Emergency Departments (EDs) worldwide are increasingly utilized for potentially non-urgent cases, imposing substantial clinical and economic burdens. In Israel, there is limited nationwide data evaluating the scope and predictors of potentially non-urgent ED utilization. This study aimed to evaluate the extent and associated determinants of potentially non-urgent ED visits in Israel. This nationwide, retrospective multi-center study analyzed 2,843,190 ED visits recorded between January 2018 and April 2022, from all general hospitals operated by Clalit Health Services and one additional public hospital (Assuta Ashdod Public Hospital), which together serve patients insured by all four Health Maintenance Organizations (HMOs) in Israel, and included only treat-and-release ED visits that did not result in inpatient admission or in-ED death. Visits were classified as urgent or potentially non-urgent using a validated multi-parameter classification algorithm. Multivariable logistic regression and Generalized Estimating Equation (GEE) models were applied to identify factors associated with potentially non-urgent visits and to account for repeated visits by the same patients. Among treat-and-release ED visits (discharged without inpatient admission or in-ED death), 49.4% were classified as potentially non-urgent. Younger age, female gender, self-referral, shorter visit duration, non-specific presenting complaints, and diagnoses in categories such as musculoskeletal disorders, dermatologic conditions were significantly associated with potentially non-urgent visits. The model showed good discriminatory ability (AUC = 0.79), and GEE analysis confirmed these associations remained robust across repeated visits. Nearly half of treat-and-release ED visits (discharged without inpatient admission or in-ED death) in our sample were classified as potentially non-urgent and might be managed in community or other more cost-effective care settings. A multifaceted approach is required, including strengthening primary care services, improving triage processes, enhancing public education, and using classification and predictive models mainly for planning and evaluation to optimize ED utilization and resource allocation.
Leshinski et al. (Sat,) studied this question.