key: cord-0903354-ma6p3u7z
authors: Westgard, Bjorn C.; Morgan, Matthew W.; Vazquez-Benitez, Gabriela; Erickson, Lauren O.; Zwank, Michael D.
title: An analysis of changes in emergency department visits after a state declaration during the time of COVID-19
date: 2020-06-11
journal: Ann Emerg Med
DOI: 10.1016/j.annemergmed.2020.06.019
sha: 950ffe96c74f319d83b3bcd9b06537ec47beb623
doc_id: 903354
cord_uid: ma6p3u7z

Abstract Objective In the initial period of the COVID-19 pandemic there has been a substantial decrease in the number of patients seeking care in the ED. An initial step in estimating the impact of these changes is to characterize the patients, visits, and diagnoses for whom care is being delayed or deferred. Methods We conducted an observational study, examining demographics, visit characteristics, and diagnoses for all ED patient visits to an urban Level-1 trauma center before and after a state emergency declaration and comparing them to a similar period in 2019. We estimated percent change based on the ratios of before and after periods with respect to 2019 and the decline per week using Poisson regression. Finally, we evaluated whether each factor modified the change in overall ED visits. Results After the state declaration, there was a 49.3% decline in ED visits overall, 35.2% (95%CI: -38.4 to -31.9) as compared to 2019. Disproportionate declines were seen in visits by pediatric and older patients, women, and Medicare recipients as well as for presentations of syncope, cerebrovascular accidents, urolithiasis, abdominal and back pain. Significant proportional increases were seen in ED visits for upper respiratory infections, shortness of breath, and chest pain. Conclusions There have been significant changes in patterns of care-seeking during the COVID-19 pandemic. Declines in ED visits, especially for certain demographic groups and disease processes, should prompt efforts to understand these phenomena, encourage appropriate care-seeking, and monitor for the morbidity and mortality that may result from delayed or deferred care.

As the novel coronavirus 2019 pandemic has spread, state governments and health systems have enacted a range of mitigation strategies and operational changes to anticipate and address an increasing number of patients with severe acute respiratory syndrome coronavirus 2 . At the same time, during the early days of the pandemic, health systems have also seen a decrease in the number of patients presenting for acute care unrelated to COVID-19 1 . Less care sought and received for these acute conditions may put patients at significant risk for preventable morbidity and mortality in the future. The characteristics of those patients who are and are not presenting to emergency departments during the early days of the SARS-CoV-2 pandemic have not yet been substantially examined in the medical literature. We report changes in the characteristics of patients and presentations to the emergency department (ED) of an urban Level-1 Trauma Center before and after the statewide announcement of a "peacetime emergency" and public health measures to respond to the pandemic on March 13, 2020 2 .

We conducted an observational study of visits to the ED of a Level 1 Trauma Center from a period of 28 days before through 28 days after the state's emergency declaration on March 13th. The ED has an annual census of nearly 90,000 and is located in a metropolitan area that over the study period had not yet seen a surge in COVID-19 infections, having seen the state's first confirmed case on March 6th. The declaration on March 13th included announcements about social-distancing measures, the closure of all non-essential business, to begin on March 16, and the closure of all schools, to begin on March 18. The state's first stay-at-home order was declared on March 25. We chose the earliest of these dates to provide the most conservative estimate of the potential effects of such measures on public willingness to seek care. We examined a cross-section of visits from February 15 to April 10, 2020, and for historical comparison, a similar period of weekdays and weekends from February 16 to April 12, 2019. Data were obtained through a systematic query of the electronic health record (EHR) as part of institutional operations and quality improvement and were therefore deemed by the institutional review board to be exempt from review. We obtained demographics, visit characteristics, and diagnoses for all visits during the above periods. Diagnoses were identified by coded individual diagnoses or according to a diagnostic grouper when there was more than one diagnostic code for the disease entities examined. The diagnoses examined were preselected by the authors to include the ten most common conditions presenting to the ED and a list of higher and lower acuity conditions provided by department and hospital administration, all of which are reported.

We estimated the ratios of before and after periods with respect to the same periods in 2019 using Poisson regression by including the period, year and period year interaction as fixed effects. The period year interaction corresponds to the log of the ratio of ratios, and thus we used this estimate to derive the percent change. In addition, we estimated the decline per week by including week as a continuous variable, and the interaction with period, year and period-year

week, corresponding to an interrupted time series analysis. We evaluated the goodness of fit test for this model and scaled the model to correct for over-dispersion. We estimated the percent change with respect to 2019 for the total number of ED visits, by patient characteristics, diagnoses, and the decline per week. We used type III p-values to evaluate whether characteristics and diagnoses modified the ED volume decline by including a second order interaction for each factor with period and year. In addition, we analyzed total ED and total hospital exclusive charges both overall and by acuity using a log normal distribution and a difference in differences to estimate the percent change. All analyses were performed in SAS 9.3 (SAS Institute, Cary, NC).

After the state declaration, the ED experienced a gradual but significant decline from an average of 250 daily visits for the 28 days before to an average of 167 daily visits for the 28 days after. This represented a 7.7% (95% CI: 1.1 to 13.7%) weekly decline in ED visits, a 49.3% decline overall, and a 35.2% (-38.4 to -31 .9%) decline with respect to 2019. We found significant changes in the decline after the state declaration by patient demographics and visit characteristics, particularly in patient age, gender, race, insurance, arrival mode, and disposition.

There were significantly disproportionate declines in ED visits by patients under age 18 (-60.1%) and over age 65 (-41.3%), women (-40.2%), White and Asian patients (-37.8% and 40.2%, respectively), patients with Medicare (-40.8%) and other insurances (-74.1%; e.g. liability, nofault, workman's compensation), as well as ambulatory patients (-38.1%) and those who left prior to evaluation or discharge (-75.6%).

When we examined diagnoses, we noted significantly decreases in the proportions of patients presenting with syncope (-70.5%), cerebrovascular accidents (-58.3%), abdominal pain (-43.3%), urolithiasis (-70.0%), and back pain (-50.7%). We also saw significant increases in the proportions of patients presenting with upper respiratory infections (-10.0%), shortness of breath (25.1%), and chest pain (-13.1%). For all other conditions, the declines in presentations were proportionate to the overall change in ED visits.

When we looked at charges, there was a 32.8% (-36.1 to -29.4%) reduction in total ED charges, and a 23.2% (-27.6 to -18.7%) reduction in hospital charges, with respect to 2019 when adjusted for inflation and without changes in billing or reimbursement rates between periods. Decreases in ED charges by acuity level were generally proportionate to the overall adjusted decline in ED visits, though the decline in ED charges for acuity level 5 visits was smaller (-26.2%, -43.5 to -3.5%). The decline in hospital charges for acuity level 1 visits was disproportionately large (-42.1%, -59.6 to -17.2%), while the decline in hospital charges for acuity levels 2 and 4 visits were smaller than the overall adjusted decline in ED visits (-19.2%, -26.7 to -10.9%; -17.6, -30.9 to -1.7%).

Echoing anecdotal reports, we noted a temporal association between our state's emergency declaration and a gradual but significant overall decline in daily ED visits. We also noted significant though smaller declines in both ED and hospital charges, a difference that may be attributable to changes in acuity distribution. Both phenomenon have been experienced by EDs and health systems across the country, prompting staff furloughs and other responses even as preparations are made for a growing number of patients with SARS-CoV-2.

In addition, we found significant changes in ED patient demographics and visit characteristics.

We found significant proportional decreases in visits by the overlapping populations of patients over 65 and Medicare recipients. There were also significant disproportionate declines in visits by pediatric and ambulatory patients, women, and certain racial categories. We also noted a disproportionate decline in the number of patients who left prior to evaluation or discharge. That change may be due in part to the substantial operational changes made in response to the pandemic. Those changes have further streamlined patient triage, rooming, and evaluation, potentially decreasing triage bottlenecks that can occur when the department is more crowded.

We also found significant changes in the diagnoses of patients presenting to the ED. We saw significant proportional increases in ED visits for upper respiratory infections, chest pain, and shortness of breath. These are symptoms that might bring patients to the ED with concerns for COVID-19 infection, but they may also represent exacerbations of and presentations for other conditions. The most concerning finding of this report, however, is the overall decline in patients seen for acute and potentially life-threatening conditions unrelated to COVID-19 1 . One might expect to find, as we did, a disproportionate decrease in presentations of less acute conditions such as back pain and other non-specific pain. However, we also found unexpectedly disproportionate declines in visits for conditions of substantially higher acuity like syncope, cerebrovascular accidents, and urolithiasis, similar to those noted elsewhere for presentations of myocardial infarction 4, 5 . Similar trends in patient presentations have been seen around the world 6 , during prior disease outbreaks 7 , and in the outpatient setting 8 .

These changes in patient presentation patterns may reflect concerns about contracting COVID-19 in health care settings, overburdening the health care system with unrelated complaints, and adhering to public health recommendations 1 . It is also possible that they are related to changing patterns of activity leading to a change in disease incidence or to the increased availability of other venues for seeking care such as telemedicine. Changes in activity noted during the pandemic might contribute to diminished air pollution, traffic, and infectious disease transmission. Such phenomena could potentially contribute to improved population health over the long term and to a decreased incidence of exacerbations of respiratory conditions, COVID-19 infections 3 , or traumatic injuries in the short term. However, most of the other conditions seen and managed in the emergency department would not be expected to suddenly decrease in incidence. Alternative care venues like telemedicine have also been developed and are expanding, offering valuable ways to provide continued outpatient care. Telemedicine providers can care for many non-acute concerns and ambulatory-care sensitive conditions, but patients with concerning symptoms or significant diagnostic uncertainty may still need to be referred to the ED for evaluation 8 . Most local telemedicine alternatives were not substantially established or bolstered until at least two weeks after the state announcement. As such, telemedicine would not be expected to have a major impact on the number of patients presenting with acute medical conditions over the study period.

These are still early days, and the timeline for the COVID-19 pandemic and resulting changes in patterns of ED utilization promise to be much longer than that covered in our preliminary study. This study is limited by its single center and cross-sectional nature, the short study time frame, and lack of adjustment for multiple potentially confounding factors related to patients and their presentations to the ED, including disease severity, comorbidities, and COVID-19 risk factors. However, we have found notable differential changes in the demographic factors, visits characteristics, and diagnoses of presentations to the ED. Further efforts should and are being made to reassure and affirm the appropriateness of seeking emergency care, 9 particularly for the groups and disease processes that have been highlighted here and elsewhere. 1, 4, 5 Further research will also be needed to examine these and other factors contributing to delayed or deferred care and to monitor for the morbidity and mortality that is likely to result 5, 8 and which may already be occurring 10 . 

Where Are All the Patients?

Fear to Encourage Sick Patients to Seek Emergency Care

Walz advises no gatherings over 250 people as MN COVID-19 cases hit 14. Minnesota Public Radio News

Exposure to air pollution and COVID-19 mortality in the United States

Reduction in ST-Segment Elevation Cardiac Catheterization Laboratory Activations in the United States during COVID-19

Journal of the American College of Cardiology

Outbreak on ST-Segment-Elevation Myocardial Infarction Care in Hong Kong