Emergency Medicine

Temporal associations between emergency department and telehealth volumes during the COVID-19 pandemic: A time-series analysis from 2 academic medical centers

a b s t r a c t

Background: The COVID-19 pandemic compelled healthcare systems to rapidly adapt to changing healthcare needs as well as identify ways to reduce COVID transmission. The relationship between pandemic-related trends in emergency department (ED) visits and telehealth urgent care visits have not been studied.

Methods: We performed an interrupted Time series analysis to evaluate trends between ED visits and telehealth urgent medical care visits at two urban healthcare system in Colorado. We performed pairwise comparisons be- tween baseline versus each COVID-19 surge and all three surges combined, for both ED and telehealth encounters at each site and used Wilcoxon rank sum test to compare median values.

Results: During the study period, 595,350 patient encounters occurred. We saw ED visits decline in correlation with rising telehealth visits during each COVID surge.

Conclusions: During initial COVID surges, ED visits declined while telehealth visits rose in inverse correla- tion with falling ED visits, suggesting that some patients shifted their preferred location for clinical care. As EDs cope with future staffing during the ongoing COVID pandemic, telehealth represents an opportu- nity for emergency physicians and a means to align patients desires for virtual care with ED volumes and staffing.

(C) 2022

  1. Introduction

The emergence of COVID-19 as a global health threat in early 2020 suddenly and dramatically changed both health needs and the Delivery of care. In response to the growing infectious threat, healthcare systems began planning for local outbreaks by identifying anticipated chal- lenges, including decreasing patient volume and the inability to provide some care by traditional means [1]. In February 2020, the Centers for Disease Control and Prevention (CDC) responded to the grow- ing COVID threat and advised healthcare systems to adopt safety protocols, such as telehealth, that reduce or eliminate potential infectious exposure [2].

Telehealth is the use of two-way telecommunication technology that allows clinicians to provide care remotely. Telehealth showed rapid expansion through the initial portion of the pandemic with

* Corresponding author at: Department of Emergency Medicine, University of Colorado School of Medicine, 12505 E 16th Avenue, Aurora, CO 80045, United States of America.

E-mail address: [email protected] (E.M. Reno).

telehealth visits in some healthcare systems growing by over 50% in the first quarter of 2020, as compared to 2019 [3,4].

Concurrently with telehealth’s increasing adoption in early 2020,

emergency department (ED) visits decreased substantially. This decline was caused by at least three types of behavioral change: (1) exposure to all types of disease and injury was reduced, as schools and workplaces closed and most non-essential travel was suspended; (2) sick and in- jured patients avoided seeking medical care due to concerns for contracting COVID-19; and (3) routine or elective care was postponed [5,6].

While evidence suggests that telehealth volume increased nationally during the COVID-19 pandemic, it’s unclear how much of this increase was from acute unscheduled healthcare, and how much was from a transition of routine outpatient care to a telehealth model to reduce in- fectious exposure [3,4]. The correlation with telehealth use and ED vol- umes has not yet been established.

The goal of this study was to evaluate the temporal association be- tween emergent/urgent telehealth utilization and ED volume through- out three COVID surges in Colorado.

https://doi.org/10.1016/j.ajem.2022.01.046

0735-6757/(C) 2022

  1. Methods
    1. Study design

We performed an interrupted time series analysis to evaluate associ- ations between COVID-19 case surges and number of daily ED and tele- health encounters in two large, urban healthcare systems in Colorado from January 1, 2019 through December 31, 2020. The study was ap- proved by the Colorado Multiple Institutional Review Board and granted a waiver of consent, and is reported in accordance with the Transparent Reporting of Evaluations with Nonrandomized Designs guidelines.

    1. Setting

Denver Health (DH) is a large, urban public health care system in Denver, Colorado whose main hospital, Denver Health Medical Center, includes both adult and pediatric EDs and urgent cares with >125,000 combined annual patient encounters. Denver Health also provides the NurseLine, a free, 24-h acute unscheduled call-in medical service avail- able to all Colorado residents. NurseLine calls are received by Trained nurses who can: (1) treat lower-risk conditions via protocolized care (e.g. antibiotics for Uncomplicated urinary tract infections); (2) refer for urgent or emergent in-person follow-up; or (3) consult the NurseLine on-call emergency physician.

UCHealth is a large, urban academic health care system in Aurora, Colorado whose main hospital, University of Colorado Hospital (UCH), includes an adult ED, which has >100,000 annual patient encounters. UCHealth offers a fee-for-service, 24-h video-based Virtual Urgent Care to all Colorado residents. Encounters are staffed by advanced prac- tice providers and emergency physicians, who can treat a variety of lower-risk conditions, or refer patients for in-person follow-up.

Both the Denver Health NurseLine and UCHealth Virtual Urgent Care served to triage calls relating to urgent medical needs during the COVID- 19 pandemic, resulting in medical advice and referrals for testing and in-person urgent and emergent follow-up.

    1. Study sample

All ED and telehealth encounters from January 1, 2019 through December 31, 2020 were included, regardless of age. There were no exclusions.

    1. Time series

COVID-19 case surges were chosen as interruptions of the time se- ries and were identified by upward inflection points in Colorado’s COVID-19 epidemiological case data to correspond with the following four distinct time intervals: [1] January 1, 2019 through March 24,

2020 (pre-COVID-19 baseline); (2) March 25, 2020 through June 15, 2020 (first surge); (3) June 16, 2020 through September 12, 2020 (sec- ond surge); and (4) September 13, 2020 through December 31, 2020 (third surge). The study period was chosen to begin on January 1, 2019 to measure a pre-COVID-19 baseline with an adequate duration to account for secular trends. The first interruption was identified as March 25, 2020, where an upward inflection point of reported cases in Colorado was demonstrated, and also the date when a state-wide stay-at-home order was implemented. Second and third surge time pe- riods were chosen based on the transition point from downward to up- ward inflections in the case data [7]. Date of encounter, age, and gender were extracted from each electronic health system (Epic, Epic Systems Corporation, Verona, WI).

    1. Statistical analysis

Descriptive statistics were used to summarize the demographic characteristics of patients and stratified by visit type and time period.

Median values with 95% confidence intervals (CIs) were reported for each time series. We further performed comparisons between baseline versus each COVID-19 surge and all three surges combined, for both ED and telehealth encounters at each study site [8]. We used the Wilcoxon rank sum test to compare median values, while also estimating the dif- ferences between medians with 95% CIs. Medians were chosen as the comparator for the effect estimates, as opposed to level (y-intercept of each series) and trend (slope), because our intention was to compare the overall effect of COVID-19 on each time series rather than the imme- diate effects or rates of change. We also estimated the differences be- tween the changes in baseline versus each COVID-19 surge and all surges combined of median daily encounters between ED and tele- health visits at each institution [8]. This method of analysis was utilized in order to compare the effect of COVID-19 on two unscheduled, con- ceptually inverse encounter types, within two geographically similar but operationally distinct healthcare systems. All analyses were per- formed using SAS Enterprise Guide Version 7.1 (SAS Institute Incorpo- rated, Cary, NC).

  1. Results
    1. Study sample characteristics

During the study period, 595,350 total encounters occurred. Stratifi- cation by time series and setting, including baseline characteristics, are summarized in Appendix A.

    1. Main results

Results of the interrupted time series analysis demonstrated signifi- cant differences (all p < .0001) of median daily encounters between baseline versus each COVID-19 surge, and baseline versus all surges combined, in both ED and telehealth visits at Denver Health and UCHealth (Table 1).

Table 1

Interrupted time series analysis comparing differences of median daily encounters be- tween baseline and COVID-19 surges, stratified by setting.

Daily encounters

n Median (95% CI) ? (95% CI) p

Denver Health ED

Baseline

154,963

346

(342-349)

Ref

Ref

1st Surge

18,026

219

(205-229)

-127

(-139 to -115)

<0.0001

2nd Surge

24,957

281

(275-289)

-65

(-73 to -57)

<0.0001

3rd Surge

30,898

284

(279-293)

-62

(-70 to -54)

<0.0001

All surges Telehealth

Baseline

73,881

79,613

269

181

(261-276)

(176-184)

-77

Ref

(-91 to -59)

<0.0001

Ref

1st Surge

21,341

257

(248-269)

76

(65-87)

<0.0001

2nd Surge

20,484

227

(222-240)

46

(35-57)

<0.0001

3rd Surge

26,415

241

(224-253)

60

(45-71)

<0.0001

All surges

68,240

243

(236-251)

62

(44-80)

<0.0001

UCHealth ED

Baseline

123,244

276

(274-279)

Ref

Ref

1st Surge

17,157

207

(198-212)

-69

(-77 to -62)

<0.0001

2nd Surge

21,516

243

(237-247)

-33

(-40 to -26)

<0.0001

3rd Surge

26,898

245

(241-250)

-31

(-37 to -25)

<0.0001

All surges

65,571

233

(229-240)

-43

(-55 to -25)

<0.0001

Telehealth

Baseline

7227

9

(8-10)

Ref

Ref

1st Surge

8859

97

(82-113)

88

(73-103)

<0.0001

2nd Surge

5331

60

(56-64)

51

(48-55)

<0.0001

3rd Surge

8381

74

(69-78)

65

(61-69)

<0.0001

All surges

22,571

70

(67-73)

61

(54-73)

<0.0001

Abbreviations: CI = confidence interval; ED = emergency department; IQR = inter- quartile range; Ref = reference.

Table 2

Difference-in-differences of median daily ED and telehealth encounters from baseline to each COVID-19 surge, stratified by setting.

Difference in differences (encounters)

(95% CI)

p

Denver Health

Baseline vs 1st surge

-51

(-55 to -47)

<0.0001

Baseline vs 2nd surge

-19

(-23 to -15)

<0.0001

Baseline vs 3rd surge

-2

(-6-1)

0.2113

Baseline vs all surges

-15

(-19 to -11)

<0.0001

UCHealth

Baseline vs 1st surge

19

(15-23)

<0.0001

Baseline vs 2nd surge

18

(14-22)

<0.0001

Baseline vs 3rd surge

34

(30-38)

<0.0001

Baseline vs all surges

18

(14-21)

<0.0001

Abbreviations: CI = confidence interval; vs = versus.

The difference-in-differences analysis showed that changes in me- dian daily ED versus telehealth encounters at Denver Health differed be- tween baseline versus 1st surge (p < .0001), baseline versus 2nd surge (p < .0001), and baseline versus all surges combined (p < .0001), but not between baseline versus 3rd surge (p = .21). Results of the difference-in-differences analysis of UCHealth were all significant (all p < .0001) (Table 2).

Fig. 1 illustrates the results of both the interrupted time series and difference-in-differences analyses, using multiple line graphs with 95% CIs of median daily ED and telehealth encounters at Denver Health and UCHealth. Across both institutions, an inverse direction- ality existed between median ED versus telehealth encounters from the pre-COVID-19 baseline to the first surge, as ED encounters de- creased, while telehealth encounters increased. The first to second surge demonstrated the same inverse relationship between ED and telehealth encounters, but in the opposite directions, as ED encoun- ters increased, while telehealth encounters decreased. From the second to third surge, all ED and telehealth encounters increased. Scatter plots demonstrating total encounters per day, stratified by time series, in each of the four study settings, is included in Appendix B.

  1. Limitations

The study period ended during the third COVID-19 case surge, and while it included the period of greatest incidence in the third surge and overall, incorporating the end of third surge, in addition to later surges, could have identified additional longitudinal trends in encoun- ter volumes. Our analysis included descriptive statistics of age and gen- der, but did not evaluate associations between these or other demographic nor socioeconomic variables (e.g. Insurance type, zip code) with changes in ED or telehealth encounters. In addition, our anal- ysis did not include encounter-level variables, including chief com- plaint, duration of encounter, and telehealth encounter disposition (e.g. advised to go to ED).

  1. Discussion

Our study demonstrated that significant changes occurred in acute unscheduled in-person and telehealth encounter volumes within two unique healthcare systems during the COVID-19 pandemic. We identi- fied immediate and overall decreases in ED volumes, consistent with previously described data [6]. Reasons for this drop in visits are likely

Fig. 1. Median (95% confidence intervals) daily emergency department (ED) and telehealth encounters at baseline and each COVID-19 surge, by institution.

multifactorial. stay-at-home orders may have contributed to a de- creased spread of other community acquired illnesses, a reduction in some injuries (e.g. car accidents), and a desire for patients to avoid emergency departments for fear of contracting COVID. Of note, certain Types of violence including firearm and domestic violence rose during stay-at-home orders [9]. Cancelling of surgical procedures likely resulted in decreased post-operative visits.

While ED volumes decreased, this study showed concurrent 34% and 678% increases in median acute unscheduled telehealth volumes across the two sites. This is noteworthy, as it indicates that patients still sought emergency and urgent medical care, but rapidly transitioned from the ED to telehealth models. Of note, these visits were independent of any ambulatory clinic visits that transitioned to telehealth as they served urgent and acute medical needs. Additionally, staffing this surge was sourced solely by ED providers.

As a result, acute unscheduled telehealth presents a unique opportu- nity for EDs, both in terms of managing patient influx, and to dynami- cally manage staffing. With the ongoing COVID-19 pandemic and with future pandemics, patient volumes may continue to fluctuate. Tele- health can be used to rapidly change staffing models, transitioning phy- sicians to and from a virtual setting as needed to meet patient demand and ensure appropriate staffing in the ED. Many EDs decreased staffED shifts during the initial surges of the pandemic, and telehealth shows that physicians can be transitioned to alternative Clinical sites to decrease staffing cuts [10].

Emergency physicians have been readily adaptable in clinical prac-

tice due to the evolving and unpredictable nature of the ED, highlighted by the growing adoption of acute telemedicine services at many hospi- tals. Despite this adaptability, the emergency medicine workforce will likely face a physician surplus over the next decade and reducED patient volume from events like the COVID-19 pandemic could exacerbate this oversupply. Emergency medicine physicians, with their inherent adapt- ability are uniquely poised to understand and adjust to changing clinical care technology.

Author contributions

All authors conceived the study. CD, ME obtained data. BL, JH ana- lyzed data and performed statistically support. ER and BL drafted the manuscript. BS provided project oversight. All authors contributed to the revision of the manuscript.

Funding

This research did not receive any specific grant from funding agen- cies in the public, commercial, or not-for-profit sectors.

Credit authorship contribution statement

Elaine M. Reno: Conceptualization, Data curation, Project adminis- tration, Supervision, Writing – original draft, Writing – review & editing. Benjamin Li: Writing – review & editing, Formal analysis, Data curation, Conceptualization. Morgan Eutermoser: Conceptualization, Data curation, Supervision. Christopher B. Davis: Supervision, Data curation, Conceptualization. Jason S. Haukoos: Writing – review & editing, For- mal analysis, Supervision. Bradley Shy: Writing – review & editing, Writing – original draft, Supervision, Resources, Investigation, Data curation, Conceptualization.

Declaration of Competing Interest

The authors report no conflicts of interest as it pertains to this project.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2022.01.046.

References

  1. Subba SH, Pradhan SK, Sahoo BK. Empowering primary healthcare institutions against COVID-19 pandemic: a health system-based approach. J Family Med Prim Care. 2021;10(2):589-94. https://doi.org/10.4103/jfmpc.jfmpc_1416_20.
  2. https://www.cdc.gov/coronavirus/2019-ncov/hcp/facility-planning-operations. html?CDC_AA_refVal=https%3A%2F%2Fwww.cdc.gov%2Fcoronavirus%2F2019-ncov

%2Fhcp%2Fguidance-hcf.html.

  1. Koonin LM, Hoots B, Tsang CA, et al. Trends in the use of telehealth during the emer- gency of the COVID-19 pandemic – United States, January-March 2020. MMWR Morb Mortal Wkly Rep. 2020;69:1599. https://doi.org/10.15585/mm6943a3.
  2. Monaghesh E, Hajizadeh A. The role of telehealth during Covid-19 outbreak: a sys- tematic review based on current evidence. BMC Public Health. 2020;20(1):1193. Published 2020 Aug 1. https://doi.org/10.1186/s12889-020-09301-4.
  3. Czeisler ME, Marynak K, Clarke KE, et al. Delay or avoidance of medical care because of COVID-19-related concerns-United States, June 2020. MMWR Morb Mortal Wkly Rep. 2020;69:1250-7. https://doi.org/10.15585/mmwr.mm6936a4.
  4. Hartnett KP, Kite-Powell A, DeVies J, et al. Impact of the COVID-19 pandemic on emergency department visits – United States, January 1, 2019-May 30, 2020. MMWR Morb Mortal Wkly Rep. 2020;69(23):699-704. Published 2020 Jun 12. 10.15585/mmwr.mm6923e1.
  5. Colorado Department of Public Health and Environment. Colorado COVID-19 data. Accessed April 4, 2020. https://www.covid19.colorado.gov/data; April 4, 2020.
  6. Warton EM. Time after time: Difference-in-differences and interrupted time series models in SAS. Proceedings of the SAS Global Forum 2020; 2020 Available at. https:// www.sas.com/content/dam/SAS/support/en/sas-global-forum-proceedings/2020/4 674-2020.pdf.
  7. Boserup Brad, McKenney Mark, Elkbuli Adel. Alarming trends in US domestic vio- lence during the COVID-19 pandemic. Am J Emerg Med. 2020;38(12):2753-5.
  8. www.washingtonpost.com/health/2021/01/04/er-doctors-covid-jobs/.

Leave a Reply

Your email address will not be published. Required fields are marked *