Blake, an alumnus of the MD-MBA program at Geisel and Tuck, collaborated with Tuck professor Jim Smith on a paper modeling interventions that could eliminate Hepatitis C in people who inject drugs (PWID) in New Hampshire.
As a native of Bedford, NH, Drew Blake MED’20, T’20 grew up in the context of the long emergency known as the opioid crisis.
Hearing constantly about the high rates of addiction and overdoses, and their attendant societal impacts in New Hampshire, inspired Blake to matriculate at the Geisel School of Medicine at Dartmouth and work on issues of drug addiction and harm reduction in his home state. During his first two years at Geisel, Blake worked as the assistant to Governor Hassan’s advisor on addiction and behavioral health, and co-founded Project 439, New Hampshire’s first needle exchange program.
Shortly after he began the MBA program at Tuck, in 2018, Blake saw how a business education could equip him with important skills to combat public health problems. He was attending a talk at Geisel about eliminating the Hepatitis C virus (HCV) and noticed that the studies on the topic were using Excel to model outcomes in various hypothetical scenarios. Blake was immersed in Excel at the time, as a student in Professor Jim Smith’s core course, Decision Science (now Analytics), and he began wondering if he could use what he was learning to build a model for eliminating HCV in New Hampshire, where it mostly strikes people who inject drugs such as heroin and fentanyl. At the end of Fall term, he cautiously approached Smith to pitch him the idea. “I didn’t really know I had just come up to one of the foremost decision scientists in the country, whose wife studied HCV and whose daughter was in medical school,” Blake recalls. Smith, the Jack Byrne Distinguished Professor in Decision Science at Tuck, was happy to work on the project with Blake.
I didn’t really know I had just come up to one of the foremost decision scientists in the country, whose wife studied HCV and whose daughter was in medical school.
The result of their collaboration is a paper recently published in JAMA Network Open: “Modeling Hepatitis C Elimination Among People Who Inject Drugs in New Hampshire.” It’s the first study of HCV elimination in people who inject drugs (PWID) that examines the full range of interventions available to the public health community.
Chronic HCV, which affects 2.4 million people in the U.S., has been called a silent killer because it is easy to transmit and causes no symptoms for many years. Only after a decade or more of having the infection does it present in patients as liver cancer or cirrhosis of the liver, and by then patients are very ill and require aggressive treatment. The development of direct-acting antivirals has greatly improved the outcomes for people with HCV, but the virus continues to spread quickly in PWID, a population with low screening rates, low treatment uptake, and an ongoing risk of reinfection. Blake and Smith’s paper presents a mathematical model that evaluates how improvements in testing, treatment, and harm reduction services could impact the prevalence of HCV among PWID. “We used New Hampshire as an illustrative setting because it has high rates of injection drug use and, like many states, has underdeveloped infrastructure to track and treat HCV among PWID,” they write.
To model the three interventions in an integrated and nuanced way, Blake and Smith built a dynamic, compartmental differential equation that simulates the spread of HCV among PWID. One of the central challenges of the study is that it is difficult to find good data on PWID in New Hampshire. Estimates from survey data suggest the state has 8,000 people in this category, but the actual number may be somewhere between 5,000 and 10,000. And many of the parameters in the model have a similar level of uncertainty. To deal with this, Blake and Smith used a Monte Carlo simulation, a tool mentioned in the core Analytics course. “Since many of the inputs are uncertain, it’s hard to have confidence your prediction,” Smith says. “One way to gain confidence is to put a probability distribution around those uncertain inputs and randomly draw input values. Doing this, we observed that if you are aggressive about each intervention, you can nearly eliminate HCV in this population relatively quickly.”
New Hampshire is trying to expand needle exchanges, testing, and treatment. Now we need to identify where to put resources to get rid of bottlenecks in the process.
Creating a complex model and authoring a peer-reviewed paper is not what Blake expected to do when he started at Tuck. He found Decision Science to be one of the most intimidating courses, and Excel was like a foreign language to him. But he appreciated how the course helped him structure decisions and analyze data, so he took every other analytics course available, and did independent studies in the topic. Together, those experiences gave him enough competence to read a technical paper on HCV and “the foolishness to think we could recreate what they were doing,” he says.
“And eventually, we did,” Smith adds.
At the very least, Blake and Smith hope their paper sparks conversations about attacking HCV among PWID. The paper already has more than 700 views on JAMA’s website, and the co-authors presented it recently to infectious disease specialists at Dartmouth. “People at Dartmouth and across the state are trying to do exactly what we modeled,” Blake says. “We used these interventions with feedback from them. New Hampshire is trying to expand needle exchanges, testing, and treatment. Now we need to identify where to put resources to get rid of bottlenecks in the process.”