Precision Policies: How Epidemiology Can Help Address Drug Overdose and Firearm Violence

The average lifespan in the United States is four years shorter than in other large, wealthy countries in part because of deaths due to substance abuse. It shouldn’t be that way, according to Dr. Magdalena Cerdá, DrPH, MPH ’99.

Cerdá and her research team spent a year compiling and comparing laws on harm reduction across nearly 500 municipalities. She was shocked to find less than 20% of U.S. counties have community-based naloxone services.

“This is just unacceptable when we know that naloxone works to reverse overdoses to prevent deaths,” said Cerdá, director of the Center for Opioid Epidemiology and Policy in the Department of Population Health at the New York University Grossman School of Medicine. “When more than 80,000 people [annually] are dying from an overdose, we should have naloxone everywhere.”

Cerdá was back on campus in May as part of Yale School of Public Health’s Distinguished Speaker series. In her current roie, she studies how social contexts and drug and health policies shape drug use and urban violence. By applying machine learning and novel modeling techniques in the context of social epidemiology, she and her team hope to develop evidence-based policies that improve population health.

Helping communities determine how best to use targeted interventions like naloxone to avert drug-related violence and deaths is a key focus of Cerdá’s work.

She has studied the Prescription Drug Monitoring Program (PDMP), which requires medical professionals to check patient databases before prescribing and dispensing opioid drugs. All 50 states have adopted PDMP to help limit opioid supply. But reductions in overdoses have varied, which Cerdá said shows that overarching laws and regulations are not always a one-size-fits-all solution. In the case of the PDMP, states with the highest decrease in overdose deaths were not those with stronger prescription opioid restrictions. Instead, the best results occurred in states with laws authorizing harm reduction services. These can include naloxone distribution, increasing the availability of fentanyl testing strips to help drug users determine the concentration of fentanyl they are about to consume, and syringe service programs that provide sterile needles and other tools to make drug use safer.

“The idea with harm reduction is to meet people who use drugs where they are and to give them the tools that they need to be able to live lives of health and dignity,” Cerdá said.

Another significant factor in the U.S.’s lagging life expectancy is firearm violence. The U.S. has the world’s highest rate of civilian gunownership, and one third of gun homicide perpetrators are under the influence of drugs or alcohol. Separately, people convicted of driving under the influence (DUI) are almost three times more likely to commit a violent crime than those without a conviction.

Using agent-based modeling techniques to simulate people’s interactions, Cerdá and her team studied potential interventions to reduce firearm violence. Their models showed that restricting gun purchases for people with prior DUI arrests or convictions resulted in only a 2% change in predicted gun violence. It was more effective to expand the pool of people restricted from buying firearms to people with risk factors such as misdemeanor or felony convictions. Also effective were blanket policies such as raising the price of firearms through an excise tax.

Cerdá and her colleagues have also studied two well-known community-based violence reduction efforts in Chicago, Becoming a Man (BAM) and Rapid Employment and Development Initiative (READI Chicago). These programs work to prevent firearm violence through providing youths mentoring, therapy, and employment support. Cerdá and her team modeled the potential effects of these interventions on participants, their networks, and their neighborhoods. The BAM program significantly lowered violence among individuals but had little effect on neighborhood shootings. The READI program was more effective in reducing neighborhood shootings. That suggests focusing efforts on highly connected individuals whose associations with gangs and other groups in the community can lead to broader population-level changes.

“In cases like these, novel agent-based modeling techniques can help us to predict not only what interventions are most valuable, but to whom they are most valuable, for bigger effects,” Cerdá said.

Cerdá says the data shows that bottom-up, population-level interventions can reduce both the harms of drug use and violence. Epidemiologists, then, must meet people where they are in their communities to devise interventions specific to populations.

Cerdá and her team hope to continue using advanced prediction modeling to tailor efforts to reduce firearm violence and opioid harms. “This is one of my passions,” Cerdá said. “I started this work during my doctoral dissertation and I've continued it throughout my career.”