A Spatio-Temporal Joint Model of Opioid Overdose Deaths and Treatment Seeking in Ohio Counties

• Activity Number: 358 - SPEED: Statistics in Epidemiology
• Type: Contributed
• Date/Time: Tuesday, August 1, 2017 : 10:30 AM to 11:15 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): David Kline* and Staci A Hepler and Erin McKnight and Andrea Bonny
• Companies: The Ohio State University and Wake Forest University and Nationwide Children's Hospital and Nationwide Children's Hospital
• Keywords: epidemiology ; applied ; spatial ; Bayesian ; joint model ; areal

Abstract

Opioid misuse is a national epidemic and a significant public health issue due to its high prevalence of associated morbidity and mortality. As a state, Ohio has been hit as hard by the opioid epidemic as any in the country. As part of its surveillance effort, the Ohio Department of Mental Health and Drug Addiction Services collects counts of individuals seeking treatment for opioid misuse and of those who die from opioid overdose for all 88 counties in the state. While typical analyses consider each of these rates independently, we examine how county-level rates of treatment seeking and overdose deaths jointly vary across space and time. In addition, we explore their association with social environmental characteristics. To do so, we use a dynamic latent trait model with a spatio-temporal conditional autoregressive process. We discuss the statistical considerations of this approach and its implementation for this application. We also present the results from the analysis and discuss how they can be interpreted and applied clinically.