Substance Use
Decomposing the US drug overdose crisis Mathew Kiang* Mathew Kiang Kiang Kiang Stanford University
Despite recent modest declines, the US continues to experience high drug-related mortality at over 30 deaths per 100,000 population in 2023. Previous work has found variation in drug-related mortality across geography, sociodemographic characteristics, and underlying drug. However, this work has largely focused on describing these patterns across a single axis and using overlapping drug categories, limiting their ability to inform health policy.
Here, we address these limitations by using demographic decomposition methods on the NCHS restricted-access multiple cause of death data from 1999 to 2023. Using Kitagawa decomposition, we quantify the contribution of mutually-exclusive drug categories to the change in age-standardized national drug-related mortality rate by both geographic (states) and sociodemographic (race-ethnicity-age-sex-specific) subgroups.
When examining the contribution of different drugs to the change in the national drug-related mortality rate, we found distinct geographic patterns. Since 2020 xylazine-involved mortality contributed approximately 0.25 to 0.5 deaths per 100,000 population to the national drug-related mortality rate with the majority limited to just two states (NY, PA). In contrast, deaths involving both methamphetamine and fentanyl contributed 0.4 to 1.4 deaths per 100,000 to the national rate and the majority of this contribution came from West Coast states (CA, WA, OR). Similarly, deaths involving cocaine and fentanyl contributed between 0.3 to 0.8 deaths per 100,000 but were spread across Eastern US states (e.g., NY, MI, OH, PA, MD, VA). While 2023 saw a 1.25 deaths per 100,000 decline in the national rate, this decline was driven primarily by six states (IL, IN, OH, PA, NC, FL) and almost entirely by reductions in fentanyl-involved deaths.
Demographic decomposition methods provide actionable insights into the US drug overdose crisis, revealing not only who and where people are most impacted, but also which substances are driving change.