Travel distance and sociodemographic correlates of potentially avoidable emergency department visits in California, 2006-2010: an observational study

Use of the hospital emergency department (ED) for medical conditions not likely to require immediate treatment is a controversial topic. It has been faulted for ED overcrowding, increased expenditures, and decreased quality of care. On the other hand, such avoidable ED utilization may be a manifestation of barriers to primary care access. A random 10% subsample of all ED visits with unmasked variables, or approximately 7.2% of all ED visits in California between 2006 and 2010 are used in the analysis. Using panel data methods, we employ linear probability and fractional probit models with hospital fixed effects to analyze the associations between avoidable ED utilization in California and observable patient characteristics. We also test whether shorter estimated road distances to the hospital ED are correlated with non-urgent ED utilization, as defined by the New York University ED Algorithm. We then investigate whether proximity of a Federally Qualified Health Center (FQHC) is correlated with reductions in non-urgent ED utilization among Medicaid patients. We find that relative to the reference group of adults aged 35-64, younger patients generally have higher scores for non-urgent conditions and lower scores for urgent conditions. However, elderly patients (≥65) use the ED for conditions more likely to be urgent. Relative to male and white patients, respectively, female patients and all identified racial and ethnic minorities use the ED for conditions more likely to be non-urgent. Patients with non-commercial insurance coverage also use the ED for conditions more likely to be non-urgent. Medicare and Medicaid patients who live closer to the hospital ED have higher probability scores for non-emergent visits. However, among Medicaid enrollees, those who live in zip codes with an FQHC within 0.5 mile of the zip code population centroid visit the ED for medical conditions less likely to be non-emergent. These patterns of ED utilization point to potential barriers to care among historically vulnerable groups, observable even when using rough estimates of travel distances and avoidable ED utilization.