General, or Johns Hopkins. There’s a little bit of a Wild West, gold rush mentality now, as everybody tries to make a splash in healthcare through digitization. Many startups will fail. This is a hard market to sell into, and if you’re not successful getting into 100 hospitals, you’re not going to be able to make it as a boutique product, produced just for UCSF.

S+B: How much of this innovation is being accomplished by other players in healthcare: the insurers, the pharmacy chains, the Walmarts?
I think there’s going to be more innovation out there than in here. In many ways, hospitals are legacy systems. I mean, we are dinosaurs. If, by 2025, someone has succeeded in disrupting the healthcare system, we’d love to think that we’re going to be the disruptors. But if you’re a kid in a garage somewhere, or if you’re an Apple, you think of us as precisely what needs to be disrupted. We’re part of the cost structure. We’re part of what isn’t working. Since my hospital is in the Bay Area, I’m trying to be close to the leading edge of innovation. But history would say that we’re probably in some trouble.

S+B: How do you see the relationships evolving between the upstart innovators and the legacy hospitals?
It’s really unpredictable, because each side has assets the other doesn’t have. At some level, you can imagine that each enterprise is trying to take advantage of new partnerships. At another level, we compete against one another for dollars. Now that we have digital data, that’s a very valuable resource. There’s not much that Apple, Google, or Amazon can do unless they get ahold of digital patient data.

So that makes for a lot of interesting bedfellows. In some ways, the relationship between hospitals, such as UCSF, and Google, Amazon, and Apple is less fraught than the relationships UCSF might have with insurers, because [those two parties] compete for the same dollars. We don’t compete with the tech companies; we mostly see synergies. But they may ultimately be out there trying to eat our lunch, at least in certain parts of the healthcare system. If we end up seeing true disruption from the Silicon Valley crowd, it’s hard to believe that any part of the existing healthcare ecosystem is entirely safe — no matter how prestigious your hospital is or how busy you are today.

Innovation’s Side Effects

S+B: Does this use of data and machine learning lead to exponential change? In other words, the learning curve you talked about with diabetes — does it lead to faster and better innovations?
No, it seems to be pretty incremental. We haven’t found much in medicine that creates that exponential effect.

S+B: There’s no Moore’s Law in medicine?
It doesn’t seem that way yet. We’re fighting against the inexorability of human disease. The things in medicine that have made almost exponential changes in life expectancy are things like antibiotics or a handful of highly effective medicines — like statins for cholesterol. If you look at mortality for things like heart disease and strokes, they’ve come massively down.

Of course, everybody is looking for the next magic pill. Isn’t there a penicillin for everything else? Well, not for aging, and not yet for cancer. We’ve made progress, but it’s been painstaking and incremental, and sometimes one step forward and half a step back.

One of my favorite stories involving technology and medicine is gallbladder surgery. Thirty years ago, if you needed your gallbladder out, you would have had a thing called an open cholecystectomy. It would leave you with a scar from here to here [gestures across half the abdomen]. You would have spent a week in the hospital, and a month recovering. A few patients in 100 would die of the surgery, 10 to 15 percent would suffer major morbidity like a postoperative infection, and the costs were fairly high — about [US]$20,000, back when that was real money.

Then we invented laparoscopic surgery. Once the surgeon learned how to do the procedure, which took a little while, the mortality fell close to zero. The morbidity was much lower than it had been. You were out of the hospital in a day rather than a week, and back to work in five days rather than a month. It’s a perfect example of a breakthrough technology that makes things better, safer, and cheaper.

So here’s the question: How much money do we spend as a country on taking gallbladders out now compared with 30 years ago? The answer is more. How could that be, since the technology is better? Well, in the old days I didn’t send you for a gallbladder operation unless you were at death’s door. Now if you have a little bit of belly pain and we do an ultrasound and you have a gallbladder stone — which many people do — we say, “Oh, you should have that taken out.”

So as the technology gets better, our threshold for using it changes. That’s been true of cataract surgery and hip replacement as well. In healthcare, the normal curves for increasing productivity and saving money through technology don’t compute. If you develop transformative and disruptive technology that could save the system oodles of money, there’s a decent chance it will also increase demand. People are going to buy more of it (mostly paid for through health insurance) and the costs are going to go up, and you’ll need new specialties to run it.

S+B: What do you think will be at the heart of progress in digital medicine? What makes good design and good practice?
For me, one epiphany was a recognition that the design is only one element of a much broader set of strategies. This is obviously a classic business problem. There has to be a sense of what are you trying to do, and how do you achieve innovation? And then, what is the technology that supports that achievement?

We’re not particularly good at change management or systems thinking in medicine. It’s not the way doctors are socialized, at least not until recently. So we didn’t have a ton of strategic capacity. And then on top of it we put in a technology that wasn’t very good, without much understanding of how a business takes advantage of technology to rethink the world. We’re just now having to learn all that stuff.

S+B: But you’re more optimistic than you were a couple of years ago?
I’m more optimistic in that…my goal, my North Star, has always been to figure out ways of delivering care that improve quality, safety, and efficiency. That’s not a goal of living to 120, or bringing healthcare costs down by 50 percent. That’s just making things better.

I do see us entering an era in which we are learning to use the technology in new ways, and we’re starting to make headway. Even today, as a frontline doctor, I’ll often see a patient who I know would have died of [his or her] condition a decade ago, but is alive because of some medical innovation. And that feels good and is a driving force.

Author profile:

Get s+b's award-winning newsletter delivered to your inbox. Sign up No, thanks
Illustration of flying birds delivering information
Get the newsletter

Sign up now to get our top insights on business strategy and management trends, delivered straight to your inbox twice a week.