Emergency department visits for emergent conditions among older adults during the COVID-19 pandemic

Abstract

Background/objective: Emergency department (ED) visits have declined while excess mortality, not attributable to COVID-19, has grown. It is not known whether older adults are accessing emergency care differently from their younger counterparts. Our objective was to determine patterns of ED visit counts for emergent conditions during the COVID-19 pandemic for older adults.

Design: Retrospective, observational study.

Setting: Observational analysis of ED sites enrolled in a national clinical quality registry.

Participants: One hundred and sixty-four ED sites in 33 states from January 1, 2019 to November 15, 2020.

Main outcome and measures: We measured daily ED visit counts for acute myocardial infarction (AMI), stroke, sepsis, fall, and hip fracture, as well as deaths in the ED, by age categories. We estimated Poisson regression models comparing early and post-early pandemic periods (defined by the Centers for Disease Control and Prevention) to the pre-pandemic period. We report incident rate ratios to summarize changes in visit incidence.

Results: For AMI, stroke, and sepsis, the older (75-84) and oldest old (85+ years) had the greatest decline in visit counts initially and the smallest recovery in the post-early pandemic periods. For falls, visits declined early and partially recovered uniformly across age categories. In contrast, hip fractures exhibited less change in visit rates across time periods. Deaths in the ED increased during the early pandemic period, but then fell and were persistently lower than baseline, especially for the older (75-84) and oldest old (85+ years).

Conclusions: The decline in ED visits for emergent conditions among older adults has been more pronounced and persistent than for younger patients, with fewer deaths in the ED. This is concerning given the greater prevalence and risk of poor outcomes for emergent conditions in this age group that are amenable to time-sensitive ED diagnosis and treatment, and may in part explain excess mortality during the COVID-19 era among older adults.

Keywords: ED visits; care-seeking; emergency care; older adults.

FIGURE 1

Emergency department (ED) visit counts for select conditions by age category. Smoothed daily visit counts for acute myocardial infarction, stroke, sepsis, fall, and hip fracture are reported by age category. Shaded areas represent the timeframe for the early pandemic period, March 29–April 25, 2020, as defined by the CDC, and the post‐early pandemic (April 26–November 15, 2020). Data are drawn from a national quality registry of community EDs, and the sample includes 164 EDs across 33 states [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 2

Incident rate ratios for emergency department (ED) visits for acute myocardial infarction (AMI), stroke, and sepsis by age category. Each box is an unadjusted Poisson regression model for visit count for select time‐sensitive conditions AMI, stroke, sepsis, fall, and hip fracture among specific age groups with reported incident rate ratio. Error bars represent 95% confidence intervals. The models include categories for pre‐pandemic (January 1, 2019–March 28, 2020), early pandemic period (March 29–April 25, 2020), and post‐early pandemic (April 26–November 15, 2020). Data are drawn from a national quality registry of community EDs, and the sample includes 164 EDs across 33 states [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 3

Emergency department (ED) deaths in ED, counts, and incident rate ratios by age. Smoothed daily counts for deaths in ED are reported by age category. Shaded areas represent the timeframe for the early pandemic period, March 29–April 25, 2020, as defined by the CDC, and the post‐early pandemic (April 26–November 15, 2020). Data are drawn from a national quality registry of community EDs, and the sample includes 164 EDs across 33 states. Each box is an unadjusted Poisson regression model for death in ED counts, among specific age groups with reported incident rate ratio. Error bars represent 95% confidence intervals [Color figure can be viewed at wileyonlinelibrary.com]