Social and demographic characteristics of frequent or high‐charge emergency department users: A quantile regression application

Objective Heavy users of the emergency department (ED) are a heterogeneous population. Few studies have captured the social and demographic complexity of patients with the largest burden of ED use. Our objective was to model associations between social and demographic patient characteristics and quantiles of the distributions of ED use, defined as frequent and high‐charge. Methods We conducted a cross‐sectional analysis of electronic health and billing records of 99 637 adults residing in an urban North Carolina county who visited an ED within Atrium Health, a large integrated health care system, in 2017. Mid‐quantile and standard quantile regression models were used for count and continuous responses, respectively. Frequent and high‐charge use outcomes were defined as the median (0.50) and upper quantiles (0.75, 0.95, 0.99) of the outcome distributions for total billed ED visits and associated charges during the study period. Patient characteristic predictors were: insurance coverage (Medicaid, Medicare, private, uninsured), total visits to ambulatory care during the study period (0, 1, >1), and patient demographics: age, gender, race, ethnicity, and living in an underprivileged community called a public health priority area (PHPA). Results Results showed heterogeneous relationships that were stronger at higher quantiles. Having Medicaid or Medicare insurance was positively associated with ED visits and ED charges at most quantiles. Racial and geographic disparities were observed. Black patients had more ED visits and lower ED charges than their White counterparts at most quantiles of the outcome distributions. Patients living in PHPAs, had lower charges than their counterparts at the median but higher charges at the 0.95 and 0.99 quantiles. Conclusions The relationships between patient characteristics and frequent and high‐charge use of the ED vary based on the level of use. These findings can be used to inform targeted interventions, tailored policy, and population health management initiatives.