A spatio-temporal Bayesian model to estimate spatial patterns of risk for drug-involved emergency department visits in the greater Baltimore-Maryland region

Keywords: Opioid crisis, nonfatal emergency department visit, spatio-temporal Bayesian model, Baltimore-Maryland region
Session Type: Virtual Paper
Day: Friday
Session Start / End Time: 4/9/2021 04:40 PM (Pacific Time (US & Canada)) - 4/9/2021 05:55 PM (Pacific Time (US & Canada))
Room: Virtual 8

Abstract

The ongoing US Opioid Overdose Epidemic presents a major public health challenge. Nonfatal emergency department (ED) visits are a key opportunity to prevent mortality and measure the extent of the problem. Estimating locally sensitive, recent rates of ED visits can provide a more accurate understanding of the drug use landscape. Using spatio-temporal Bayesian models and a sample of ED visits from the Maryland Emergency Department Drug Surveillance (EDDS) system, this study estimates both the frequency and risk of heroin-, methadone-, and cocaine-involved ED visits across the greater Baltimore Maryland region at the Zip Code Tabulation Area-level (ZCTA). These models consider space-time patterns while also adjusting for area-level measures of confounding factors related to the social determinants of health, propensity for drug issues, and location relative to health services. ED visits at the ZCTA-level demonstrated positive spatial autocorrelation across drug-demographic strata. Models indicated that higher scores in a deprivation index, share of Medicare claims, and adjacency to a sampled UMMS hospital increased the risk of ED visits. Maps of posterior median risk and posterior exceedance probabilities showed an elevated risk for heroin in downtown Baltimore and Glen Burnie, for methadone from Baltimore city center extending northward towards Pennsylvania, and for cocaine in a North-South pattern across the state cutting through Baltimore. This modeling approach used a sample of ED visits from EDDS to estimate locally sensitive, recent rates of drug-related morbidity across a large metropolitan area. These estimates identify high-risk areas and can inform local health system planning.