Emergency Medical Services
Abstracts
.png "Catherine Wolff, MS photo")
Background: The opioid epidemic has grown rapidly in North Carolina (NC), with deaths increasing by nearly 40% annually since 2015. EMS are an important partner in treating and tracking opioid overdoses; however, existing data linkage methods between EMS encounters and subsequent ED visits do not allow for comprehensive outcome analysis. We developed novel methods to enhance linkage performance and allow for analysis of ED visit data following EMS hand-off.
Methods: We identified all NC EMS encounters between 1/1/2017 – 11/30/2017 with documented naloxone administration and transportation to the ED, excluding EMS transports outside of NC or to military hospitals, using data from NC’s Disease Event Tracking and Epidemiologic Collection Tool (NC DETECT) and EMS Performance Improvement Center. A non-random subset of 12 counties was evaluated via manual chart review to create a “gold standard” linked dataset that informed development of novel linkage methods using demographic, time, and destination data. Descriptive statistics were performed comparing linked versus unlinked records. ED diagnosis codes were evaluated to determine the proportion of EMS suspected opioid overdoses confirmed by ED data.
Results: We identified 12,089 EMS encounters treated with naloxone and transported to the ED. Among these, 1,906 (15.8%) were included in the 12-county subset, with existing linkage performance of 64.8% (1,154/1,781) and 2.0% false-linkages. Following implementation of an iterative linkage method, performance improved to 91.0% (1,620/1,781) with 0.05% false-linkages. No differences by age or sex were found between linked and unlinked encounters. Only 27.2% of linked ED records included a diagnosis code for opioid overdose.
Conclusion: Through an iterative linkage approach, EMS and ED data linkage performance improved greatly while reducing the number of false-linkages. This improved linkage allows for evaluation of patient outcomes to inform opioid overdose surveillance and treatment, and may be helpful for state ED surveillance efforts more broadly. Future research directions include efforts to better understand the low proportion of EMS suspected opioid overdose patients receiving an ED diagnosis of opioid overdose and exploring applications of machine learning approaches to further improve linkage performance.