NIH All of Us program gearing up for 'precision engagement,' Eric Dishman says
It may seem overhyped at the moment. And its technical and clinical complexities are certainly way beyond the grasp of most hospitals. But precision medicine has enormous promise – more than we may yet even realize – for transforming the way care is delivered to patients.
Eric Dishman is working as hard as anyone to turn that promise into reality. More than once he has been called the "Face of Precision Medicine." As director of the All of Us Research Program at the National Institutes of Health, he is leading the cohort creation for one of the biggest and most ambitious research initiatives ever: the Precision Medicine Initiative launched by President Obama in 2015.
Dishman's goal is to enlist 1 million Americans – "from all walks of life and parts of the country" – as a representative group that can help researchers draw new insights into the ways genomics vary across diverse patient populations.
That's a tall order. But he says he's confident that he and his NIH colleagues will be able to overcome substantial clinical, technological, social and political challenges to making it happen.
If anyone has shown the power of optimism, it's Dishman, whose life was saved five years ago – after more than two decades undergoing costly but ineffectual cancer treatments – thanks to insights derived from a last-ditch whole-genome sequencing. We spoke with him in Boston earlier this month at the Healthcare IT News Precision Medicine Summit about the challenges and opportunities ahead.
Q. The last time I spoke with you was in 2012, back when you were an Intel Fellow, and we talked mainly about patient engagement. What have you learned about it since then, and how will you put that to work on a massive patient engagement project such as this?
A. We're going to have to do precision engagement if we're going to do precision health. Because understanding and trust varies so greatly for even medical expertise or a government-funded study, so we need to understand the unique needs of different parts of the population and address those.
We did some user-centered design work right away. We started doing not only focus groups but qualitative interviews and some survey work and we have this persona segmentation that cuts across race and ethnicity and gender and age, and it's really about skepticism about data and data privacy and research and those kinds of things.
When we did the continuum, there were some folks who said, "These are going to be the most impossible people to get!" And I said, "That's exactly who we need to get." Because these are people who have been left behind by traditional biomedical research. We've known from the beginning that we can't think that it's people like us who we need to get to come into the program.
I hired this amazing woman, a physician named Dara Richardson-Heron, and she was CEO of the YWCA. She's now our chief engagement officer. She has huge experience dealing with how you hear and act upon community concerns, all across the country. She's coming in to help us figure out our strategy.
One of the things we're doing – and we've already put out a funding announcement, and could soon announce the first few of what could be hundreds of awards over time – is community partner awards: think not-for-profits, churches, senior centers, having them be ambassadors for the study locally.
A lot of the feedback has been to use local radio, and local, trusted, intermediaries, to help explain what precision medicine is, explain the implications and explain the fact that they and people like them haven't been included in traditional biomedical research, so you haven't been included in the targeted cures that are being worked on. We have a micro-targeting massive engagement strategy on steroids.
Q. When you talk about getting a cohort "from every walk of life," does that just mean casting a wide net? Or do you have specific, detailed demographics you're hoping to account for?
A. For our UBR – underrepresented in biomedical research – we want 50 percent of the million people to be those who are racial and ethnic minorities, so we will over-recruit to make sure one in two people will come from that world. And we want three out of the four people to be UBR. That means women, lower economic status, rural areas.
When you start to add that up, it's shocking: How come no one ever addressed this before? But it's also shocking because it's just such an undertaking. I'm trying to bring people in who have expertise in targeted marketing from industry: sophisticated analytics to let us know where our methods or medium or message is failing, and then how do we swarm to try different things to change that.
We're trying to learn from people who have done successful recruitment of diverse populations, and skeptical populations before. But even then they've never done it at anything like this scale. And they might have done so by having two or three volunteers to bring one person in – if I do that times a million, I'm not sure I'll be able to pull this off. I've started to wonder if we might need to hire another million people to help be buddies and translators to help pull in the million-person cohort.
Q. What does IT need to do better? Are there evolutions that need to happen to be prepared for precision medicine?
A. I think on all of the technology fronts, we're close but not quite there. The first way I think of technology making a big difference is moving data collection to the edge of the network, if you will: having mobile devices, having smartphones, having wearables that people are buying and using. Being able to collect data types that will help us not only be able to understand nutrition, but can be a data type that feeds understanding about where people are physically in the world, and the impact of the environment on them. Understand their patterns of daily life and deviations from those patterns.
Then when you combine that with medical data, and with genomic data, we're going to run different analyses on the blood and urine we're getting out of version one of the protocol that would look at chemical exposures and those types of things. You start to get a rich picture.
So in terms of where those sensor technologies are, it's pretty amazing what we can do today. But I think the problem is going to be as we start to look at the runs of different consumer products, even though it's the same product, different manufacturing runs have different levels of calibration, different sensor accuracy. It's like, wow, how do we collect for a quality data sample? Will statistics make up for the noise that occurs in these consumer devices? And what's next? What can we count on, even three years out, five years or 10 years out that's going to be almost free and widely available in consumer devices, so we don't spend our efforts making our own wearables, knowing the consumer market will address it.
Q. What about clinical technology, specifically EHRs? Many say they're still not quite equipped to handle the voluminous and specialized genomics data that fuel precision medicine.
A. If you think about EHR data as a resource for research, first of all, it was never designed, the data wasn't captured in a way the researchers would want it to be. You can make up for that, in terms of large numbers and comparison. People keep telling me, "Eric, once you guys even get EHR data of 150,000 people, there's analyses you can run to help discover digital phenotypes for diseases. And people are sharing their own digital phenotypes with us, a more accurate assessment of what people's true health condition is. It's there in the EHR, we've just never had the analytics tools to help discover it.
At the same time, the interoperability is still key. And I'm hopeful there. We've done this program Sync for Science, with six or seven big EHR vendors and the Office of the National Coordinator. And they've all been incredibly helpful using this FHIR interface to make it sort of like a blue-button for EHR data: "I, as a participant, want to share my EHR data with the All of Us Research Study."
So we're going to pilot that in August and September at 14 sites. And in all the tests so far it's doing really well. So as those roll out, and the updates continue from EHR vendors across the country, and small hospitals and clinics get those updates, that technology is in there for a lot of other people to exploit and develop a lot of other apps and services on top of. I think it will take a couple more years for that to scale out as part as normal EHR enterprise updates. But I think that's pretty exciting for us.
Now, of course, what that's going to give you is the EHR data for the place you happened to be treated at last. So we're also looking at some of the EHR aggregation companies – I did one myself and it was like, holy cow, they pulled up data from clinics and hospitals I didn't even remember I went to. If you're trying to pull a longitudinal record together, it's probably a combination of getting your current EHR data, and you may need to use one of these aggregators, pulling from clinical and claims and lab and all of that.
Q. And how about humans? Some physicians are skeptical of precision medicine. Some are maybe just set in their ways and don't want to change how they've done things for years?
A. Our first challenge on the clinical side, or the clinician side – or the provider side, I should say, more broadly – we have an outreach program to them: If their patient walks in the door and says, "Hey, I want to be part of this All of Us Research Program," a) have they heard of the All of Us Research Program? And, b) hopefully they'll have a good impression of us and will say, "Yes, you should check them out." We're reaching out to professional societies – not just doctors, but nurses and allied health workers and a wide range of folks to get that message out.
There's skepticism about the latest, greatest, newfangled thing, and with precision medicine it's well-earned. The number of hype cycles they've all been through, and eventually there was value to them, but never when it was first talked about. This is probably true for precision medicine from most of them. The advanced oncology places, some of the precision medicine work on diabetes, some on Alzheimer's – you're starting to see progress.
When I was at Intel we were doing ethnographic work of top pioneering clinics in different fields on the edge of precision medicine. And you'd find variability just within that practice. The physician in the room on the left was into the genomics and precision medicine and was offering lifesaving cures to their patients. And the doctor in the room right next to them was like, "Ehhh, I can't learn all this, I don't know what whole genome sequencing is.” Or they're just skeptical of it. So even at some of the top tier places around the world, the variability is huge.
So we have a massive education job to do, and this cannot just be a few CME credits here and there. I think we're getting out in front of it, in terms of thinking what med school needs to be for the new physicians that are being trained. But how do we train the workforce that's out there that this is real? And at the same time, not overhype it? I think that's one of our challenges.
Q. Will we see a day where precision medicine is not just the bailiwick of academic medical centers, and its techniques have become more commonly used at smaller hospitals? Or are we always going to have to depend on the large organizations to help those who don't have the resources or expertise?
A. I think there's two answers to this question of who gets to play in the biomedical sandbox. On the research side, I suspect you're always going to have academic medical centers being at the forefront of new technologies and new tools and new research. And you sort of want that, because they're set up to build the clinical evidence that there's a "there" there. Otherwise, don't bother the rest of the healthcare ecosystem with a bunch of snake oil.
At the same time, the cost curves on things like sequencing and pretty sophisticated technology that can be put into a consumer wearable are so low that it's allowing a lot of other people to play in a way that haven't in the past. And if you combine that with at least some percentage – even if only 20 percent of consumers start pushing saying they want precision treatment, that's enough to disrupt the apple cart and get people to start saying, "We have to start figuring out ways to deal with this."
I think back to the days when the Internet was just starting out. I was back at Intel and we were studying family care and physicians all over the country. They were just starting to see these websites that would pop up and say, "Hey, there's this great new cure," and they were like, "This internet thing is horrible – now all my patients want to do is talk about this thing they saw on a website."
Of course, over time, those curmudgeons said, "We have to learn how to use this internet technology to do scheduling and have our own controlled portal and all of that." Now it's commonplace, even in small practices.
I think the same thing will happen with precision medicine: There will be enough consumer demand – and enough "there" there in terms of them being real cures in cancer and beyond, that they're going to have to pay attention to it. And they'll have to move from skepticism to, how do you actually bring it into practice. Our program is going to last 60, 70, 80 year – I can't even think out that far. By that point, precision primary care will probably be the name of the game.
Q. President Trump's budget wants to cut NIH's budget by almost $6 billion. Are you confident you can manage this project, even if that happens?
A. The Congressional response in terms of support for all of NIH, especially in terms of support for what people are now calling the "innovation package," the 21st Century Cures – the Cancer Moonshot, our program, which in the bill is still called the PMI Cohort Program – I feel quite confident. We wouldn't have started the beta launch if I wasn't confident we had the funding to make this a useful, meaningful study.
The original plan was for our budget to eventually ramp to about $400 million a year. And now that we're into it, I think we really need it. But if we don't get to that budget, between our base budget and 21st Century Cures, it's just going to have to be a trade-off discussion. Will it just take us longer to get to a million people? Or will we get to a million at the same speed but we won't have the diversity of data right out of the gate? It's just all a matter of timing of when we get there. Right now I feel like we're on a good course to get to a million people in about four, four-and-a-half years, with a really rich data set.
I think as early as a year from October we're going to have close to 100,000 people in it. Then it begins to be a really useful resource – even, for certain kinds of questions, if you have 10,000. By the end of this summer it'll be a useful resource.
And I think success will breed success. As people start using it and saying, "Gosh, I wish you had this richer data, and we start evaluating it, I could imagine making the case to Congress in future years that, "Hey, we want funding to go out and do these particular assays, and here's what that would be enable."
My other hope is a public-private partnership. We've already had people coming from the pharma industry, the medical device industry and others saying, "Look, we realize there are some huge big-ticket items you need to do, and everything is times a million people." Competitors are coming together to talk about putting money into the program so we can accelerate whole-genome sequencing or the use of some wearable or something, in a precompetitive way, with all the data open to the public.