TrestleTree LLC of Fayetteville has purchased the rights to a digital tool that determines the level of risk for opioid abuse and addiction that patients face before their first opioid prescription is written.
Bradley Martin, who heads the Division of Pharmaceutical Evaluation & Policy at the University of Arkansas for Medical Sciences, and his team developed the tool, which uses an algorithm to assign risk.
Financial details of the deal were not disclosed. UAMS BioVentures, the university’s technology licensing office and startup incubator, is a private nonprofit.
TrestleTree was founded in 2001 and has nearly 100 employees, CEO Ted Borgstadt told Arkansas Business. It works with people who have chronic conditions and with their families to change patients’ behavior. The goal is to lower health care costs for its clients — large employers, health insurance plans and hospital systems.
TrestleTree excels at changing the behavior of people who don’t want to change their behavior, Borgstadt said. Its “coaches” build trusting relationships to help accomplish this.
The firm plans to use the new tool to enhance what it already does. The tool will be online, integrated into TrestleTree’s existing database “in weeks, not months,” he said. Coaches are already assigned to some patients who will receive opioid prescriptions, like those awaiting surgery, but the tool will help determine whether a patient needs a coach for a longer time because of greater risks for opioid abuse.
“The thing that intrigued us was the ability to look at a population and have some sense of probability, of risk, that someone might misuse or abuse or hit the slippery slope with opioids before they have ever taken their first prescription,” Borgstadt said. “So the opportunity to really get ahead of an issue for an individual was something that was very intriguing to us, and it seemed to fit what our clients were asking us to do as well.”
Martin, the developer, said work on the tool began in April 2017 with a $46,000 grant from the UAMS Translational Research Institute. His team explored more than 300 variables in developing the tool. The models within it, including the one that predicts overdoses and the model that predicts abuse disorders, uses 30 to 60. Variables include the amount of opioids prescribed, whether someone has also been prescribed other drugs and what type of pain is being treated.
Also, he said, testing found that the models being acquired are at least 90 percent accurate. Just 80 percent is required for use in a clinical setting.
Sign up for Daily E-news
