Geography of poly-substance drug mortality

Abstract

BackgroundThere has been a rise in drug overdoses and death in recent years. We examine the spatial distribution of drug related deaths in the U.S., by typing places by multiple problem substances and predicting the types with potential covariates. MethodsUnits of analysis are N = 3079 counties in the 48 conterminous United States. Drug death rates by substance are drawn from confidential cause-of-death mortality files from the National Vital Statistics System and independent variables are from several official sources. Latent profile analysis is used to identify drug clusters by classifying counties into profiles based on mortality rates across six drug types in 2017–19 and change from 2000 to 02. A multivariate general linear model is used to explore unconditional mean differences by latent profile across a number of variables. ResultsLPA identified 7 classes. These clusters vary by key covariates, with a cocaine and illicit opioid problem concentrating in relatively affluent urban areas, a methamphetamine problem in rural areas, a prescription drug problem concentrated in a few states, and a syndemic in places that were the origin of the opioid pill epidemic of preceding decades. ConclusionsThere is no longer a single drug crisis in the U.S. Rather, the problem is diverse spatially. A single national drug policy response is unlikely to be equally effective in all parts, but findings speak to responses that have promise in particular contexts.