Multiple Determinants of Vulnerability for Emergency Department Visits for Heat-Related Illness in California 2005-2008 Warm Seasons

Abstract

Background: High temperatures are associated with risk of Heat-Related Illness (HRI) and other acute clinical outcomes. Information on HRI risk factors is rarely place-based; decision-makers need local-scale (e.g., neighborhood) data to inform strategies to reduce heat health impacts. Objective: In multi- determinant multi-level models assess relations between daily maximum temperature (Tmax) and emergency department (ED) visits for HRI and vulnerability factors, i.e., biological susceptibility (e.g., sex, age, race), physical environment (e.g., land-use, air quality), or socioeconomic (e.g., neighborhood assets). Methods: CA Office of Statewide Healthcare Planning and Development ED data for 2005-2008, restricted to the warm seasons (May 1-October 31) were used. Cases were defined as a HRI diagnosis (International Classification of Diseases (ICD) ninth revision, clinical modification (ICD9-CM) code 992.0-992.9). Non-HRI cases were retained as controls. Exclusions: If date of visit or residence ZIP code (ZC) was missing. ZC were converted to Zip Code Tabulation Area (ZCTA) to link census tract data to patients. Hierarchical generalized linear models with a logic link (SAS v9.3 PROC GLIMMIX) were used. For each exposure (daily Tmax, daily O3 or PM10) models included both the ZCTA-specific 4-year seasonal mean (Tmaxmn), and ZCTA-day (i.e., ED visit date) deviation from that mean (Tmaxdif). Interaction terms were used to evaluate effect modification. Results: HRI risk was positively associated with Tmaxdif (OR 1.02 95%CI: 1.016, 1.019, p<.0001), and co-exposure to O3 (Tmaxdif*O3dif) increased the risk (OR 1.051 95%CI: 1.043, 1.059, p<.0001). Risk varied by age group, sex and race/ethnicity, with girls (especially those ≤ 10 years), African American and Hispanic children, and elderly Black at greatest risk. ZCTA %-impervious surfaces, % multi-family homes were positively associated with HRI. These and other results pointed to interventions to reduce HRI risk.