Innovations in Modeling Existing and Emerging Policies to Improve Warning Systems for Opioid Overdoses - PROJECT SUMMARY/ABSTRACT This application seeks three years of dissertation funding to have a computational PhD candidate confront a pressing public health issue using three interrelated, interdisciplinary aims supported by concomitant mentorship and training that will prepare them for a future academic career in using statistical, computational and mixed- methods techniques to develop and analyze interventions to address the opioid crisis. Over 9.5 million people reported using opioids in the US in 2020, during which time there were ~257 opioid-related overdose deaths per day. Fatal overdose rates have grown nationally by 274% from 2013 to 2020, largely due to the growing presence of synthetic opioids and other additives, primarily found in street drug supplies. Street-obtained substances with unexpected composition, such as cocaine containing fentanyl or heroin with novel fentanyl and benzodiazepine derivatives, have been causing overdoses en masse. To mitigate the scale of these mass injury events, over 3000 agencies in 49 states use the Overdose Detection Mapping Application Program [ODMAP], which features a “spike alert”-based warning system. ODMAP issues a spike alert when overdose counts exceed preset thresholds within 24 hours to help mobilize rapid public health responses to prevent overdoses and save lives. The state of Connecticut has one of the highest overdose rates in the country, with 39.1 out of every 100,000 people experiencing a fatal overdose in 2020. To address this crisis, it has implemented one of the most progressive evidence-based overdose spike response systems in the nation, with each ODMAP spike alert undergoing an extensive manual review by the Department of Public Health that occasionally culminates in a public health alert. The effectiveness of this system rests on its ability to accurately identify spikes and rapidly mobilize a public health response to save lives, but it is unclear 1) if the system has any effect on overdose rates 2) how first responders, harm reduction organizations and health systems make use of the system to rapidly respond to overdose spikes 3) if the system can be modified to more accurately identify spikes and motivate rapid responses to save lives. I therefore propose to 1) estimate the causal effect of Connecticut’s current spike alert system on subsequent overdose-related outcomes; 2) assess utilization of the current system, barriers to overdose prevention and opinions on alternatives to the status quo; and 3) develop and simulate the impact of alternative spike alert strategies on overdose-related outcomes. To address these Aims, I will use a combination of cutting-edge causal inference, mixed-methods, space-time regression and epidemiological modeling techniques, along with integrated data sources and guidance from key stakeholders. These findings will provide actionable advice to improve Connecticut’s current spike alert system, can motivate future policy work to address the overdose crisis and provide a framework for other health departments looking to implement spike alert systems that are responsive to stakeholder needs and can more effectively save lives than the status quo.
