Democratizing digital healthcare: a bumpy road up front

Gavin Allen: The subtitle of your book mentions hope, hype, and harm. Of those three, which do you think is currently winning the day?

Dr. Robert Wachter: It's a mix, but I think I'm a little bit more hopeful. I wrote the book to understand why healthcare's transition from analog to digital was so painful. Why is this so hard? Why is it so bad? But the final chapter was actually quite optimistic. It's going to take years, probably decades, but we actually will get to a much better place, and there's no question that healthcare will be transformed. The experience will be better, safer, less expensive and more patient-centric.

Gavin Allen: So there will be a digital nirvana, we just haven't reached it yet?

Dr. Robert Wachter: It'll have unanticipated consequences. For instance, today we have controversies over privacy, or censorship, or what digital tech is doing to our kids' brains. All those sort of things are unanticipated consequences. Yet none of us would want to go back to a pre-digital age.

I think that's likely to be the story in healthcare. We will be able to give patients better, more convenient, more accessible, more evidence-based care. When? Parts of it will happen in the next few years. There'll be parts of it that will take 10 or 15 years to fully occur. But there will also be tremendous unanticipated consequences.

If you think about what's happening with the media in the US, it used to be that we got all of our news from three networks. Nobody would want to go back to that anymore. And yet, of course, we've introduced tremendous amounts of chaos, a lot of misinformation. A version of that will happen in healthcare, too.

Gavin Allen: What has the most potential to transform healthcare? 

Dr. Robert Wachter: I don't think we've seen anything that has truly transformed patient care. For example, I went to see my doctor last week. A patient portal allowed me to schedule my visit online, and I quickly got to see my laboratory studies and see the doctor’s note online as well. But the essence of the transaction was pretty traditional.

Where things get transformed is, not only would I see my lab results, but I would get information telling me what they mean. The doctor would get information and decision support, recommending certain treatments or tests. All of those things are possible, but relatively few of them actually happen today. So I think we're still in this transitional period. 

Gavin Allen: Digitalization has great power to break down data silos, enable teleconsultation, and improve service quality. Are we entering a golden age of hospitals? Or, as we have more remote consultations, will we be using hospitals less in the future? 

Dr. Robert Wachter: All those things you said that hospitals can do, those sound great. The problem is that, for the most part, they don't do them yet.

I think about digitization in four stages. The first stage is that you digitize the record itself. That has happened in hospitals in the United States and in many other countries around the world, where data are now collected digitally and stored in electronic health records.

The second is all of the parts need to connect. That's happened in patchy ways. My hospital’s computer system connects easily to another hospital's computer system if they happen to have the same brand. If they don't, it doesn't connect very well. 

Third, you need to glean meaningful insights from those data. Fourth, you start doing something with those insights: transforming the way you're organizing care. I think we've done relatively little of the third and the fourth. I think we're mostly at this foundational stage. In the average American hospital, we're not taking advantage of the capacity of digital the way a different industry might.

True, if I prescribe medicine to a patient who’s leaving the hospital, I can send it electronically to the pharmacy. That's nice. But the potential of digital is that I can really understand what's going on because the computer recommends the right course of treatment based on evidence, based on the literature, based on 1,000 patients like mine. It says, “Doctor, it doesn't look like you thought of this diagnosis, whereas the patient's characteristics fit this diagnosis.” Almost none of that happens today.

Gavin Allen: Why not? Is the data science not quite there yet, or is it a lack of expertise in analyzing the data? 

Dr. Robert Wachter: It's a combination. There's a lot of nuance in data about patients. Analyzing almost everything depends on the context.

Also, one of the great obstacles is it's going to be quite expensive for hospitals and healthcare systems to buy the analytic capability, whether they have to hire the individuals themselves or hire companies to do this work. Many healthcare systems work on a margin that's relatively small. The investment needed to take full advantage of digital tools, I think for many hospitals, they don't see what the return on investment will be. That's sad, because it may be that the return on investment is that the patients get better care. But in many healthcare systems around the world, hospitals don't get paid extra if patients get better care. They're paid purely on volume.

Human connection needed; human expertise, less so

Gavin Allen: Are you confident that there will always be a place for the doctor, for the human at the heart of healthcare in future?

Dr. Robert Wachter: Yes, I'm confident, but I have two biases. One, I'm a human being, so I'm rooting for the humans. Second, I'm a physician who spent 15 years training and 35 years practicing. So the idea that what I've learned is not useful, or won't ultimately be useful, is a little too painful for me to acknowledge.

When you think about the human touch in healthcare, it has two elements. One: Is the insight, expertise and experience of a human physician truly valuable and irreplaceable? I would be circumspect about declaring that too strongly, because in every other field where computers have replaced people, or partly replaced people, the instinct in the beginning was to say, “This is too complicated, it's too nuanced.” But ultimately, if a physician can do it, a computer can probably figure out how to do it, too. After all, my insights are born of hearing a data stream and looking a patient in the eye, and there's nothing about that that a computer could not replicate. 

The second element is the human connection. I actually don't mind a computer telling me I should increase my blood pressure medicine, or reminding me that I need to lose weight. But I don't want a computer telling me I have cancer or that I may be getting to the end of life, and have some very hard decisions to make. Those are profoundly human kinds of endeavors. So, as we think about the role of the human doctor going forward, we have to figure out which part of it are we talking about? The human connection, or the human expertise? I expect that the human expertise part will be chipped away.

But with the human connection, I think many people feel like, “I need a human to talk to me.” Health is different from, say, financial services. I care about my money. It's meaningful in my life. But I would be perfectly happy never to talk to a human again about my money and my investments, if the computer can replicate all the things that I need to do, and deliver them to me in a way that's efficient.

For the foreseeable future, a lot of healthcare tasks will be done by humans and augmented by computer. Looking at an X-ray, looking at a skin lesion, looking in the back of someone's eye – computers just are incredibly good at it, and getting better. For a while, there will be a human to take a second look to be sure the computer got it right, partly because of legal liability. No digital company wants to be on the hook for missing a diagnosis of cancer, for example. But over time, computers will have more of a role in those very discrete diagnostic skills that are really about visual pattern recognition.

Gavin Allen: I am wearing a wristwatch that gives me some heart monitoring and other health data. What will the future be for the patient experience, in terms of having power – almost literally – at their fingertips?

Dr. Robert Wachter: The pattern of digitalization in every industry is democratization: consumers having the ability to do more self-management. That’s mostly a change for the better, but patients can potentially be very anxious about these findings. If your watch tells you that you might have atrial fibrillation, you may not have any idea what that means, or what to do about it. So the patient sends a message to the doctor, at least in the United States, saying, "Doctor, my watch tells me I've got atrial fibrillation," or "Doctor, I just got my labs back, and my magnesium is low. What does it mean?"

I can tell you that the physicians are basically being broken by this amount of data flow. Patients still need a credentialed expert to help them interpret all of this information. When you talk to an American physician, the level of burnout is massive. And a lot of it comes because of digital. We're giving patients more and more information. But we're not giving them enough information to actually manage themselves fully. What you've created is a massive amount of confusion. So we're going to have a very bumpy five or 10 years as we democratize healthcare.

Business-class healthcare?

Gavin Allen: You've spoken a lot about democratization, and probably the key mantra for Huawei is about connecting the unconnected and bridging the digital divide, so everyone benefits from technological advances. How much can technology help bridge that divide by reducing inequalities in healthcare?

Dr. Robert Wachter: I'm a bit skeptical that wearables will make a difference. You asked before about Hope, Hype, and Harm. In that area, I think it's more hype than anything. I think those things will be tested out on the well-to-do and the privileged, but I'm not confident that they'll actually be getting better health or better health care. I think we’ll have a generation of the “worried well” who are constantly checking all these digital signals that aren't all that meaningful.

As to the broader question of the digital divide, you could make the argument that it will actually worsen care for people of limited means – at least in the short term. Because do they have smart phones? Are they digitally literate? Will they understand all these data signals? I think that over time, I would actually argue the opposite. I would argue that for a patient in, let's say, a rural place in the United States, or in many countries in the world, they have absolutely no access to a doctor, in general, or to a primary care doctor or someone to look at this skin lesion and tell them whether it's cancer or not. The ability to scale expertise and get them what they need, I think, will over time be markedly enhanced by their ability to get a lot of their care handled and managed via their computer or their smart phone.

So those who are developing digital tools need to have that sensibility. We can't only be developing tools for triathletes in San Francisco. We have to be asking, “How does this help poor people get better care and have better health?”

One of the concerns I have about the digital divide is what I talked about earlier, in terms of empathy and human connection. I think it's likelier than not that your health insurance may pay for care that's digital, and if you actually want to see a human being, that will cost more money. So we may see a stratification of care, where the first layer of the care most people get is fully digital: It's all telemedicine, or it's algorithmic and AI-based. And for many people, that will be better than the care they get today. But if you want to have a half-hour with a real-life human doctor, that may be the equivalent of business class on an airplane:  something you can get, but it costs a substantial amount of money and is out of reach for many people.

I do think digital offers the capacity to have a lower health care divide. It can help people who lack money get better access to care, better access to the information they need to manage their health, than they currently have. At the end of the day, I think the digitalization is actually good for health equity, and will improve the care of people who have limited means.

Gavin Allen: It sounds like everyone has a responsibility in that relationship: the health care provider, the technology company and of course policy makers, to ensure that even if there is a premium on the human touch, that it's not at the cost of those core benefits to everyone.

Dr. Robert Wachter: Yeah, I think that's fair. And you know, the world of digital is like that, in the beginning: it's often market-driven, and therefore available first to those who can afford it. How did the electric car get developed? By Tesla, at least in the United States. It’s a high-end status symbol, a computer on wheels. It created a market and some buzz. And then, our entire country basically began saying, “We need electric cars for everyone.” And now, you're seeing the price points come down, you're seeing other companies enter the fray.

Health care is probably going to be like that. The tools that get developed tend to be for the well-to-do that can afford them in the beginning. 

Digital health is ready for prime time

Gavin Allen: You're someone who's tweeted prodigiously about COVID over recent years. Do you think that even amidst what was clearly a global tragedy, the one tiny glimmer of light out of it was that people really started to understand the benefit of technology within healthcare?

Dr. Robert Wachter: In some ways less than I would have thought, but it's a great question. The two areas where it clearly, kind of, broke through: One was telemedicine, and at least in the U.S. and I suspect around the world.

Technologically, telemedicine was ready for prime time five years ago. But there were a number of regulatory barriers to it, and neither patients nor clinicians were comfortable with it. For example, at UCSF, where I work, about 1% of our outpatient visits were telemedicine. One month into the pandemic, that number was 70%. It's now come down to about 30%, but it would have taken two decades to reach 30% had it not been for the pandemic. So a lot of the regulatory barriers came down, and patients and doctors became comfortable with the technology.

The second thing that COVID accelerated is dashboards, which let you present data in visually attractive ways that the patient can use to find what they need. During COVID, dashboards were created for people to follow the state of the epidemic in their community or city. People saw the data in nice, well-visualized, user-friendly ways. We're past the days where you say, "I wanna take a look at the data" and someone sends you an Excel spreadsheet.

I think that's going to be very important. You go from just having raw data to having tools that people can use to take care of themselves better.

What didn't hit its tipping point, surprisingly, was artificial intelligence. I thought this would be the time that we would see more AI taking over self-help for patients, or changing the way we organize our practices. It made some incremental advances, but I don't think it was transformative.

Gavin Allen: So how do we get a real kick-start to that effect? Who’s going to provide that?

Dr. Robert Wachter: I think it's already happening, in a way. In the US, health care digitalization was impossible until we had electronic health records, until the foundational data about patient visits and patients’ overall health status was stored in electronic form.

Before that, the business case for a hospital or a doctor's office to invest in electronic health records was simply not there. So you had this vicious circle of everybody saying, “Healthcare care is the last industry that's still analog, we need to digitize,” but the private sector was not willing to invest. I was on Google's Advisory Board in 2007 when the company started Google Health. A year later, Eric Schmidt came in and told us he was disbanding Google Health. He said this is too hard for us. And I said, Okay, if it's too hard for Google, that's pretty hard. But the real problem is, if the data are not in digital form, there's nothing Google or any other company can do. That's the absolute minimum. That's table stakes.

So the US government in 2010 or so invested $30 billion to incentivize hospitals and doctors to digitize, and it worked. And so for a relatively small investment, $30 billion in a $4 trillion industry, they were able to get hospitals to make this change, and now, essentially 100% of US healthcare data is stored in electronic form. That has unleashed a tremendous investment by the private sector. The question now is, How do we take that data and make use of it, in order to improve care?

We have a lot of resistance to digital transformation because every existing entity in the health care system has a self-interest in fighting change. Physicians are part of that. Doctors are better lobbyists than taxi cab drivers. You supplant the taxi cab industry with Uber, you’ll have a lot of unhappy taxi cab drivers. But if you make doctors unhappy, we will go out there and say, “You're killing people with this,” which is hard for a taxi cab driver to do.

But I think the biggest obstacle is, “What is the business case for investment?” If my tool is going to make care better or safer or cheaper, who benefits from that? In every other industry, when I decide to go out and buy an iPhone and spend 1,000 bucks, that money is coming out of my pocket. In healthcare, who pays? Well, sometimes it's the patient, sometimes it's the employer, sometimes it's the government, sometimes it's the insurance company.

And the incentives to improve quality are very muted, because patients can’t evaluate the quality of the care they're getting. They know, Did the doctor seem nice? Or, did the clinic look like a nice facility? But they actually have no idea whether this was high-quality, evidence-based care. 

Hospitals often have little incentive to be efficient. My hospital, if somehow we figured how to stop doing X-rays because we figured out some more efficient way of care for patients, we’d go out of business by next Monday. We are dependent on having a full hospital, with our MRIs and CAT scans humming. I'm not saying that's a good thing, but it is the way we're organized.

All these factors make digital transformation harder in healthcare than in other industries, where, if you come up with a better mouse trap, and you do something better and cheaper, customers are gonna buy your product. In health care, you can do that and still go out of business.

Gavin Allen: Do you think that will change at some point? 

Robert Wachter: No, I don't think it can. Because you need to have an insurance function, because of the cost of health care. So there needs to be a third party involved.

If a billionaire who needs a heart transplant gets hit by a car, they can pay for that out of pocket. For the rest of us, that will bankrupt you very quickly. So there's always going to be a third party that covers that risk. And as long as there's a third party, you now have two people in the room making a purchasing decision: a patient and an insurer. That’s why you see a disproportionate amount of digital focus in the consumer space, where it really is a direct-to-consumer kind of relationship: help me lose weight; or, if I'm an athlete, help me to figure out how to run better or faster – those sorts of things.

But that's actually not where we need the help. We need to help someone with multiple conditions: an older person with multiple chronic illnesses, who's trying to manage their heart failure, their diabetes, their blood pressure, their prior heart attack. That just turns out to be wildly complicated. There’s almost never going to be a direct-to-consumer relationship, because there has to be a third-party insurer. Otherwise, the vast majority of people would not be able to afford the care that they need to stay alive. 

Decisions, decisions

Gavin Allen: What do you think Huawei might create that could be a transformative breakthrough in the future?

Robert Wachter: I think it would be a version of solving a problem no one has yet solved: computerized decision support.

That may sound dull. But Amazon or Netflix make recommendations that say, “People who like this book, also like that book.” That makes it easier for me to decide what to buy, or watch.

When I go on to my computer system in my office or hospital, I get remarkably little intelligence out of the computer, which is why most doctors dislike their computer. They're getting all these messages from their patients saying, "I need your help, doctor, tell me what this means." And they're spending a huge amount of time inputting information into the computer. But they're getting remarkably little insight and intelligence out of the computer to help them guide their practice.

When we try to put in things in an alert format – “Doctor, did you realize this patient is allergic to this medicine?” – we massively overdo it. I spend most of my day looking at these alerts and saying, “No, that's wrong, that's meaningless.”

I don't know whether Huawei is in that position, but we must figure out how to take all of this data that's sloshing around, turn it into meaningful intelligence, then give it to patients and doctors in forms that actually make it easier for them to manage health care. That's really the Holy Grail here. When that happens, I think we will begin to experience real digital transformation.

Gavin Allen: Is there anything we didn't touch on, that you really want to convey?

Robert Wachter: In 1994, Erik Brynjolfsson, a scholar of digital transformation, coined the term the productivity paradox of information technology. When it came to ICT, what he saw in every industry was lots of hype. “We're gonna digitize, and it's gonna massively improve everything.” And then the technology came in, and a year went by, and nothing happened. Two years, five years – nothing. And finally, somewhere between Years 5 and 10, you started seeing the promised gains in productivity.

So, the good news about the productivity paradox is that it always resolves itself, usually in years 5 through 10. The question is, Why does it take so long? Why don't you see gains beginning in Year 1? The answer that he, and other scholars, came up with was that two things have to happen.

Re-imagining medicine

First, the technology has to get better. So you need version 2.0, 3.0, 5.0, and you don't get those until the system is actually being used and people realize it's just not as good as it needs to be. They give feedback and companies respond, and you have those iterative improvement cycles. That never happens until the technology is being widely used.

But the real insight he came up with, and it's extraordinarily important for health care, was the second way the productivity paradox got resolved. It was what he called “reimagining the work.” In Version 1.0, when people first install new technology, they almost always replicate old, familiar processes. With digital tools, we just replicate what we were doing on paper. So maybe they're a bit more reliable, or there are some new efficiencies, or 10 people can view something at the same time rather just one person. But fundamentally, you haven’t changed how the work gets done.

It takes a while until people come in and say, “Why are we doing this thing this way?” That never happens in Year 1. I think that's part of what we're seeing with technology and healthcare, in the US, at least. In 2008, fewer than one in 10 hospitals had electronic health records. By 2018, a decade later, fewer than one 10 did not. So we went from analog to digital. Maybe if you take the mean year, about 2015, when we're seven years into having our data in digital form, we're in the midst of this productivity paradox, where we digitized, but really didn't change anything fundamentally. That's why a lot of exciting stuff is happening in the consumer space, because you actually can change what an individual patient does.

But in terms of fundamentally changing the health care system, the consumer stuff is always going to be at the margin. People who are sick, who really need medicines, who need tests, will always have to see a doctor, go to a clinic or hospital. So we're seeing a version of the productivity paradox.

I'm not sure how long it will take to get out of it. In the average industry, it took five or 10 years. But it's very different to think, “How do we reimagine the way we organize the factory floor, or organize the way people buy books?” versus, say, “How do we reimagine the way that we deliver health care?” – when, if we get it wrong, someone can die, and when we have very powerful constituencies that may fight change. So that's why I think we're more on the order of 10 to 20 years, rather than a five- to 10-year journey.