In 2020, Ohio-based nonprofit the SOAR Initiative launched a pioneering “bad batch” alert system. Its platform allows anyone to submit secure, anonymous reports of drugs they’ve encountered in the local supply, which may be dangerously adulterated or sold as something other than what they are. These reports are vetted by SOAR’s outreach staff, and those deemed reliable are texted to thousands of subscribers in the community. The model has been replicated by community organizations and health departments across the country.
Now, armed with a $200,000 grant from the OneOhio Recovery Foundation, SOAR is reimagining how to best collect and communicate this crucial information.
Much of the funding—which comes from the controversial foundation created to distribute Ohio’s opioid settlement funds—will go towards modernizing the existing technology. But SOAR also plans to expand its investment in a less conventional approach: deploying community outreach workers, who collect and distribute information in person.
“They build the trust with the community. The participants they serve, they know.”
While other alert systems have increasingly moved towards automation, usually by feeding emergency medical services data to powerful algorithms designed to predict overdose spikes, SOAR (for which I previously served as director of operations) has maintained its commitment to a bottom-up approach. It employs eight outreach workers in strategic locations across Ohio, each tasked with keeping a pulse on the community’s needs, verifying alerts and distributing harm reduction supplies.
Tonja Catron, SOAR’s executive director, told Filter that the community-based knowledge of people who use drugs is the “frontline of defense” against overdose. SOAR’s outreach workers, who are deeply embedded in these communities and have their own lived experiences of substance use, provide a space for people to safely share time-sensitive information about the drug supply.
“They build the trust with the community,” Catron said. “The participants they serve, they know. So they are always gonna … report to them first.”
In the age of algorithms and artificial intelligence, SOAR’s commitment to boots-on-the-ground information gathering may seem old-fashioned. However, Catron argues that it keeps the organization in alignment with harm reduction’s core principle of elevating the voice of people who use drugs.
“These are coming from real people, and most of our alerts are coming from people who use drugs,” she said. “It’s not coming from a computer system, just analyzing trends and putting that out. It’s coming firsthand from drug users.”
By leveraging the community’s collective expertise, SOAR often gets word of overdose spikes or high-risk substances before public health catches on. This allows its outreach workers to respond to risks quickly.
health departments’ alert systems have increasingly focused on automated, algorithmic methods.
“When we send out those alerts, then we have people that are ready to meet the community with naloxone or test strips or linkage to care,” Catron explained.
An added benefit of this strategy is the financial support it provides to outreach workers. Most have provided services to their communities on a volunteer basis for years, paying for supplies themselves. For those who have been justice system-involved, working for SOAR can also bolster their resumes for future employment opportunities.
This community-based approach is quite different from the recent trajectory of health departments, whose alert systems have increasingly focused on automated, algorithmic methods.
For example, a study recently published in the journal Addiction found that non-fatal overdose data can be used to predict fatal overdose surges. This allows under-resourced health departments to more effectively mobilize support, since non-fatal overdose data is more quickly and easily available.
“The processing time that it takes to get the fatal data varies a lot depending on local capacity,” lead study author Thomas Patton, a researcher at the University of California San Diego Department of Medicine, told Filter. “These [nonfatal] data are at least routinely available.”
The study found that both methods were quite reliable in highly populated areas, though their accuracy declined in smaller counties.
Using emergency department and emergency medical services data from California and Florida, the research team tested two overdose detection algorithms commonly used by health departments. The algorithms use different mathematical approaches, but both were designed to automatically detect “aberrations”—substantial deviations from the number of overdoses that would be considered “normal” in an area at a given time.
The study found that both methods were quite reliable in highly populated areas, though their accuracy declined substantially in smaller counties.
“We didn’t see one method really dominate over the other,” Patton said, emphasizing the value of enlisting multiple detection strategies—including cutting-edge techniques like wastewater surveillance—when possible.
A major benefit of this approach is that most health departments already have access to emergency medical data, which is usually reported in real time. Implementing an alert system based on the “aberration detection methods” proposed in the paper would be relatively cheap and simple.
On the other hand, Patton also highlighted the importance of community-led reporting initiatives like SOAR’s bad batch alerts, calling them “inseparable” from these algorithmic approaches. Without information from the community, health departments might be relying on the wrong indicators.
“We don’t know, as drug use evolves and changes over time, what’s gonna come up next,” he explained. “We might not have the data needed.”
Grassroots organizations, however, are more hesitant to collaborate with health departments, reflecting ongoing tensions between the two groups. Catron said SOAR is far more willing to share health department information with its subscribers than it is to share its own reports with health departments.
Unless public health systems can build more trust with grassroots communities, it’s unclear whether the two approaches can converge.
“Anytime a coroner sends us a report, we’re sure to send it out,” she said. But when it comes to sharing information provided by SOAR’s subscribers, “There’s not a lot of trust there still with public health.”
This stems from the relationships between public health and law enforcement. To ensure people feel comfortable submitting reports, SOAR avoids partnerships with any perceived risk of bringing law enforcement into the loop.
While SOAR’s model is powerful, it has its limitations. It takes time to build the large network of community informants, outreach workers and institutional connections necessary for a successful system. It’s also expensive; SOAR has been awarded over half a million dollars of funding to build and maintain its text message alerts and app (which was discontinued in 2024) since 2020.
SOAR plans to continue expanding its network of community outreach workers, and invites organizations in other states to reach out for more information about implementing the system in their communities.
For many municipalities, algorithmic solutions that leverage existing public health data are currently the only viable option for warning the public about potential overdose spikes. As Patton and his team show, innovative new techniques, such as using non-fatal indicators, can greatly enhance the timeliness of alerts in low-capacity places.
Community-driven and algorithmic alert systems have each shown value, and each continue to receive significant private and public funding. But unless public health systems can build more trust with grassroots communities, it’s unclear whether the two approaches can converge.
Photograph of SOAR workers and volunteers courtesy of the SOAR Initiative