Background

The severe acute respiratory syndrome coronavirus 2 (SARS-COV-2) pandemic has caused over 173 million cases worldwide, with over 3.7 million deaths, by June 2021 [1]. Healthcare provider (HCP) infections are thought to comprise a notable proportion of these cases [2]. The emergency department (ED) is a high-risk setting of exposure to SARS-COV-2, especially given the performance of aerosol-generating medical procedures (AGMPs) such as endotracheal intubation [3]. Therefore, EDs have sought to follow international guidance on personal protective equipment (PPE) practices [4], increase isolation rooms [5], utilize dedicated intubation teams [6] [7], increase the use of ED telemedicine [8], and implement updated resuscitation guidance [9].

Patients with SARS-COV-2 requiring intubation [10] represent a risk to HCPs [11] given the potential aerosolization of the virus [12]. The risk is increased, in part because of the duration of exposure and proximity to the patient increase [13]. Therefore, it is important to minimize the door opening of AGMP patients, which may transmit aerosolized material and to minimize the number of HCPs present inside isolation rooms. It is also necessary to minimize doffing and re-donning of PPE, which produces self-contamination risks and over-use of resources [14].

Given the importance of maintaining closed isolation room doors, it is important to continue clear communication between HCPs across closed doors. This presented multiple challenges for our ED during the first wave of the pandemic, which prompted multiple in situ simulation exercises [15] to prepare HCPs and increase safety practices with AGMPs. A variety of communication methods across the closed doors were trialed, including a “call bell” system (paging the nurse desk), a whiteboard and marker through glass windows, and finally a commercial baby monitor system, which was in use during our baseline period. Communication remained the most common challenge cited in debriefs after ED resuscitations.

Given the communication challenges across isolation rooms along with the risks of the continued door opening and potential doffing and re-donning, we conducted a quality improvement (QI) initiative with the aim to decrease isolation room door opening and improve HCP communication clarity.

Methods

Study setting

Toronto General Hospital (TGH) is a quaternary care adult academic 371-bed medical center part of the University Health Network in Toronto, Ontario, Canada. TGH is a referral center for SARS-COV-2 critically ill patients. The ED sees approximately 55,000 patients per year. The TGH ED has 24 beds in the acute zone, of which 17 are isolation rooms. Isolation rooms have glass doors with curtains outside them (for privacy) or anterooms leading to the patient room. Baseline communication with the outside team was through door opening or by using a commercial baby monitor system.

Participants

Our HCP participants consisted of ED nurses, physicians, and trainees on shift during the data collection period. Patient encounters were included if the patient was greater than 16 years old, placed in an isolation room, and met any of the potential infectious symptoms for SARS-COV-2 (Additional file 1: Appendix 1), regardless of presentation acuity.

Study design

This QI initiative was completed in a prospective, multi-stage approach consisting of a baseline period and three discrete Plan-Do-Study-Act (PDSA) cycles. The SQUIRE 2.0 Guidelines were used for study design and reporting [16]. We received a formal exemption from our organization’s research ethics board. We were supported by a local grant: The Mount Sinai Hospital - University Health Network (MSH-UHN) Academic Medicine Organization COVID-19 Innovation Grant.

Interventions

An initial period of stakeholder engagement was undertaken at daily nursing huddles and ED physician business meetings, which helped devise three sequential PDSA cycles. PDSA-1 was an educational intervention, whereby HCPs were educated about the need and rationale for closing isolation room doors through email and daily departmental huddles.

Given the limitations of educational interventions, concurrent search was done for a better information communication technology (ICT). PDSA-2 was the introduction of an ICT developed by a co-author (CC) called the TQC 200 “the transceiver” [17]. The transceiver was initially developed for use in sports but was adapted to our local ED setting to replace the existing use of the commercial baby monitor system. The transceivers are wireless radiofrequency wearable paired headsets that allow for two-way uninterrupted communication across closed isolation room doors (Fig. 1). HCPs entering an isolation room would wear one transceiver headset as part of the PPE donning process, and the remaining team members would wear paired transceiver headsets. Team members would then be able to talk to each other freely without pushing any buttons. Devices would be disinfected according to approved infection prevention and control protocols and plugged in to recharge in between use. PDSA-3 was mainly to embed and sustain the change that was noted. It included a local clinical champion (registered nurse) who was identified to demonstrate, remind, and support nurses and physicians in their use of the new technology available. During this phase, further refinement of the transceiver was also made based on ongoing clinician feedback.

Fig. 1
figure 1

Transceiver headset

Full size image

Data collection

A dedicated research coordinator (SG; not involved in transceiver development) directly observed patient encounters that met the inclusion criteria throughout the baseline period and PDSA cycles. The total number of minutes HCPs spent during a clinical encounter (an episode of going into the room until exiting it) and the total number of minutes of door opening during that encounter were recorded. Data collection consisted of a convenience sample that occurred during research coordinator availability, between 09:00 and 15:00 on weekdays, between May and July 2020.

Baseline surveys were distributed to nursing staff and included quantitative (Likert scale) and qualitative (two open-ended questions) portions. Following the conclusion of PDSA 3, a follow-up survey was collected (Additional file 1: Appendix 2). Likert scale surveys included rating communication clarity, communication errors, the need for opening isolation doors to communicate, using other means to communicate (e.g., hand gestures, whiteboard), or having to alter clinical workflow to communicate adequately. The follow-up survey included questions comparing the baby monitor system to the transceiver. Finally, open-ended questions were asked about the team’s communication approach. Surveys were collected in the ED during HCP shifts. All surveys were developed locally and piloted with a sample of HCP prior to administration. Transceiver-specific qualitative data was collected during PDSAs 2 and 3 to allow for rapid improvements during the corresponding PDSA cycle.

Outcome and balancing measures

The primary outcome measure was the duration of the isolation room door opening, which was the percentage of time the isolation room door was left open between the HCP(s) entering and exiting the room for that corresponding patient encounter. Our aim was to decrease isolation room door opening by 50% over a period of 3 months. Our secondary outcome measure was HCP-perceived communication clarity across closed isolation room doors. This was measured using a Likert scale survey. Our aim was to increase HCP-perceived communication clarity using an ICT by one point on a five-point Likert scale over a period of 3 months.

Our balancing measures were HCP-perceived communication errors across the closed isolation room door during patient encounters, perceived use of workarounds across the closed door (e.g., hand gestures), and HCP pausing their workflow to communicate with team members outside rooms (all using a Likert scale). We also measured the percentage of patients seen in the ED who were placed in isolation rooms, i.e., failed triage screen presenting with one or more infectious symptoms to monitor the infectious landscape.

Data analysis

We utilized Statistical Process Control ([SPC] or Shewhart) XbarR charts [18] to assess for special cause variation. SPC charts were completed with QI Macros© (Version 2018.04, KnowWare International Inc., Denver, CO, USA) for Microsoft© Excel© (Microsoft Corporation, Redmond, WA, USA, Version 14.5.9). Centerline calculation utilized formulae [19] and control limits utilized rules recommended by the Institute for Healthcare Improvement [20]. Four discrete periods of data were collected (baseline and three PDSA cycles). For the baseline and follow-up surveys, we used a two-tailed Mann-Whitney U test to assess for statistical significance between means, with a significance level set at p<0.05. The qualitative (open-ended questions) underwent thematic analysis [21].

Exclusion criteria for points used to create the SPC charts included: subgroups with less than three data points according to accepted rules [22], patients who passed the triage infectious assessment (no infectious symptoms reported), patient interactions lasting less than 5 min (chosen as a minimum to include meaningful clinical interaction), and patient interactions when HCP were discharging patients from the ED (e.g., disconnecting from monitors and patient leaving).

Results

Direct observation of 174 patient encounters was completed over 33 days. Ninety-five encounters were included in the final analysis, while 79 met exclusion criteria and were removed from the analysis (38 did not fail the infectious screen; 25 encounters were less than 5 min; 16 subgroup days had less than three data points). Door opening (primary outcome) increased from baseline (n=40) with a mean of 72.97 to 96.93% in PDSA-1 (n=28; p=0.04). Then, there was a statistically significant decrease as compared to the baseline in PDSA-2 (n=6; mean 1.58%; p<0.001) and in PDSA-3 (n=21; mean 1.47%; p<0.0001). The primary outcome across the study period is illustrated in the SPC chart shown in Fig. 2. The baseline period and PDSA-1 met the criteria for special cause variation, i.e., a possible external influence of the system that would need investigation. An improvement was noted by PDSA-2 and 3, with no further special cause variation.

Fig. 2
figure 2

Isolation door opening XbarR statistical process control chart. CL center line; UCL upper control limit; LCL lower control limit; PDSA Plan-Do-Study-Act; medians are adjacent to the control limits

Full size image

The themes that emerged from the baseline nursing staff qualitative survey (when HCPs were using baby monitors) were varied (Table 1). Positive findings included the facilitation of isolation door closure, increased HCP safety, and some positive effects on communication. Negative effects pertained to workflow and difficulty with the quality of communication, and they exceeded the positive findings in both frequency and intensity.

Table 1 Qualitative responses during baseline period
Full size table

The qualitative themes identified post-PDSA-3 nursing staff survey included increased HCP safety, facilitation of door closure, positive communication effects, and increased patient confidentiality (Table 2). Moreover, positive effects were noted on workflow including being hands-free, improved efficiency, and portability. Themes also included some challenges with the new workflow, device design, and sound quality with the initial roll-out.

Table 2 Qualitative responses after PDSA-3
Full size table

Our secondary outcome of HCP-perceived sound clarity had a statistically significant improvement on the Likert survey (5 is best) from baseline (n=36; mean 3.36) to post-PDSA-3 (n=49; mean 4.21; p<0.001). Our three balancing measures of HCP-perceived errors, using other means of communication and alterations in workflow did not show any worsening; in fact, they all showed statistically significant improvements (Table 3). The post-PDSA-3 survey also showed the majority of respondents agreeing or strongly agreeing with utilizing the transceiver in making care for their patients less stressful, improved satisfaction with communication, and improved quality as compared to the previously used baby monitor (Fig. 3).

Table 3 Baseline and post-PDSA-3 nurse Likert survey
Full size table
Fig. 3
figure 3

Post-PDSA-3 Likert survey questions. PDSA Plan-Do-Study-Act

Full size image

Throughout the PDSA cycles, specific feedback was also obtained to improve the transceiver system and better integrate it into the workflow. Feedback and resultant actions are illustrated in Table 4. Finally, changes in patients presenting to the ED who failed the infectious screen are shown in Additional file 1: Appendix 3 and appear to show a slow general decline in patients meeting the criteria as the study period progressed.

Table 4 Feedback and actions for transceiver system improvement
Full size table

Discussion

Our QI project with [1] HCP education, [2] the introduction of the transceiver, and [3] engagement through a local nurse champion substantially decreased door opening for infectious patients beyond the apriori 50% aim (primary outcome). Our secondary outcome also showed a significant improvement, but narrowly missing our a priori one Likert scale point. This improvement in communication was also supported by qualitative feedback, which demonstrated a positive experience with the transceiver as a communication device. While there have been many reports of frontline HCPs using baby monitors in the ED [23], operating room [24], and intensive care units [25], our qualitative surveys revealed substantial problems with their use pertaining to communication clarity and workflow, especially in high-acuity situations when they are even more important for safe patient care.

The special cause variation in the baseline period (Fig. 2) coincided with the addition of new curtains placed in front of isolation room glass doors, which was organized by ED leadership and was unrelated to the conduct of our project. As a result of this increased privacy, HCPs often kept the glass doors open, thereby posing a potential risk to others. A significant decrease in door opening was noted after the introduction of the transceiver by PDSA-2, and it was maintained through PDSA-3. A search for potential confounders leading to this notable decrease such as a new local policy, incentive, or critical incident did not reveal any contributors.

The educational intervention was not noted to have a difference on decreased door opening, but we posit it may have helped prime the participants towards further PDSA cycles along with the transceiver introduction. Moreover, we leveraged a local nurse champion in PDSA-3 to further support and coach HCPs in keeping with the technology acceptance model [26]. While this was anecdotally appreciated, a further change was not captured on the SPC chart given the already low values by PDSA-2.

We coupled the introduction of the transceiver with ongoing feedback. The introduction of (ICT) into a complex work ecology such as in the ED necessitates ongoing stakeholder engagement and feedback, also referred to as participatory design [27]. This feedback led to ongoing improvements in design and use as illustrated in Table 4. Moreover, key factors that enable the successful adoption and integration of ICTs in healthcare include portability, user satisfaction, and communicability [28]. The main focus of the QI approach was to engage relevant stakeholders at every stage of the project, thereby optimizing adoption and integration into the workflow. A large proportion of HCPs found that using the transceiver decreased their stress during patient care and increased their satisfaction in communication with team members.

Qualitative surveys also highlighted further improvement opportunities, such as optimizing communication strategies with multiple team members speaking in parallel. While allowing continuous communication for all users (instead of only one at a time like talkie-walkie push-to-talk systems) facilitated efficient communication, there was some confusion when multiple people spoke at the same time, as would be inherent in real-time in-person interactions. To optimize continuous communication and shared mental models for resuscitation patients [29], in situ simulation [30] may improve team performance.

Limitations

Our pragmatic QI approach used a convenience sample based on the research coordinator availability. We also did not collect patient demographics (to reduce risk to participants), which may limit generalizability. Moreover, (CC) who developed the transceiver was not involved in study design, data collection, and data analysis to mitigate conflict of interest. Data collection was done as soon as patients entered the isolation rooms prior to the performance of infectious illness screening done by nurses, which led to a proportion of encounters being excluded. Moreover, we could not identify an adequate validated survey for our initiative, so a new instrument was developed to capture important measures instead.

The SPC chart shows special cause variation in the baseline period. Methodologically, the baseline should extend until no further special cause variation is noted. However, a change in practice was noted (the addition of curtains, leading to greater door opening), which highlighted the need to proceed for potential HCP safety. We elected to continue with PDSA-1 after seeing the following week of baseline data stabilize.

Conclusion

Our iterative QI approach with HCP education, transceiver introduction, and leveraging a local champion demonstrated significant decreases in the door opening and improvement in communication clarity without increasing communication errors, workarounds, or alterations in the workflow. Future work may include scaling this initiative to other EDs, operating rooms, and intensive care units.