A spatial framework for selecting sentinel sites for drug and alcohol related early warning for the United States.

Abstract

Objective: The absence of a robust early warning system in the U.S. remains a pressing issue, particularly for preventing overdose-related epidemics in vulnerable areas. This research provides a timely framework for selecting sentinel sites for drug- and alcohol-related early warning. The system, once in place, will be crucial for informing and implementing timely and effective interventions for overdose prevention. The sentinel strategy-specifically, the selection of hospital sites for the Drug Abuse Warning Network (DAWN) national overdose surveillance system-is a step toward addressing this urgent need.

Data and methods: Sentinel counties were selected using a geographic framework based on a modified version of the CDC's Social Vulnerability Index (SVI), incorporating measures of overall vulnerability, drug-related mortality, and alcohol-related mortality. Principal component analysis (PCA) was applied for data reduction and composite scoring. Mortality data were obtained from the CDC WONDER Multiple Cause of Death (MCOD) database, while nonfatal overdose data came from the U.S. Department of Transportation's NHTSA opioid overdose tracker. Data on High Intensity Drug Trafficking Areas (HIDTAs) were included to assess overlap with vulnerable counties. The American Hospital Association (AHA) data on DAWN-eligible hospitals and emergency department (ED) visits were used to identify potential hospital sites within selected sentinel counties for the DAWN early warning system.

Results: Principal component analysis of standardized county-level indicators revealed significant geographic disparities in overdose vulnerability. Counties in West Virginia, New Mexico, southern Texas, the Dakotas, and Alaska exhibited high social vulnerability. Drug-related mortality was highest in West Virginia, while alcohol-related mortality was elevated in New Mexico and South Dakota. Vulnerable counties overlapped with HIDTA regions in Arizona, New Mexico, Texas, Florida, South Dakota, Kentucky, and West Virginia, with additional overlap in southern California. Nonfatal overdose rates from NHTSA were also elevated in many of these counties, supporting their designation as potential sentinel counties for early warning.

Conclusion: Vulnerable, high-risk counties with elevated drug- and alcohol-related mortality may be key drivers of the ongoing overdose epidemic in the U.S. These counties are geographically dispersed, underscoring the need for a targeted spatial strategy. Sentinel hospitals can be selected within these high-risk areas to detect emerging drug trends, novel substances, and evolving slang terms. This framework strengthens the application of spatial epidemiology in surveillance and offers a replicable approach for site selection in regions where resource constraints limit active surveillance.