LAW ENFORCEMENT RISK MODEL TO COMBAT OPIOID RECIDIVISM

Abstract

Bergen County, New Jersey, has seen opioid-related overdoses and deaths spike in the last few years. One of the challenges in addressing this epidemic is that “at-risk” individuals may encounter multiple segmented domains such as law enforcement, recovery services, and healthcare institutions, but no one agency has oversight of all the contacts. Each encounter with at-risk populations, including those who suffer from opioid addiction or who may recidivate, becomes a data record in a system. This thesis asks how can law enforcement leverage such data sets to address the opioid epidemic and battle recidivism? This research examined law enforcement arrest data and overdose reporting in Bergen County, analyzing which risk factors in recidivism could be discerned using statistical information, cross-tabulations, Pearson’s chi-squared tests, and data modeling from the Cox proportional hazards model. The results showed that no demographic profile was more likely to have another overdose or death, and theft arrests coincided with a decreased chance of overdose, despite law enforcement’s presumption of the contrary. The strongest predictor of an overdose was a prior overdose, with the risk increasing for each additional overdose. Additionally, having any contact with law enforcement was an indicator of a significantly higher chance of overdose or death. Thus, each interaction between law enforcement and an observed opioid abuser is a critical point for intervention.