Reducing Drug-Related Mortality Using Predictive Analytics: A Randomized, Statewide, Community Intervention Trial

点击翻译按钮获取中文摘要

PROJECT SUMMARY Overdose deaths have skyrocketed in the United States since 1999. The epidemic has prompted widespread federal and state actions, yet the number of people who die of an overdose continues to increase. In light of the accelerating and rapidly evolving overdose epidemic, new strategies are needed to identify communities most at risk, and to utilize resources more effectively to curb overdose deaths. To address these public health priorities, we will develop a forecasting tool to predict overdose deaths before they occur, and then conduct a randomized, statewide, community-level intervention to evaluate resource targeting based on these predictions. The study will take place in Rhode Island, a state with the 10th highest rate of overdose fatality in 2016. The study has two phases. First, we will develop a predictive analytics model that forecasts future overdose mortality at the neighborhood-level, using publicly available information and data from a multicomponent overdose surveillance system. This tool, called PROVIDENT (Preventing Overdose using Information and Data from the Environment) will be used to predict the likelihood of magnitude of future overdose deaths in every neighborhood across Rhode Island. Next, we will conduct a randomized policy experiment to evaluate whether targeting overdose prevention interventions to neighborhoods at highest risk reduces overdose morbidity and mortality. The state's department of health will receive PROVIDENT model predictions for half of the 39 cities/towns in Rhode Island. Within these cities/town, the health department will work with stakeholders to target overdose prevention interventions to neighborhoods with the highest probability of future overdose deaths. Interventions include efforts to: (1) prevent high-risk prescribing (through academic detailing and other educational efforts); (2) expand access to opioid agonist therapy, including buprenorphine and methadone; (3) increase naloxone distribution (through community and pharmacy-based efforts); and (4) expand street-based peer recovery coaching and referrals. Control cities/town will continue to receive these interventions, but without targeting to specific neighborhoods. Fatal and non-fatal opioid overdose rates in the control cities/towns will be compared to those that received the PROVIDENT model predictions. To achieve these aims, we will leverage a unique partnership between an academic institution and a state's health department, which allows for unprecedented access to and sharing of population-based overdose surveillance data. Our results will improve public health decision-making and inform resource allocation to communities that should be prioritized for evidence-based prevention, treatment, recovery, and overdose rescue services. If found to be effective, the PROVIDENT forecasting model will be disseminated to other states, which could adapt the tool to guide resource allocation and maximize public health impact. In sum, this project is highly responsive to a top research priority of the National Institute on Drug Abuse, and directly addresses one of the nation's most challenging public health crises.