In Conversation with…Eric J. Topol, MD
Editor's note: Eric J. Topol, MD, is Director of Scripps Translational Science Institute, Professor of Genomics at The Scripps Research Institute, and Editor-in-Chief of Medscape. We spoke with him about his new book, The Patient Will See You Now: The Future of Medicine Is in Your Hands, and more general issues regarding the role of patients in our new world of digital health care.
Listen to an audio excerpt of the interview (.MP3 | 11.1 MB | 8 minutes, 5 seconds)
Dr. Robert M. Wachter: What is thesis or elevator speech about your latest book?
Dr. Eric J. Topol: The Patient Will See You Now is about the great inversion of medicine; it's bottom up whereby the patients will generate their own data. With computer processing, this provides a rich, granular, high-definition view of each person's medical essence and that changes everything. Democratizing medicine—that is what the book is about. Not just digitizing human beings, but where that can go in terms of the power of information and placing the individual at center stage for their health care, if they want it.
RW: What do you see as some of the advantages that are beginning to accrue?
ET: People like feeling like they are in control by having their own information, generating it, and seeing it—the average person looks at their smartphone 100 to 150 times a day. They are in constant touch with what is going on. The metric of interest about their body—that is just the beginning because as we go forward that same smartphone is a platform for doing labs, diagnosing infections (like the pathologist in your pocket), and doing some of the physical exam in addition to all the sensors. All that information can be supported through algorithms connecting with cloud and even to supercomputing. There is digital infrastructure to support each individual and get that diagnostic monitoring side of medicine covered in a different way than it has been traditionally. People who have been early adopters are gleaming because they have never had access like this. Previously they had to struggle to get their data and now it is free flowing. It is on their own devices, and they are feeling emancipated.
RW: How do we separate wheat from chaff here? You think about the physician who has advanced knowledge and skills and training trying to wade through the extraordinary amount of digital data that they are now seeing. How does a patient do that?
ET: That is a very important issue that still has to be fully addressed. We're good at generating lots of data but the whole filtering, processing—the best analytics—that is not the strong suit today. Some of the sensors are quite good at that. For example blood pressure, giving a nice screenshot summarizing data for a day, a week, or a month. But some sensors are not as good and have not been integrated yet. It will not be long before people are using multiple nodes of information that they want to integrate. For example, some chronic condition that needs to be managed, like glucose or blood pressure but also lifestyle issues, environmental centers, and labs. At the moment, we don't have the algorithms, data science, code doctors, if you will. The field has not been developed nearly the way it should be to get us to the virtual medical assistant, which integrates all relevant data about an individual and provides great data visualization back to that person. Once it does, we have a tremendous opportunity to help people, even to preempt illness. People who are at risk or already had a particular condition manifest having that flow of data analyzed, as you say, signal to noise. So that we only get signals to patients and moreover to doctors. That needs a lot of work before it is suitable for where this can go and be highly effective.
RW: You paint a picture in which the data will outstrip the analytics, at least in the short term, and also overwhelm the ability to sift through it, filter, and give it to whoever is using it, be it patient or health professional. Do you see those competencies coming from the existing world of medicine, hospitals, delivery systems, and IT companies, or will a whole new layer be developed? If that is true, where do you see that emerging from?
ET: It is a whole new layer as you've described it. We're now seeing for the first time the invasion of the major tech companies into health care. Previously they were very weak and not sustained efforts like Google Health and Microsoft HealthVault. But recently, there is a major campaign from the likes not only of Google and Facebook but Apple, IBM, Microsoft, Salesforce, Intel, and Qualcomm. These new entries are welcome because now we're talking about what has been done in other sectors to come into health care and the capabilities of dealing with big data, real-time streaming and the deep learning. Enormous data sets but now on an individual basis and ultimately at even a population level. This is important, and we have not seen it start in the health care space until very recently.
RW: What kept them out before, and what is getting them in now?
ET: What has changed is that we have things like wearable sensors and other ways to capture data that we did not have before. We made this big transition from the Fitbit fitness era to now the medical side of capturing information. We could not get blood pressure with each heartbeat, all the time, real-time streaming while you're sleeping, while you're stressed in traffic. Now you can, and over time that generates terabytes of data. Then every single condition that you can think of has a data tsunami type of story. Because of this powerful microprocessor in our pockets, you have this ability to do labs and all these things that we could not do before. These tools have transformed the landscape and led to a remarkably opportune time to change medicine and data-fy it and make it into the data science that it has not been from the origins of the profession.
The new tools are the biggest things. Then as it happens all these companies have done extremely well with mobile devices and connectivity and there has been tremendous growth. It is the kind of thing that has made Apple the highest market cap company in the world and led to many other tech titan companies' grand success. Obviously all these companies have moved to mobile platforms like Facebook and Google. This is the next big growth area as you can only go so far with travel, retail, and finance. This one is wide open. From a business point of view, all these big tech companies see that this is uncharted territory. Previously, this was just about electronic health records and health information systems and hospitals. Now we're talking about consumer-level mega expansion, and it feeds right into the growth strategies of these large successful tech companies.
RW: Let's turn to some of the safety issues, and one that readily comes to mind is patient self-management: it is sometimes wonderful and sometimes patients don't get it right. What do you worry about in terms of the locus of control shifting to patients?
ET: Plenty of issues come to mind. Certainly this is not about treating oneself. This is more about acquiring diagnostic or monitoring data and still having an intimate relationship with a doctor to help guide the appropriate treatment. There is oversight by physicians and the apps that have been introduced like doing a cardiogram, diagnosing a skin lesion, or doing an ear exam on a child, they are good in terms of capturing high-quality data and accurate algorithmic interpretation. But they are not treatment stories. That is where doctors come into play. Today we have 12 million serious diagnostic errors in this country. The fourth leading cause of death in this country is medication errors. We can do better than that. The hope is not to put the burden on patients but focus on processing this information to help get us into a safer zone than we have been. Right now, I would not think that we have a whole lot of bragging rights to the safety that exists in American medicine.
RW: I'm surprised to hear you draw a fairly bright line between the data collection and analysis, which will shift more and more to patients, but the treatment staying within the province of the physicians. That strikes me as almost the same thing. As power shifts to patients, my guess is that the same companies that are coming up with algorithms for diagnosis will come up with new, interesting, and maybe very positive ways of patients beginning to treat themselves. The power of the professions and the existing institutions may erode or be compromised. Do you see this as an inevitable shift of the treatment moving in the direction of the patients, or do you think that line will stay fairly well intact?
ET: Well, I wrote about the doctorless patient, but I never conceived that the treatment side would be picked up as a do-it-yourself story. There are many reasons for that. When the FDA reviews all these strategies, whether they are apps and ads to smartphones and other devices, they are very concerned—rightfully so—that erroneous information can lead to maltreatment. Having physician oversight of the data to be sure that there has not been some egregious malfunction is quite important. While there could be recommendations, for example if you have an app that enables you to examine your child's eardrum, it says there is an infection and maybe suggests consideration for antibiotics, and even gives a couple of choices that might be particularly effective. But still I do believe it is the doctor's domain to render what is the right treatment if any treatment is indeed indicated. I don't see do it yourself getting into treatment anywhere in the decades ahead. That is where we have to rely on judgment, knowledge, experience of physicians, and their appropriate guidance.
RW: It was interesting reading the 23andMe chapter in your book, about this tension between two philosophical camps out there. One says that the democratization of data and information are such that people should be able to choose whether they are using this information in new ways. They can make those choices, and it is not the job of authorities like the FDA and others to choose for them. Another says that people have to be protected against bad information or bad choices. You've spent time in the last 5 or 10 years immersed in this world of patient choice and patient empowerment. Where have you come down on that scale?
ET: I think in general as physicians we don't give enough respect to patients as to their capabilities. It has been a very paternalistic profession. That is how we were trained, and in the book I trace it back all the way to long before Hippocrates. So we need to let go there, but it is not for everyone. Many patients like having autonomy of physicians. That is fine. In the last couple of years, many large consumer surveys asked whether they would like to have full access to all their information. If people want to have their DNA sequenced, or their 23andMe profile, that is their right. It is their data and they should own it. Obviously that is far from where we are today, where doctors and hospitals own people's medical information, and that is not the right fit. Over the next several years, when people are generating terabytes of data, there will be an overriding amount of data that is not in the traditional electronic record. No health system today is capable of ingesting all these different sensor data, sequences and everything else. Right now the new data that people will be generating and want to have access to is homeless. We have to figure out where it will live and how it will be easily accessed and utilized to help preserve health.
RW: There is an obvious tension in having the data be accessible to everyone everywhere in an era where hackers seem to want access to it and seem to be pretty good at getting it. How do you think through that knotty problem?
ET: Well it is knotty; there is no question about that. There is a "secure versus cure" theme. When you ask individuals if they are willing to share their data, 80% even 90% say absolutely yes, as long as it is kept private and it is de-identified. We are terribly vulnerable, not just to hacking. When you type terms into a Google search or in an Amazon purchase, this information is being sold by data brokers. This will hold us back. There is this enormous opportunity if we protect the privacy of one's data and it is shared. For those who are willing—and the vast majority are—then these resources can be helpful for individuals with that same diagnosis to match up with others with similar demographics and like conditions to see the best treatments and outcomes. Instead of the way we deal with patients today, where everything is siloed and we don't learn from them to help other patients. We do it in very bizarre ways like publishing articles, but not in a way that is this open medicine concept, which is exciting to me. If we can get 1.3 billion people on Facebook, why not get millions of people on social media to improve health, if we can protect their privacy and prevent hacking. We could have an extraordinary medical resource to help promote better care, not for research though it could be for that too, but primarily for promoting better care.
RW: The role of the government in all of this is pretty complicated in terms of promoting the things that you talk about, but also potentially getting in the way of innovation if they are too heavy handed. How do you think the government is doing and what do you think it should do?
ET: It could put laws in effect. White papers have been published with no action but with many good concepts. We need to actualize the protection of medical privacy. HIPAA is terribly outdated. It needs to be rebooted from scratch. There have been some signs of light at FDA but much more is needed. When will we see legislation that cements the idea that people own their medical information? Not doctors and hospitals but the people own it, and they are the ones who call the shots as to how it is shared with doctors and hospitals. That legislation will not be coming soon, but I sure would like to see it. I think that is what we ultimately need.
RW: Alerts are a problem that we tend to think about mostly in health care institutions now, but may increasingly be an issue for patients as well. As soon as you're monitoring anything you create the capacity, maybe the obligation, to let people know when something seems amiss. Certainly in the hospital world, we recognize that we're over-alerting people and folks have begun tuning out even some important alerts. How do you think through that problem?
ET: The hospital example is horrendous, and if we cannot do better than that then this will go nowhere fast. I don't understand over the years how that never got cleaned up—where the alerts are nonstop and a complete nuisance. We've made amazing strides with the software in smartphones. If we get the right players in this mix, we have the capability to do it right and not have simulation of the farcical things that happen in a hospital because of the constant alerts. The signal-to-noise is not differentiated properly. We have to do better because the consumer will not tolerate it, and certainly doctors will not tolerate it if consumers are getting alerted and anxious, contacting their doctors incessantly.
RW: Most people believe the notes and documentation have deteriorated markedly as we've gone electronic. Do you have thoughts about how that can get better and what the documentation looks like in the future?
ET: I thought you did an excellent job in your book describing that mess. What I have tried to put forth is that during the encounter with the patient there should be no keyboard. There should be voice activation of natural language processing of the doctor and the patient, and that should be edited by both of them. There should be machine learning so that within a matter of x number of patients on the doctor's side it becomes quite easy to edit. It is amazing to me that that has not already happened. It is not that hard to do. But that would get rid of keyboards. It would be full text instead of these truncated non-English, non-useful way of keeping records today. I would like to see the conversation that goes between doctor and patient fully captured and processed the way it ought to be, but it has not happened yet.
RW: When a patient comes to see Dr. Topol in San Diego, what is his or her experience like, and how is that different than going to see the cardiologist down the road?
ET: We talk a little bit, of course, and then instead of checking a pulse, I will do a cardiogram on the smartphone. I don't use a stethoscope; I use a high-resolution ultrasound to examine the heart. I hardly ever have to order a full echocardiogram. We review the patient's data because almost all my patients have a smartphone and almost all have hypertension along with other chronic conditions that they can track like diabetes. If I have not received that data beforehand, I will review it during the visit to see how well their blood pressure is being monitored. What are the issues surrounding that? And what else? By the way, I share the ultrasound exam in real time with the patient and point out the findings so that instead of them being sent to an echo lab where they are not allowed to know what is found by the sonographer, I'm reviewing it with them. I try to avoid medications. I try to avoid supplements. Obviously I'm prescribing stuff, but I would guess at a much lesser rate than many colleagues, and a lot more apps than meds.
RW: You and I trained in the same program. Tell me if you were in charge of the training today, what should residents and students be learning that they are not currently learning?
ET: Residents are not being trained to be good telemedical doctors, which will become increasingly the norm of routine appointments instead of having people visit physically. Simulating for that encounter that is not face-to-face physically but face-to-face virtually. That is one thing they should learn. Training physicians should include using handheld ultrasound throughout their exam, and there is little purpose of using the old stethophone from 1816. Being facile with what apps are available in terms of not just the sensors but also the labs, the physical exam tools, all the things that could be useful for patients. Trainees, residents, and students should know about that as well. These are not in the typical curriculum of the 140 medical schools, but they hopefully someday will be integrated.
RW: I'll ask you a harder question in terms of the training—how would you train physicians to have the humility that is necessary to flip the classroom and have the patient be more in charge?
ET: In my contacts with students and residents they seem to understand already that classic autonomy is caving quickly, and they have no problem with that. Actually patients having their data all flowing freely to them, it is a natural thing. It is not like you have to train doctors. The patients will be training the doctors saying, "I have all this information, it is all processed, what do you think? What would you suggest I do about it?"
RW: Interesting. It sounds like your sense is the young generation today gets that power relationship and is comfortable with it. In some ways, we untrain them and we just have to stop doing that.
ET: I see this a lot, like the old directed consumer ads: "ask your doctor." Unfortunately these ads are still on TV, but they have profound effects. Because the consumer coming in stating "I want this drug" is very powerful, instead of "well here is my data for all these different metrics, and now what do you think I ought to do about it?" We will see a lot of that over time if the medical community allows it to happen.
RW: Talk about the reaction of the medical community to what you've been saying, not just the increased power of patients relative to physicians but also this movement toward electronic enablement of patients. Some of that is controversial. How do you feel the profession has treated you and has handled this?
ET: It is very mixed. It is in some ways concerning because some physicians are not willing to think about a different model of rendering health care. I can understand that resistance to some degree because we're talking about perhaps the most significant change in the history of the profession. At the same time, I see this as an inevitable path. What is concerning is that it is either "all technology, depersonalizing and no humanity, no human touch"—or it is all human touch. That type of dichotomy is absurd. You can promote these technologies that we've discussed and greatly enhance the intimacy of the physician–patient relationship. That unwillingness to accept that—"this is all iDoctor stuff"—and it is more of the same of treating the scan or the lab test rather than the patient. That is not what this is. This is the patient having their data and asking for the wisdom and the experience of a physician. I reject that notion, and obviously this whole line of reasoning is what the naysayers use to try to take this on.
RW: So you see a path of taking the best of both worlds. And you see the naysayers as tending to be hyperbolic, on one side or the other.
ET: They are basically saying well you're computerizing medicine and you're trying to make this into some type of techno-festival. These "techno-optimism to the extreme" comments have been used. I don't accept that. I say that we're leveraging technology to take medicine to a much higher plane. We're now giving patients empowerment that they never had. The term consumer-driven health care has been out there for a long time, but we did not have the tools to provide that. I do believe that is the biggest switch now and that is what sets this off. Someday this will take hold; it is just a matter of when.