Joshua Barocas, MD, an infectious disease physician at Boston Medical Center, led a years-long study published in 2018 that showed the number of people with opioid use disorder in Massachusetts was not only growing, but was vastly underestimated because many people with OUD were unknown to the healthcare system.
Now, Barocas is a 2020 winner of the National Institute on Drug Abuse (NIDA) Avenir Award, which recognizes innovative early-stage researchers in HIV and addiction science. His project will develop an improved surveillance network to more accurately gauge the scope of local substance use and inform clinical care and policy decisions.
Barocas spoke with HealthCity about the need for better drug surveillance and the importance of community-engaged research in expanding trust, reach, and impact.
HealthCity: What problem are you aiming to solve with this project?
Joshua Barocas, MD: We have excellent surveillance systems for a number of diseases and health issues, but our surveillance system for drug use is not working in the United States. The overdose crisis and the growing number of outbreaks of HIV among people who use drugs are red flags.
“We have excellent surveillance systems for a number of diseases and health issues, but our surveillance system for drug use is not working”
We rely on two flawed assumptions in our approach to our substance use surveillance system. One is that we tend to count people just by doing surveys or looking in administrative claims databases. The highest-risk populations are those people we don’t even know are at risk. Those are the people we desperately need to find. The problem with that is you don’t know who you’re missing. The second assumption is that we think we understand what puts people at risk for HIV, for overdose, or for loss to follow-up. We don’t ask people how risk factors might be changing over time.
We don’t know what we don’t know. How can we possibly be allocating resources adequately when we don’t even know the size of our populations? How can we target resources when we don’t even know how risk factors may differ in different communities? What I want out of this project is literally a complete overhaul the public health surveillance system for drug use disorders. We need to make sure that our most marginalized populations — people who are incarcerated, people who are unstably housed or experiencing homelessness, and other people who might be hidden in plain sight — are not continually forgotten.
HC: What spurred this work?
JB: I’m a denominators person. All of this grew out of a project that I did with investigators from Boston Medical Center, Boston University, and the Massachusetts Department of Public Health in which we tried to figure out the prevalence of opioid use disorder — the denominator — in Massachusetts. We did what’s called a capture/recapture analysis, effectively borrowing from methodology in the fish and wildlife world. If you’re trying to estimate the size of a population of fish in a lake, you have to do some sampling. We took those principles and put them into a statistically rich analysis.
We were able to use the linked, longitudinal database that we are lucky to have here in Massachusetts and calculate the number of people with opioid use disorder who had at least one interaction with the healthcare system where they were recognized as having opioid use disorder. We then applied statistical models and were able to estimate the unknown population of people with OUD. What we found was eye-opening. Right around 50% of people who have opioid use disorder in Massachusetts were estimated to be unknown, meaning that they have not been identified in the Massachusetts healthcare system in a given year. So we’re probably undershooting funding and resources for our interventions by at least 50%.
HC: How can drug surveillance be done in a way that reaches more people who are being missed?
JB: We need to stop assuming, as researchers, that we know the right questions to ask without community input. This isn’t meant to sound like a big brother program — it’s the opposite. We need community members to be involved so that we can attempt to quantify the problems that they are experiencing. And we need to be able to disseminate these findings and make them digestible for the general public.
When we make the community an equal partner that has equal voice in how we approach research and clinical care, then we’re more likely to reach more people. It starts to instill more trust in the system. We’re more likely to engage people, get them into care, and keep them in care.
HC: The project name is “Development of a Novel Community-Based, High-Performance Surveillance Network for Drug Use.” Can you define a surveillance network?
JB: I use the word network instead of system, and I’ll explain why. A surveillance system is something that’s usually set up with a top-down approach from folks at high places in the government and the ivory towers of academia when we’re trying to get a handle on a health issue. We say we need to understand the prevalence of obesity in the U.S., for example, so we deploy surveys in X number of communities, survey a pre-determined number of people, and define the population of obesity. In the end, we throw these numbers into a report or on a website. We tell people “You can look at them, absorb them, do what you want with them — but this is how it’s going to be.”
When I say a surveillance network, it’s community based. People know their communities better than I do. What might be working in community A might be applicable to community Z, but not so much to all of the communities in between. Of course, there are likely going to be risk factors that are more universal, but we rarely try to understand the size of those at-risk populations. For instance, people who have experienced incarceration are likely some of the highest-risk people for overdose — across the board. If that’s the case, then what we really need to know is how big that population is. Other factors — risk or protective — are less explored. For example, in one of our communities, access to religious clergy might be a protective factor. We might only find this out using discussions with community members. Now, it’s our responsibility to go find out (using data) if this is something that might be protective more widely and how big the population of people with opioid use disorder with access to clergy might be. Then we implement interventions that are appropriately scaled to this population.
HC: What is the plan for engaging communities and exchanging this information?
JB: Overall, I hope that this becomes a bidirectional process whereby we listen to what community members are telling us and, together, frame the research questions and interventions, and then make the findings accessible to the community.
This is not an easy task, and I recognize it will take time to build trust. We are going to try to engage people from all aspects of the community. That includes government officials and public health officials, people who work for community organizations, harm reduction workers, people with lived experience, clergy members, business owners, and others who literally have boots on the ground. This is a project that is meant to serve them. I hope to bring people together so that they feel empowered to use this system.
The technical aspects of the project — machine learning, advanced statistical methods — will all be driven by what we learn through the community engagement process. Our models will reflect what is important to the community. Our results will, hopefully, be used to effectively target interventions to the highest-risk parts of the population that we wouldn’t otherwise have known about.
HC: This sounds difficult to carry out. How do we start to solve this problem if it requires many different interventions specialized to the communities themselves?
JB: It doesn’t mean that we have to create new interventions. What I aim to do is give communities the tools to say, “These are the risk factors and protective factors for HIV or overdose in our community. This is the actual size of those populations at risk. From the toolbox of interventions that already exists, which ones should we deploy, and what’s the scope needed?”
If we can start to understand who’s at risk and the size of that population, then I’m hoping we end up with a more equitable distribution of resources and the ability of communities to then implement or adapt existing interventions that will work for their community. I’m hoping it will help decisionmakers and policymakers allocate resources appropriately to scale.
HC: Besides improved surveillance accuracy and resource allocation now, what are your hopes for the project’s impact in the long run?
JB: There are a couple of things that I’m hoping for. One is that this will go beyond Massachusetts and the process will be translatable to different counties or cities or states. I hope that this type of process replaces the outdated system we use nationally. From a big data standpoint, I think one of the biggest pushes is to help different states and locales understand the power in having linked datasets like we have here in Massachusetts, and what that can do for resource allocation.
The nice thing about this Avenir award is that it’s high risk, high reward; the goal is that you do something transformative, something that actually helps move the needle. My ultimate goal is not to have a new paper or an interesting manuscript in a high-impact journal, but that I’m part of a paradigm shift from this top-down approach to more of a leveled-out, community-engaged approach to how we conduct research.
This interview has been edited and condensed.
