Aim design and rationale

In the first phase of our work, we will conduct in-depth interviews with emergency physicians to explore the facilitators and barriers to the adoption of COVID-19 guidelines into practice amidst a changed healthcare sector landscape. Using a sociotechnical systems framework,34 we will also conduct detailed field observations in the ED to describe guideline and checklist use at the bedside and identify potential unintended systems consequences of guideline adoption including risks and hazards to patient and clinician safety. In response to the COVID-19 pandemic, local task force teams have iteratively designed and implemented a series of new guidelines, clinical decision pathways, and companion checklists to support both patient and clinician safety and ensure compliance with standards of care. However, rapid changes in knowledge of COVID-19 disease processes and presentations as well as resource availability require frequent updates to these practice guidelines. While daily updates from local task force teams are delivered via email and print materials, the rate and systemic adoption of these protocols are still unknown.

Participants and setting

In months 1–6 of the study period, beginning in January 2021, we will recruit ED healthcare providers (attendings, residents, nurses and technicians) with experience treating COVID-19 (and suspected COVID-19) patients as participants in 30–45 min qualitative interviews. Eligible participants shall include all providers having worked at least one clinical shift caring for patients with COVID-19 since March 2020. All participants will be targeted via email and recruitment fliers with QR codes that will be posted in the ED. Participation will be on a voluntary basis. Observations will be conducted at the Yale-New Haven Health (YNHH) ED, situated in a busy, 1541-bed, urban, tertiary care referral academic hospital in New Haven, Connecticut, USA. Both resident and attending physicians actively working at the YNHH ED will be enrolled as part of the system observations. Participants will provide verbal consent at the time of recruitment and scheduling for interviews and/or observations.

Qualitative interviews with resident and attending physicians will be based on normalisation process theory (NPT),35 a sociological theory that identifies, characterises, and explains key mechanisms that promote and inhibit the implementation, embedding and integration of new health innovations, technologies and clinical practices. Quality experts have increasingly used NPT to inform feasibility studies and adoption evaluation of complex healthcare interventions.36 Interview topics (table 1A) will organise around the four core constructs within NPT to explore barriers and facilitators of guideline adoption as well as factors contributing to clinician stress and burnout. All interviews will be audio recorded and professionally transcribed weekly for analysis.

Field observations in the ED will be conducted on a biweekly basis for randomised 8-hour blocks throughout the baseline study period. Observers will further assess the adoption of COVID-19 clinical care guidelines by completing one or more of four established guideline checklists. Using a sociotechnical systems framework, we will use field notes to capture systems features related to guideline use and care provision (table 1B). A human factor expert and a trained research associate will double code a subset of observations to assess inter-rater reliability. We will calculate a Cohen’s kappa for inter-rater reliability.

Outcomes

From Aim 1, we will obtain qualitative data on preparedness, adoption of COVID-19 care guidelines, and perceived risks and hazards to clinician and patient safety through qualitative interviews with emergency physicians and human factors observations of ED physician shifts. These data will inform the refinement of CRI:SIS to ensure that we embed information that will improve preparedness and mitigate stress during clinical care for our participants.

Data analysis

Qualitative interviews will be analysed for emergent themes of adoption facilitators and barriers. We will use Dedoose (SocioCultural Research Consultants),37 a collaborative qualitative software package, for thematic analysis and data organisation. We will start with a systematic, inductive approach through an initial round of open coding, then achieve consensus on major themes38 through an iterative analytical process as more information is added after additional interviews using the constant comparative method from grounded theory.39 Field notes will be analysed using a deductive coding method based on the sociotechnical systems approach.34 Items marked as completed or deficient on the checklist during field observations will be aligned with observed system factors to identify critical system interactions impacting adoption. We will stop interviews and field observations when we reach data saturation, an accepted technique for ensuring richness of data in qualitative research.40 We approximate that this will occur at 25–30 physician interviews and 75–80 hours of observations.

Aim study design and rationale

Our innovative approach applies a fully adaptive simulation programme that can rapidly shift between remote virtual telesimulation and in-person modalities developed and implemented by our team. High-fidelity simulation activates participants’ emotional or affective state and allows the development of necessary cognitive and psychomotor skills in clinical practice.41 To achieve similar benefits, we created our virtual telesimulation technology with the goal of retaining as many cognitive and affective learning features of the live simulation environment as possible while adapting the simulation experience to a virtual videoconferencing platform.42 This rapidly adaptive format allows us to maintain continuity of simulation delivery while responding to fluctuations in local public health restrictions, including business closures and social distancing. Applying this rapidly adaptive approach, CRI:SIS will address healthcare worker stress through preparedness and engagement. Experts have increasingly raised concerns regarding the wellness of the front-line health workers that are directly diagnosing and managing critically ill patients during this pandemic.6 However, much of the current attention in clinical research is currently focused on healthcare system preparedness, diagnostic testing, and medical treatment of patients with COVID-19.43 Our simulation scenarios embedded within CRI:SIS will focus on health worker preparedness to mitigate physician anxiety and stress as our main goal and use health worker physiological measures as our primary outcome of interest.

This section of our study will be a randomised controlled trial assessing the impact of CRI:SIS as a simulation-based preparedness intervention on physician stress levels through changes to HRV on shift while caring for patients with COVID-19 and post-shift anxiety as measured by STAI. In order to address patient safety in the current pandemic, we must first address the safety needs of our clinicians caring for these patients. Clinicians require support along a range of psychological needs, including basic safety through proper use of personal protective equipment (PPE), social support fostering teamwork, and preparedness for clinical challenges, including difficult conversations regarding patient care.14 If these needs are not met, COVID-19 presents increased risk of infection, anxiety and burnout for clinicians.8 14 The syndrome of burnout in physicians increases risk of patient safety incidents, extends patient waiting times and reduces patient satisfaction.44 45 Evidence suggests links between physiological measures of acute stress and the emotional exhaustion subscale of burnout.16 During the pandemic, our team has developed and piloted a COVID-19 simulation intervention designed to support preparedness for the clinical stressors physicians will likely encounter caring for patients with COVID-19 in the ED. We aim to test CRI:SIS to determine its effect on mitigating physician stress through a two-arm randomised controlled trial at two clinical sites with a primary outcome of change in on-shift HRV as a physiological measure of stress and a secondary outcome of post-shift anxiety related to the care of patients with COVID-19 as measured by STAI,33 a commonly used measure of trait and state anxiety in clinical settings to diagnose anxiety and in research as an indicator of participant distress.

Participants and setting

In months 3–12, we will recruit resident or attending physicians actively treating acutely ill patients with COVID-19 (and suspected COVID-19) at either of the two YNHH ED campuses. These two campuses are geographically and structurally unique academic ED sites: (1) York Street Campus, the tertiary care referral centre with four resuscitation bays, 56 beds, and average adult volumes of 100 000 visits per year; and (2) St. Raphael Campus, an urban community hospital with two resuscitation bays, 35 beds and 65 000 visits per year. Eligible participants shall be working full-time in one of these two EDs with an anticipated average of 26 hours/week (three to four shifts/week) for attending physicians and 45–50 hours/week (four to five shifts/week) for resident physicians. Eligible participants will be selected for recruitment based on proximity to working clinical shifts in the ED at either one of the two YNHH campuses and contacted via email. Participation will be on a voluntary basis. Exclusion criteria will include use of a beta blocker and/or antiarrhythmic medication, active thyroid dysfunction and pregnancy.

Protocol and randomisation

We will enroll eligible participants across a period of 12–15 months. At the time of enrolment, participants will completea written consent, then will be fitted with an appropriately sized Hexoskin smart shirt. Participants will be asked to sit quietly for a 5 min baseline session to capture each individual’s baseline heart rate and HRV. The Hexoskin shirt contains three sensors to capture participant heart rate via ECG, respiratory rate, minute ventilation and movement/activity. This non-invasive technology will capture moment-by-moment physiological data necessary to calculate HRV changes as measures of stress during the care of patients with COVID-19 and suspected COVID-19. Following capture of these baseline data, participants will be asked to complete the 40-item STAI survey. All participants will wear a Hexoskin smart shirt underneath their standard scrub shirt for four consecutive shift data collections. Following the completion of the first two data collection shifts, participants will be randomised to either the control or intervention arm and scheduled for the remaining study sessions (two additional data collection shifts±the intervention session). Participants will be divided into two groups based on experience level, junior (PGY1-3) and senior (PGY4+). Prior to the start of the clinical trial, numbered envelopes with random intervention and control designations will be prepared by a statistician and evenly split into junior and senior groupings. After completion of two shifts, a corresponding envelope will be opened, designating the participant to either the control or intervention grouping. For participants randomised to the intervention arm, data collections will consist of two shifts prior to the intervention and two shifts following the intervention. A research assistant will be present at the start of each participant’s shift to confirm data capture and log shift start time on the recording. Following end-of-shift sign-out, a research assistant will administer the 20-item state subscale of the STAI and conduct a debriefing of 5–10-min with the participant to capture qualitative data on perceived stressors experienced during the shift. Participants will be asked to continue HRV data recording for 20 min following the end of each data collection shift to assess for return to baseline HRV. The primary outcome of interest will compare the change in HRV from baseline to the last 5 min of the shift prior to sign-out as a measure of cumulative shift stress. Additionally, post-processing of the data will capture 5 min readings during the treatment of each acutely ill medical patient based on time logs of medical patient alerts as a measure of acute stress during the shift (figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Example of heart rate and event-related heart rate variability during emergency department physician shift. BPM, beats per minute.

Simulation intervention

Participants randomised to the intervention arm will receive CRI:SIS as a 3-hour simulation session. This session will include three scenarios focused on three critical areas of care for patients with COVID-19: (1) airway management procedures in patients with COVID-19 given increased risk of viral transmission to personnel and rapid respiratory deterioration in infected patients46 47; (2) new presenting symptoms and associated complications of COVID-19 (eg, hypercoagulability and cardiovascular morbidity), making accurate diagnosis and treatment of patients with suspected infection difficult48–50; and (3) caring for patients presenting with severe illness and poor prognosis adding emotional and cognitive strain to physicians as they initiate palliative care, discuss goals of care or withdraw care in the ED. In addition, all three scenarios will address negative effects on team performance during COVID-19 care from social distancing and PPE requirements through interactions with nursing and ancillary staff confederates during each scenario. We will refine our simulation scenarios from our pilot programme to reflect qualitative data obtained in the first phase of the project. Each participant will complete all three scenarios within a 3-hour block between 1 and 5 days prior to a clinical shift. This simulation session will be offered on Tuesdays to clinicians, educators, and administrators recruited from outside institutions who are experiencing a surge in COVID-19 presentations in their geographical region (see Ethics and dissemination section).

Control

Participants randomised to the control arm will participate in four shift data collections, with no additional intervention. These participants will have access to the routinely distributed COVID-19 Task Force updates, guidelines, weekly town hall meetings, and any in-service support that would routinely be available to all clinical staff as per standard operational practice in our local departments. Once enrolment is complete, all participants randomised to the control arm will be offered the opportunity to complete the simulation intervention.

Outcomes

We have previously developed a rapid cycle implementation and evaluation of a novel virtual tele-simulation intervention to improve patient and clinician safety during COVID-19. If proven successful, CRI:SIS is readily scalable and applicable at other institutions to improve and evaluate the responsiveness of healthcare delivery systems and healthcare professionals to the COVID-19 pandemic.

Data analysis

Data processing of HRV will be completed with Vivosense software to validate signal quality and mark ECG R-waves for analysis. HRV, our primary outcome of interest, will be assessed as the time-domain measure of root mean square standard deviation (RMSSD) of sequential R–R intervals. RMSSD is considered a measure of vagally mediated change and is more resistant to respiratory artefact than other HRV measures.51 We will analyse HRV using a short-term period of 5 min. These HRV periods will be captured at baseline, immediately following oncoming shift sign-out, during on-shift medical patient responses, ahead of the end-of-shift shift sign-out and post-shift. The primary outcome of interest will be measured as the change from baseline to the 5 min period prior to end-of-shift shift sign-out as a measure of cumulative shift stress. Additional analyses will examine acute stress as changes in HRV at the presentation of acutely ill medical patients during the shift. Timing for the treatment of individual patients will be captured by aligning Hexoskin timestamps with electronic timestamps of medical alerts from the ED’s electronic paging system. Electronic health record review will allow for the identification of COVID-19-positive and suspected COVID-19 medical patients. An increase in stress will be assessed as a decrease in HRV from baseline. Changes in HRV will be averaged over the two shift data collections to control for an anticipated shift-to-shift and patient-to-patient variability in stress response due to patient acuity and workload. Our second outcome of interest is the change in anxiety between the intervention and control conditions as measured by the STAI. We will assess change in both primary outcomes of HRV and STAI using repeated measures mixed models. Baseline measures will be included as a covariate. A random effect will be included for the subject variable to accommodate for repeated measures. Least squares means will be used to describe HRV and other outcomes under each intervention. Changes in HRV will be averaged over the two post-intervention shift data collections to control for an anticipated shift-to-shift and patient-to-patient variability in stress response due to patient acuity and workload. Linear contrasts with 95% CIs will be used to compare the outcomes between different interventions. Given an SD of 15 ms in HRV51 and an estimated STAI score difference of 5.8 (SD=8), we found that a sample size of 38 per group will provide 80% power at the two-sided 0.05 significance level to detect differences of 10.8 ms, an effect reflecting clinically meaningful changes to stress in prior HRV studies.20 This sample size is a conservative estimate as we expect improvements in power (or detectable effect size), given the repeated post-randomisation assessments. Nevertheless, we will enrol 42 participants per group to accommodate a potential 10% loss to follow-up.