County-Level Influenza-Attributable Emergency Department Visits and their Spatial Correlates in the United States (Preprint)

BACKGROUND Burden of seasonal influenza on emergency department (ED) visits is poorly quantified due to case ascertainment and data availability challenges. This study estimates county-level respiratory ED visits attributable to influenza using time-series models and examines spatial heterogeneity in county-level burden in three states. OBJECTIVE To estimate county-level seasonal influenza–attributable emergency department (ED) visit rates and examine their associations with socioeconomic, environmental, and chronic health covariates in three US states (California, Georgia, and New York). METHODS We used daily hospital discharge records to measure community-level influenza activity in California (2005–2018), Georgia (2010–2018), and New York (2005–2018). County-level influenza-attributable respiratory ED visit rates were estimated using quasi-Poisson time-series models that adjusted for temporal trends and environmental factors. Under a meta-regression framework, Bayesian spatial hierarchical models were then applied to evaluate associations between ED visit rates and county-level socioeconomic status (SES), environmental exposures, and chronic health conditions. RESULTS Influenza-attributable respiratory ED visit rates per 100,000 population were 226 (95% CI: 206, 246) in New York, 232 (95% CI: 206, 259) in California, and 547 (95% CI: 506, 589) in Georgia. A 10% increase in county-level poverty and uninsured rates were associated with higher influenza burden, increasing influenza-attributable respiratory ED visit rates by 160 (95% CrI: 127, 196) and 217 (95% CrI: 168, 265), respectively. Long-term PM2.5, humidity, and temperature also exhibited positive associations. Chronic conditions also increased ED visit rates by 1,476/100,000 (95% CrI: 1,167, 1,778), 588/100,000 (95% CrI: 400, 747), and 488/100,000 (95% CrI: 402, 574) per 10% increase in stroke, chronic obstructive pulmonary disease, and diabetes prevalence, respectively. These associations weakened after adjusting for SES. CONCLUSIONS Influenza-attributable respiratory ED visit rates exhibit significant spatial heterogeneity that is associated with county-level socioeconomic factors, environmental exposures, and chronic disease prevalences.