Concordance in Medical Urgency Classification of Discharge Diagnoses and Reasons for Visit

Abstract

Current policies to divert emergency department (ED) visits for less medically urgent conditions to more cost-effective settings rely on retrospective adjudication of discharge diagnoses. However, patients present to the ED with concerns, making it challenging for clinicians. To characterize ED visits based on the medical urgency of the presenting reasons for visit and to explore the concordance between discharge diagnoses and reasons for visit. In this retrospective, cross-sectional study, a nationwide sample of ED visits by adults (aged ≥18 years) in the US from the 2018 and 2019 calendar years' ED data of the National Hospital Ambulatory Medical Care Survey was used. An algorithm to probabilistically assign ED visits into medical urgency categories based on the presenting reason for visit was developed. A 3-step, look-back method was applied using an updated version of the New York University ED algorithm, and a map of all possible discharge diagnoses to the same reasons for visit was developed. Analyses were conducted in July and August 2023. The main outcome was probabilistic medical urgency classification of reasons for visits and discharge diagnoses and their concordance. We analyzed 27 068 ED visits (mean age, 48.2% years [95% CI, 47.5%-48.9% years]) representing 190.7 million visits nationwide. Women (mean, 57.0% [95% CI, 55.9%-58.1%]) and patients with public health insurance coverage, including Medicare (mean, 24.9% [95% CI, 21.9%-28.0%]) and Medicaid (mean, 25.1% [95% CI, 21.0%-29.2%]), accounted for the largest share of ED visits, and a mean of 13.2% (95% CI, 11.4%-15.0%) of all visits resulted in a hospital admission. Overall, about 38.5% and 53.9% of all ED visits were classified with 100% and 75% probabilities, respectively, as injury related, emergency care needed, emergent but primary care treatable, nonemergent, or mental health or substance use disorders related based on discharge diagnosis compared with 0.4% and 12.4%, respectively, of all encounters based on patients' reason for visit. Among discharge diagnoses assigned with high certainty to only 1 urgency category using the New York University ED algorithm, between 38.0% (95% CI, 36.3%-39.6%) and 57.4% (95% CI, 56.0%-58.8%) aligned with the probabilistic categorical assignments of their corresponding reasons for visit. In this cross-sectional study of 190.7 million ED visits among adults aged 18 years or older, a smaller percentage of reasons for visit could be prospectively categorized with high accuracy to a specific medical urgency category compared with all visits based on discharge diagnoses, and a limited concordance between reasons for visit and discharge diagnoses was found. Alternative methods are needed to identify the medical necessity of ED encounters more accurately.