Substance Use
Association of non-fatal overdose surveillance data with concurrent and future overdose deaths in Rhode Island Alexandra Skinner* Alexandra Skinner Yu Li Benjamin D. Hallowell Claire Pratty William C. Goedel Bennett Allen John C. Halifax Alexandria Macmadu Jennifer Ahern Magdalena Cerdá Brandon D.L. Marshall
Given substantial reporting delays in overdose deaths, state health departments typically rely on non-fatal overdose data to inform rapid overdose response efforts. For instance, the Rhode Island (RI) Department of Health monitors emergency medical services (EMS) responses to suspected overdose events in near-real time to detect increases in overdose activity and to direct public health resources to the corresponding neighborhoods. However, there is limited understanding about whether EMS-attended non-fatal overdoses serve as a reliable proxy for fatal overdoses. We sought to evaluate the extent to which non-fatal overdose data were associated with concurrent and future overdose deaths in RI.
We aggregated non-fatal overdose data from EMS records (2019–2022) and fatal overdose data from the RI State Unintentional Drug Overdose Reporting System (2020–2022) in 3-month intervals at both census block group (CBG) and census tract (CT) levels. We estimated rates of fatal overdose, relative to non-fatal overdose lagged by 0, 3, 6, 9, and 12 months, using negative binomial regression models offset by population size. We applied integrated nested Laplace approximation, a Bayesian spatiotemporal approach that allows for spatial autocorrelation and small area estimation.
At the CBG level, each additional EMS-attended non-fatal overdose was associated with rates of fatal overdose that were 11–21% higher than expected in concurrent or future 3-month intervals. However, at the CT level, we observed diminished predictive utility; each additional non-fatal overdose was associated with fatal overdose rates just 4–10% higher than expected (Figure 1). Forthcoming analyses will explore these associations at town and regional levels and in monthly intervals.
Spikes in non-fatal overdoses were most strongly associated with elevated overdose mortality in concurrent and near-concurrent periods. At small geographies, non-fatal overdose activity may be indicative of fatal overdose burden.
