STIMuLINK will identify risk/protective factors for stimulant-involved deaths that are actionable, for the development or adaptation of current prevention and intervention efforts. STIMuLINK will link records from fatal stimulant-involved overdoses found in Kentucky’s drug overdose fatality surveillance system to electronic medical records (EMR) from the state’s largest primary and safety net health care system, UK HealthCare, for the period 2017–2024. This innovative data linkage reduces the surveillance bias associated with medical information obtained only from coroner reports, makes EMR “gold standard” medical information available for research, improves data completeness needed for risk/protective factor ascertainment using the concept of “computable phenotypes,” and enables a comparative cohort of patients found in the EMR (i.e., those with evidence of harmful stimulant use).
Using traditional and novel statistical approaches to analyze these cohorts, STIMuLINK will examine risk/protective factors predictive of fatal stimulant overdose using the social ecological model as the principal framework.
In parallel with these aims, STIMuLINK tests the use of existing biomedical natural language processing (NLP) tools to extract additional factors from the “literal text” notes found in the Drug Overdose Fatality Surveillance System (DOFSS) and compares findings from notes written by coroners to those written by medical examiners in neighboring states. The guidance and open-source software developed from this aim will advance the science of NLP for automating the workflow for injury surveillance.
STIMuLINK’s multidisciplinary team includes epidemiologists, computer scientists, lawyers, statisticians, and behavioral scientists with experience working directly with communities affected by the current polydrug overdose epidemic. The project specifically includes ongoing communication points with vulnerable communities and clinical populations to increase the translational impact of the research.
