There may be no phrase in the English language that makes the eyes glaze over faster than “public health data infrastructure.” But any big effort to tackle urgent health problems requires data, and lots of it. A recent multi-year study from the National Institutes of Health tried to find out if county health departments could make progress against America’s horrifying opioid epidemic if they had much more robust data — the kind of information that would help them see the factors contributing to their local addiction problem. The bad news: Overall, the $350 million federal effort, called HEALing (Help End Addiction Longterm) Communities — which aimed to drive down opioid deaths by 40 percent — failed to significantly reduce fatal drug overdoses. An analysis of the study found that it was fighting significant headwinds because of the pandemic and a rise in illicit fentanyl use. But there were signs of hope in some counties, like Ohio’s Toledo-Lucas County. (Read the full story of what happened, and what officials want to do next, in The Fifty.) In Lucas County, federal funds paid for iPads to help collect and share data and offer videos to train people to use the opioid overdose reversal drug naloxone. It bought a mobile van for educational outreach and naloxone distribution. And it came with access to a staff who coordinated and analyzed data, designed interventions and helped with marketing. The data provided a clearer view of Lucas County’s drug problem, showing which ZIP codes and demographics were seeing the most overdose deaths. And it allowed a coalition of local public health officials, treatment providers, law enforcement, and others to bring deaths down 20 percent between 2020 and 2022. The study also illuminated some real challenges in using data effectively to improve public health nationwide. The U.S. is uniquely bad at collecting health data, in part because our health system is so fractured. Hospitals own their data and can be resistant to sharing it. But the ability to tackle epidemics of all kinds—from covid to drug addiction—requires being able to see who is affected and where they are located. When county departments do get data, even on basic metrics like opioid deaths, it can be years behind. There’s also a glaring lack of specialists to work with the data they do gather. The study provided each county with its own data analyst for the duration of the study. But after the study ended, many counties found themselves with reams of data and unable to pay a data analyst to work with it. While they are still using the data to send out naloxone and educational materials, they could be using the data to better understand smaller trends among certain ethnicities or other groups. For example, during the study, researchers collected all of the meeting minutes from the coalition meetings. Those meeting minutes are a rich source of qualitative data, where members of the community discuss the problems they’re running into and experiences treating substance use disorder. In theory it could reveal important new patterns or help communities to come up with plans to fill holes in their understanding of the issue. “We didn't have time to analyze the minutes,” said Nabila El-Bassel, professor of social work at Columbia University who led the HEALing Communities study in New York. This is where AI could come in. While El-Bassel says it’s very important there’s a human to review takeaways delivered by AI, technology could do a lot of data crunching. For instance, ChatGPT could analyze those minutes quickly, and help coalition members create action items. “You know how many seconds it takes to download a focus group of two hours? It takes one minute, two seconds,” she said. The 67 counties that participated in the study now have a basic infrastructure for pulling in data quickly. And it can be built on to track other health trends within people suffering from substance abuse, like the prevalence of infectious disease. She says with more and better data, AI can help unearth microtrends, like what percentage of people suffering from opiate addiction are unhoused and where they may be congregating or moving to. “If we collect data about different populations, we can really address the underserved populations fast,” she said.
| 
