Human Services

Reducing Risks for Children and Adults

Doctors Walid Gellad, Julie Donohue, and Eric Hulsey of University of Pittsburgh’s Center for Pharmaceutical Policy and Prescribing seek to develop a predictive model that, if successful, can be used by local health and human service agencies to target services and interventions for at-risk populations.

Developing Models to Identify Opioid Overdose Risk:

University of Pittsburgh

Developing Models to Identify Opioid Overdose Risk:

University of Pittsburgh

In 2016, overdoses in Pennsylvania totaled 4,627—the fifth highest in the nation—and counties across the state have experienced dramatic increases in overdose deaths. In Allegheny County, 591 individuals died from a drug overdose, and 75 percent involved fentanyl, a synthetic opioid that is 80-100 times stronger than morphine. 

With a $1,800,000 National Institutes of Health (NIH) grant, the University of Pittsburgh’s Center for Pharmaceutical Policy and Prescribing is developing machine learning algorithms to predict opioid overdose risk statewide, using medical claims data from Pennsylvania Medicaid. A $446,000 Foundation grant will support the Center’s efforts to develop a predictive model for opioid overdoses in Allegheny County, using a combination of Medicaid data and county data from the Health Department, the courts, and the Department of Human Services, which, in 2017, had interactions with 68 percent of individuals who overdosed. 

In the same way Allegheny County uses predictive risk scores to address child abuse, an initiative garnering national recognition, a successful model of predictive risk scores for overdoses will enable health and human service agencies to target preemptive interventions in neighborhoods and populations that are most likely to benefit from them.