In this month's "Discussing Data Science" episode, I talk with Josh Lee, Professor at NYU in Population Health and Medicine and an NIH-funded researcher conducting clinical trials on opiate and alcohol addiction. He is also the principal author for the opioid use disorder (OUD) treatment map, which is the most comprehensive resource for information about the evidence behind treatments for opioid use disorder in the world.
You can watch the video below or on Youtube. But if you'd prefer to read, keep scrolling. The complete transcript (edited for length and clarity) is below.
Spencer Hey (SH): Hello, my name is Spencer Hey and I'm the co-founder and chief science officer at Prism.Bio and this is Discussing Data Science.
My guest today is Dr. Joshua Lee, a professor at NYU in Population Health and Medicine. Josh is an NIH-funded researcher conducting clinical trials on opiate and alcohol addiction. He's also the principal author for the Opioid Use Disorder Treatment Map which is the most comprehensive resource for information about the evidence behind treatments for opioid use disorder in the world. This is hosted on Prism’s platform right now.
So, Josh, thank you so much for joining me today.
Josh Lee (JL): Thanks, Spencer. Great to be here.
SH: Why don't we dive right into the opioid map. I'd love to start with a little bit of the history. Prism got involved at the end of the project, but it'd be great if you could talk about how the project came to be. Why did you decide to undertake this massive effort of mapping the evidence behind treatments for opioid use disorder?
JL: It was really an opportunity that kind of fell in our lap. Although I am not a library scientist by training, what we’ve done here is a kind of library science project. We’ve visualize individual studies in order to show people how many papers came out, for example, about methadone that were randomized trials, each year, for the last 10 years.
So that kind of bibliographic insight is of interest to me because I'm always in that literature. That's my industry. That's what my team and I are doing at my institution.
And actually, before I went to medical school, my degree included a major in art history. This kind of bibliographic detective work related to what I did. I just got used to running up and down the library looking for obscure articles—like, “Wait a minute you said this in 1970 but you're referring to something in 1965 and that guy didn't actually say that”. So that bibliography and citation detective work I've always just personally kind of geeked out on and enjoyed.
But the overall project really came from a funder. Spencer, you and I were both working with The Laura and John Arnold Foundation on different stuff. One of the project officers there, Julie, was basically shopped it to us saying, “What if we gave you a little bit more funding for a new project?” Their perspective, as a grant giver working in the opiate treatment and prevention space, was always struggling to figure out what was really needed, what next, what's the holes to fill. Or in terms of policy or applied science: How should we treat people for opioid use disorder in our communities?
That's how the idea started. Julie came to us with the idea of mapping the existing evidence. We liked the idea. I was able to staff it with a postdoc or a a research coordinator type person for a year—since that’s about what the budget would allow. In terms of the size of these grants, it was a small one, but it gave us enough to get started.
Since then, we don’t work with Arnold anymore, but the project lives on because we've got it to a point where we can maintain it with some volunteer work and some internal funding (of course we're interested in more funding down the road if that becomes possible). But the map has become a thing that we can maintain within our in-house resources and we hope to keep it going as a living resource.
The goal overall for the project was to take this new methodology of evidence maps and apply it to opiate use disorder treatment, science implementation, and eventually policy. So we want to keep it focused on what we do that's practical and effective to prevent overdose, to save lives, to reduce heroin and Fentanyl and other opiate use, to reduce HIV infection within people using drugs or or opiates. That's what we're doing day to day in Bellevue Hospital where I work and where I treat patients. The map is intended to show the relevant universe of studies that can speak to these issues.
Fundamentally, we want it to be practical, something that NIDA or a state government official or somebody new to the research or someone much deeper on one end of the neighborhood versus the other could use. We want the map to show them what is functionally applicable to the real people in the present day, in terms of their condition and what kind of treatment they might consider. It's not about epidemiology or neuroscience. It's all about the treatment of opiate use disorder in human beings.
In other words: We wanted to make a tool to inform a primary care doctor's view of opiate use disorder and what we should do about it. That's my background too. I'm a primary care doctor. So we wanted to make a tool that was accessible.
And that’s why we wanted to partner with Prism for this, since you all can do things with data visualization that my team can't. Without our collaboration with Prism, the map would not be what it is today. We were interested in buying Prism’s creativity, knowledge base, and the ability to make the evidence map a usable, publicly accessible product.
Key Insights from the OUD Evidence Map
SH: When you look at the evidence map as it is today, what are the top-line insights that you’d like users to take away from it?
JL: The map today does nicely reflect what we do day-to-day, how we train our doctors and nurses and counselors to work with opioid use disorder. The map suggests that medications are extremely important to our treatment paradigms. It suggests that we are spending less time and money on non-pharmacologic or counseling or behavioral interventions. That’s not to say that non-pharmacologic interventions don't work or don't have a place, but you can quickly see the difference in research volume in the map.
Again, it’s important to point out that the map is not pointing you to the right answer (e.g., about the best treatment). The map is simply cataloging studies on a question. This, of course, has an implication as to treatment effect, and it can still inform clinical algorithms, since the most populous parts of the map indicate where we've done the most studies. Or you can filter to see where we have the highest quality studies or where the most patients have been studied.
But as the map shows, much of the extant research is clustered around medications like methadone, buprenorphine, and naltrexone. Not surprisingly, these are the three classes of FDA-approved medications for OUD in the United States.
However, the other fun part about evidence maps is the gaps. If you wanted to really look for new breakthrough areas, the gaps are where you should go. And we just know from our current practice, from our patients, and from the epidemiology that the landscape of OUD is changing all the time. For example: Does anything work in the age of fentanyl? You can't find any paper in that map that really isolates reduction in fentanyl use as opposed to other opiates or heroin use in general.
The map makes this gap very clear. So maybe we should start talking more about people smoking fentanyl in Portland, Oregon, just to quote a recent New York Times photo essay. The cartels are now pressing fentanyl into pills and people are smoking it. We need to be studying that—and using the filters with the map you can quickly see that there simply isn’t much research out there. So if I’m a research funder, that gap can guide my decisions.
Another big takeaway from the map is research on special populations. For instance, homeless individuals versus general population individuals. That's an important and, unfortunately, growing segment of the population. It reflects the lack of affordable housing throughout the country, which has nothing to do with opioid use disorder, except that it does because a lot of folks are wound up in a cluster of problems that can lead to (or stem from) addiction and opioid disorder.
So maybe we need more studies looking at homeless veterans who have opioid use disorder and are at super-high risk for overdose. We know from the map that there isn’t much out there. If you wanted to make a career as a as a young gun, that might be something to think about.
SH: That’s one of the most powerful things, I think, about evidence maps as a general approach—the way that it can sharpen your thinking around what you know and what you don’t know. It’s not that you could never figure out the big gaps in our knowledge from just doing literature searches and reading a stack of papers, but the structure that’s required by the visualization just makes those gaps immediately apparent.
JL: That’s correct. And it’s also important to contrast the evidence mapping approach with the more traditional systematic review and meta-analytic methods. A Cochrane review, for example, is typically limited to a pretty narrow question like, “Does aspirin prevent heart attacks in people that have already had a heart attack and are leaving the hospital?” Knowing that answer to that question doesn’t tell me anything about whether I should—as a 50 year old without cardiovascular disease—take aspirin every day. To answer that, I need another Cochrane review.
In the universe of opioid use disorder I would estimate that there’s probably 20 or 30 Cochrane reviews. So at best, that’s 20 or 30 questions answered. That’s a drop in the bucket of all the important questions that I need answered as a clinician—i.e., what works best for this particular patient.
The map speaks to exactly this deficit of knowledge. Using the filters, it can help to answer as many questions as there are combinations of filter parameters. Since we did a lot of coding of the data, a user can filter by study design, funder, number of people, intervention, outcome, geography. Those are all different ways that a user can slice and dice these papers so that they can quickly find, say, all the methadone studies in the 1970s from South America (of which there probably aren’t any to also make a point also about regional disparities in research).
But that kind of scope and flexibility is part of the power of the map. Again, it doesn’t tell you the answer, but it immediately tells you if there is research on the question. And if there is research, you can immediately access it.
I mean, we can do the math. There's probably tens of thousands or maybe even a million different combinations in that one map, simply because we chose a lot of filters and you can then mix and match intervention outcome and type of type of filter. This empowers you to quickly do your own systematic review.
By contrast, PubMed has this feature called “related articles” that you can use to get a general sense of what’s been published on a topic. It's always spotty and a little random (I think on the back end it's working through MeSH terms). It’s similar to what we’ve done with the map, but the map is just so much more efficient. You now already have all the opioid use disorder literature sitting there. You don’t need to ask PubMed over and over to do related article searches for different questions. That’s a big part of what I see as the key innovation and power of the map that that you can't get from existing search engines right now.
Next Steps for the OUD Evidence Map
SH: I know you've touched on this already, but let’s circle back to what you see as the next steps for the evidence map. You mentioned that there were particular gaps uncovered. You mentioned policy research. Where do you want go next?
JL: My own perspective on the opiate epidemic is: I can go back to clinic and write more buprenorphine prescriptions but it hasn't quite kept up with fentanyl. Fentanyl does seem to be isolated to North America, which suggests that it has its roots as a trade or economic issue. But I’d like to better understand: Why do we have all this fentanyl? How do we go back to abusable but safer forms of illicit opioids? To me the biggest problem that we need to solve.
So that’s where I’d like to take the map. Going back to that example from Portland (versus New York, say): The drug supply is different between the two regions. Fentanyl is packaged in different ways and used through different routes. We want to keep up with that kind of change or difference, and that has policy implications. It’s not just about the treatment of the individual patient. It's more about these macro factors that need to be sorted out or improved to get kind of macro change we want to see—to have that U.S opiate curve flatten and come back down.
Stepping back: I don't know if the evidence map is the answer or what part exactly it will play. But I see fentanyl as the most pressing problem in my field and it's not something I, as an individual doctor or prescriber or scientist, am in great position to like figure out. But that's what I want to solve.
The policy tab for the evidence map is not up there now, but it will be. We’re working on it. And when that is ready, I believe it will be a great way of looking at what we do about opioid disorder— moving away from an individual getting treatment at a certain place, in certain modalities, which is kind of how the map works now. The policy map will tell us more about what can we do for communities or entire zip codes, counties, states. It is mostly U.S. focused, but of course you could apply those policy insights from one part of the country to another or from one country to another.
What are the biggest data gaps in the field?
SH: Why don't we transition now to our three questions that we ask everybody.
First question: What do you see as the biggest data gaps in your field?
JL: This is both in the map, and something that I'm just telling you because I do this for a living: We have great treatments for opioid disorder now. We have trials that show treatment A (call it) is way better than anything else. But if nobody can stay on treatment A, then it's not very effective. And we're finding that retention on these core drugs—buprenorphine, methadone, naltrexone—is a challenge. If you can stay on them, you definitely do better.
But in fact, retention is not great. So how do I improve your chances of staying on our effective treatments over the long term? The answer there is not another drug. The drug itself has done all it can. What’s needed is something psychosocial or policy; something outside the clinic, in your neighborhood, your household, your bank account balance.
It used to be that the clear answer was to get people health insurance. And thankfully, in the United States we now have pretty high rates of insurance. Other countries probably have even better rates, but since 2008 and ObamaCare, the U.S. has made vast strides in getting more people with opioid use disorder insured—although still not enough are and we can still improve.
But that used to be the answer: How do you how do you boost treatment? You make it more affordable and accessible. Don't make it a pocketbook issue.
Even if you do that though, in a state like New York where Medicaid rates are like 97 percent, most of the patients I'm treating have no problem getting health insurance and getting these medications paid. The problem is that we can't keep people on them long enough and we're still losing people to drop out, eventual overdose, re-incarceration, in and out of the hospital, never really stabilizing. And then five years later, they've died of an infection.
There's currently so much room for improvement in terms of optimizing what we think should be guaranteed outcomes and they're not.
What excites you the most about the future of research?
SH: Question 2: What excites you most about the future of research?
JL: The great people.
If you fund it, they will come. The budgets in addiction science have been pretty good and it has attracted a new generation. We see it in our Addiction Medicine Fellowship, which is not all researchers, but through this program we see people that want to do this for a living. There's now a lot of smart people that are younger than me, that like the kind of stuff that we've been doing, and can hopefully take over and run with it for the next 50 years.
You just want to see that in the field if you care about it. It's so gratifying to turn over the wheel to somebody who has, you know at this point in your life, more energy and more ambition, is probably smarter than you. That's a lot of fun.
Of course, there's still not enough people and it's all hands on deck. We always need more help, more docs, and more researchers. But I do think that now, as opposed to when I went to med school: If you talk to med students now, they're more focused on the social determinants of health. That emphasis in training is just more common now.
The opioid epidemic has also done a number on younger generations in a way that it didn’t quite affect me and Gen X. Of course, people up and down the age ladder have had problems with opiates, but if you were in high school in the 90s and 2000s—that was a lot tougher, opiate wise in terms of what you would see in your neighborhood and community. It affected people you knew.
Those younger people are medical professionals now and they have a passion for this that was new to me when I got into the work. I don't have as deep or personal a connection to the opioid epidemic, although I've been treating it for years, because it really wasn't a thing for me in the late 80s in high school, and it was not a thing for my immediate family.
Not to say that we're free of problems or addiction. But the opioid epidemic specifically was not something that touched me. It really is different though with people born later. Then as they become trainees or new attending physicians in our health system—they are just ready to rock on this issue in a way that my generation was kind of caught us off guard. So that's fun and encouraging to see.
Wave the magic wand...
SH: Question 3: If you could wave a magic wand, what would you change about research, about the industry, about medicine? What leaps to your mind?
JL: On a personal basis, I would like to magically have written and published all my papers that have stacked up.
But back to like the public health wise… Can we just not have fentanyl?
SH: Right! I was going to say after this conversation: How about no opioid epidemic anymore?
JL: That's actually not the hope. People use drugs, alcohol, chemicals. Some are good, some are bad, some are more or less harmful.
I'm not a teetotaler. I don't think it's a reasonable expectation that humans won't use mind altering stuff or mood altering or or whatever to kind of tweak their daily life. That's fine. There's no reason to expect that would change because we've been doing it a long time. But is the only way to use opiates today in my county fentanyl? That just seems absurd.
We know that's the case. If you want to get your fix, or you need to use opioids because you're sick or you're just feeling like that's what you should do today, your most likely option is something that like easily could kill you. So you go use a bag right now. You're not quite sure what you're doing. You don't do it absolutely safely. You do it by yourself. That's why the overdose epidemic is what it is—and it has gotten so much worse year to year.
I don't know what I'm proposing here… Percocet vending machines? A lot of people died thanks to Percocet, so it's not all rosy when you think of the early days of the opiate epidemic either. But is there not some other way to manage illegal drugs and opiates, and in particular, fentanyl? Because if there isn't, then there's just no end to this until you've killed every person that is curious about opiates.
It does happen that epidemics end because they burn out in this way. We're not all smoking crack cocaine these days. That was actually a Newsweek headline from 1992, claiming that it would never end and that it would decimate entire cities. But that’s not what happened. Crack certainly had its harms, but the next generation wasn't as interested in it, in part because they were watching all those harms accrue in their neighborhood and their parents.
So maybe we just need more time and enough people to die then there's no more new users. That is one way to look at drug epidemics… But it's just so out of hand right now in terms of how dangerous is the drug supply.
I'd definitely be interested in what the DEA, FBI, NIDA, SAMHSA, President Biden, the governors of red states, blue states… Do you have any new ideas besides just trying to arrest Mexicans and offer people Narcan? Narcan is great and all, but you know it's not gonna necessarily prevent the next overdose.
SH: That is a very interesting magic wand answer, I have to say.
JL: Yeah, we need a safer drug supply. That's what I want. I don't want people to stop using drugs. I want people to do what they need to do. But right now it's just too easy for people to make mistakes with fentanyl.
SH: If feels rough to end on that note, but you know super interesting, super important. So I appreciate it.
JL: Well, yeah, I mean look: Our treatments work and they're good. So if you do have a fentanyl problem we can help you and that's great. That has been the pillar of my career. I found that I can actually do something about alcohol, smoking, and opiates. That is most of what is killing people addiction-wise in a country like ours.
SH: Thank you again, Josh, for making the time to talk. Always a pleasure.
JL: Thanks for having me.