Abstract: The Changing Epidemiology of Non-Fatal Overdose in Baltimore City: Insights from Emergency Medical Services Encounters (Society for Prevention Research 27th Annual Meeting)

582 The Changing Epidemiology of Non-Fatal Overdose in Baltimore City: Insights from Emergency Medical Services Encounters

Friday, May 31, 2019
Garden Room B (Hyatt Regency San Francisco)
* noted as presenting author
Brian W. Weir, PhD, Assistant Scientist, The Johns Hopkins University, Blatimore, MD
Chen Dun, BA, MHS student, The Johns Hopkins University, Baltimore, MD
Kelly M. King, PhD, Mobile Integrated Health Analyst, Baltimore City Fire Department, Baltimore, MD
Amy Knowlton, PhD, Professor, The Johns Hopkins University, Baltimore, MD
Introduction: In the United States, fatal overdoses increased five-fold from 1999 to 2016, reaching 72,855 in 2017. In Baltimore City, fatal opioid-related overdoses increased from 354 in 2015 to 628 in 2016. Emergency medical services (EMS) are often first responders to overdose, and EMS electronic records provide timely data on overdose with important demographic, temporal and spatial information. This study describes the changing epidemiology of non-fatal overdose in Baltimore, based on EMS encounters with naloxone administration, from 2012 through 2017.

Method: Data on EMS patient encounters with naloxone administration (1/2012 – 12/2017) were obtained from the Baltimore City Fire Department. Encounters within the city limits were included if there was a record of naloxone administration. We used geocoordinates to assign encounters to census tracts with census tract characteristics from the Baltimore Neighborhood Indicators Alliance. Independent variables included year, sex, race, and age. City-level population-based incidence rates per 1,000 persons 15 years of age or older are presented for demographic groups over time. Factors associated with incidence rates and changes in incidence rates over time were evaluated through univariable and multivariable negative binomial regression. Variation in incidence across census tracts was evaluated through multilevel negative binomial regressions with time nested within census tracts. Geographic variation is presented through choropleth maps. Analyses were conducted in Stata and ArcGIS Pro.

Results: The number of EMS encounters with naloxone administration approximately doubled from 1,557 in 2012 to 2,966 in 2015 and doubled again to 6,520 in 2017, corresponding to incidence rates of 3.0, 5.8, and 12.7 per 1,000 person-years, respectively. Incidence rates were nearly 3 times higher for males than females, similar for non-Hispanic Whites and African Americans but lower for other racial and ethnic groups, and peaked among those 50-54 years of age. Incidence rate ratios for changes over time (2017 vs. 2015) were higher for African Americans than Whites, similar for males and females, and varied by age group. Multilevel models show significant variation across census tracts in incidence in 2015 and in the increase in incidence from 2015 to 2017. Maps illustrate the unequal distribution among census tracts of EMS naloxone administration and of the dramatic increase in incidence from 2015 to 2017.

Conclusion: This study reveals the dramatic temporal, spatial, and demographic patterns of non-fatal opioid overdose in Baltimore as reflected by EMS encounters with naloxone administration. This approach can be used for sentinel surveillance, to identify areas for harm reduction or other intervention, and to inform health equity advocacy.