Emergency-Care Utilisation During Two Healthcare Crises in South Korea: A Single-Centre Comparison of Calendar-Aligned 32-Week Windows

This is the authors' abstract. We don't add key points for this paper.

Background/Objectives: Emergency department (ED) attendance may change differently from subsequent hospital utilisation during healthcare disruptions. We compared two historical crisis episodes within one Korean regional emergency medical centre. Methods: Four calendar-aligned 224-day windows before and after the onset of COVID-19 and the 2024 workforce disruption were analysed using episode-by-phase interactions. Modified Poisson models used patient-and-day two-way clustered variance. Less-conditioned and initial-acuity-conditioned models were compared. Daily counts used calendar-lag autocorrelation-consistent variance. Source-data audits, full-period segmented analyses and alternative-baseline sensitivities were added. Results: The windows included 83,998 visits by 66,096 patients after source-data recoding. The ratio of relative changes was 0.73 (95% confidence interval [CI], 0.68–0.78) for daily visits, 1.28 (1.19–1.37) for transfer-in presentation and 1.44 (1.33–1.55) for acuity-conditioned ICU admission. EMS arrival increased more during COVID-19, but the corrected physiological composite showed no clear between-episode contrast. ED-origin ICU admissions/day were 9.24 before and 9.67 during the workforce episode; the primary ratio of incidence rate ratios was 1.13 (1.05–1.22), but the prior-calendar-year sensitivity included the null (1.06, 0.97–1.15). Workforce-period mortality associations differed before and after acuity conditioning, with adjusted risk ratios of 1.28 (1.10–1.49) and 1.06 (0.92–1.22), respectively. Conclusions: Attendance contracted more than ED-origin ICU admission volume during the observed workforce episode. These findings describe utilisation, not measured capacity pressure, and depend partly on baseline and adjustment choices. The contribution is a multidimensional approach to utilisation surveillance, not universal crisis-specific effects or evidence of unchanged survival.