Ambulatory care, insurance, and avoidable emergency department utilization in North Carolina

Objective: To examine whether and how avoidable emergency department (ED) utilization is associated with ambulatory or primary care (APC) utilization, insurance, and interaction effects. Design and sample: A cross-sectional analysis of electronic health records from 70,870 adults residing in Mecklenburg County, North Carolina, who visited an ED within a large integrated healthcare system in 2017. Methods: APC utilization was measured as total visits, categorized as: 0, 1, and > 1. Insurance was defined as the method of payment for the ED visit as: Medicaid, Medicare, private, or uninsured. Avoidable ED utilization was quantified as a score (aED), calculated as the sum of New York University Algorithm probabilities multiplied by 100. Quantile regression models were used to predict the 25th, 50th, 75th, 95th, and 99th percentiles of avoidable ED scores with APC visits and insurance as predictors (Model 1) and with an interaction term (Model 2). Results: Having >1 APC visit was negatively associated with aED at the lower percentiles and positively associated at higher percentiles. A higher aED was associated with having Medicaid insurance and a lower aED was associated with having private insurance, compared to being uninsured. In stratified models, having >1 APC visit was negatively associated with aED at the 25th percentile for the uninsured and privately insured, but positively associated with aED at higher percentiles among the uninsured, Medicaid-insured, and privately insured. Conclusions: The association between APC utilization and avoidable ED utilization varied based on segments of the distribution of ED score and differed significantly by insurance type.