Exploring the Relationships of Social and Environmental Determinants of Health and Emergency Department Visit Categories Open Access
Downloadable ContentDownload PDF
Limited knowledge exists about the relationship that specific social determinant of health (SDH) factors have with ED visits and ED visit categories. Following the PRSIMA and AMSTAR guidelines, the first aim of this research project was to conduct a systematic literature review to determine what evidence exists about the relationship between area-level SDH factors and ED visits. The second aim statistically assessed the associations that social and environmental determinants of health (SEDH) factors have with ED visits and the four main NYU-Billings algorithm visit categories through the estimation of single-level and multilevel linear regression models. We assessed SEDH variables against five ED-level outcomes: total annual ED visits, non-emergent ED visits, primary care treatable ED visits, preventable ED visits, unavoidable ED visits, and county-level rates of ambulatory care sensitive hospitalizations (ACSHs) for patients admitted through the ED. Informed by the first research aim, the estimated models’ structures included 17 separate SEDH variables, and 14 county and ED-hospital control variables (e.g. percentage of county residents age 65 and older, hospital size).The systematic literature review identified an initial 4,666 PubMed and EconLit search results of which 76 assessments between an individual SDH factor and an ED visit outcome (across 43 studies) met our inclusion criteria. In synthesizing these results, we found that the most commonly assessed SDH factors were area-level income and deprivation indices. Although some limited conclusions could be drawn about these two factors, many other SDH factors were either assessed too infrequently or exhibited conflicting evidence such that conclusions about the relationships’ significance and directionality could not be concluded with confidence. However, we did find some intermediary SDH factors (e.g. unemployment) that exhibited consistent directionality despite a limited number of assessments, and thus indicated areas where future additional research could be directed to confirm the nature of the relationships.For the second aim, we obtained ED treat-and-release data from the Healthcare Cost Utilization Project State (HCUP) Emergency Department databases for 2010 from five states (i.e. CA, FL, IA, NY, NC), and used the NYU-Billings algorithm to identify our categories of treat-and-release ED visits. For ED patients that were admitted to the hospital, we obtained the 2010 HCUP State Inpatient Databases from the same five states and used the Agency for Healthcare Research and Quality’s Prevention Quality (PQI) software to create our outcome for ACSHs for patients admitted from the ED. We estimated single and multilevel linear regression models and selected the preferred model for each of our five outcomes. We found multiple SEDH variables to be significantly related to each of the preferred outcome models. For several of the SEDH variables, we observed significance across multiple outcomes (e.g. unemployment, healthy food index). Notably, we have found that in most cases the intermediary SEDH variables had a larger magnitude association with our ED visit outcomes than the more commonly assessed structural SEDH variables (i.e. income, educational attainment, race and ethnicity). This exploratory study lays the foundation to direct and inform future assessments of SEDH factors as they relate to ED utilization.