Enhancing Infrastructure for Clinical and Translational Research to Address the Opioid Epidemic

Opioid misuse and dependence continues to be a significant and growing cause of preventable morbidity and mortality in the United States (National Center for Health Statistics, 2017). Over the last 10 years, there have been dramatic increases in the number of individuals presenting to emergency rooms (ERs) across the country as a result of intentional or unintentional opioid overdose (Daly et al., 2017; Hsu et al., 2017). Many of these individuals die or have a repeat overdose presentation within a year (CITATION) and public health costs associated with opioid overdose are high (Florence et al., 2016; Maeng et al., 2017). The defining features of the opioid epidemic have changed over time. While prescription opioid misuse was initially the primary problem, increasingly the use of high potency non-prescription drugs, such as fentanyl and heroin, have been responsible for opioid-related ER presentations and mortality (Hsu et al., 2017; Rudd et al., 2016). Recent data suggest that a growing proportion of opioid dependent individuals are initiating use with heroin (or illicit synthetic opioids), rather than prescribed opioids (Cicero et al., 2017; Cicero et al., 2015) such that opioid overdose may continue to increase over the next 1-5 years (Blau, 2017). Patients presenting with opiate overdose to ERs represent a high-risk group for morbidity and mortality and a potential target for interventions to combat the epidemic. The ability to characterize these individuals could inform trials of multi-level, multi-system overdose prevention efforts, as well as translational research efforts to connect opioid-dependent individuals with effective treatments. We propose foundational work to create an interinstitutional research database and network focusing on studies in patients presenting to ERs with opiate overdose. To create this network, we will extend work by Daly and colleagues (2017) to develop a (1) e-phenotype for case identification in the ER based on EHR data, and combine this (2) data dictionary and tools extraction for coded data and natural language processing (NLP) algorithms to obtain additional data from EHRs to characterize individuals presenting to EDs with opioid overdose including: demographics, comorbidities, overdose agent and source, intentionality of the overdose and patient disposition upon ED discharge; as well as tools to facilitate better capture of relevant data in EHRs. We are currently refining the data extraction process using 2016-2017 data from the Medical University of South Carolina EHR. The funding from this proposal will be used to manually validate the e-phenotype and data extraction routines. In partnership with University of San Diego (UCSD), we will integrate the data into Accrual to Clinical Trials (ACT) clinical data model i2b2 ontology, extending the model where needed. We will also partner with CTSA’s at University of Kentucky (UK) and Dartmouth to extend the project to a more representative sample from across the country. We then propose to deploy and populate the model across the network and demonstrate (1) near real time surveillance capability for opiate overdose; (2) queries of patients counts and interactive trial design using SHRINE software. Using this funding to refine and automate the data extraction process, it could be extended throughout the ACT network and to CTSA’s nation-wide to provide timely information about the evolving opioid epidemic and a platform to accelerate research.