Managing ICU surge during the COVID-19 crisis: rapid guidelines

Abstract

Given the rapidly changing nature of COVID-19, clinicians and policy makers require urgent review and summary of the literature, and synthesis of evidence-based guidelines to inform practice. The WHO advocates for rapid reviews in these circumstances. The purpose of this rapid guideline is to provide recommendations on the organizational management of intensive care units caring for patients with COVID-19 including: planning a crisis surge response; crisis surge response strategies; triage, supporting families, and staff.

Introduction

In December 2019, a widespread outbreak of acute respiratory illness occurred in Wuhan, China [1]. A novel coronavirus, later named ‘Severe Acute Respiratory Syndrome Coronavirus 2’ (SARS-CoV-2), was identified as the cause of this epidemic [2]. The World Health Organization (WHO) termed the illness caused by SARS-CoV-2 as ‘Coronavirus Disease 2019’ (COVID-19).

Since then, this virulent organism has spread to over 200 countries worldwide and territories and officially declared as a pandemic by the WHO in March 2020.

Scope

Given the rapidly changing nature of COVID-19, clinicians and policy makers require urgent review and summary of the literature, and synthesis of evidence-based guidelines to inform practice. The WHO advocates for rapid reviews in these circumstances [3].

The purpose of this rapid guideline is to provide recommendations on the organizational management of intensive care units (ICUs) caring for patients with COVID-19. This is not intended to provide clinical guidance as we recognize that others have produced recommendations on the clinical management of COVID-19 [4, 5]. Further, the intent is not to duplicate high quality existing advice regarding Mass Critical Care or Crisis Surge Response [6,7,8,9]. This rapid guideline focuses specifically on key questions about how to manage ICU surge during COVID-19, which have not been addressed elsewhere.

Methods

Panel selection

Panel selection focused on expertise, availability, diversity and ability to contribute within very short timelines during the pandemic. The core group and members of the steering committee for this project were all members of the panel and leadership of the Surviving Sepsis Campaign COVID-19 guideline [5]. A steering committee was constituted for the panel (YMA, MDC, WA, GC, LE) who nominated potential additional panel members with prior expertise in emergency preparedness, critical care, infectious diseases, and guideline development. The co-chairs (YMA, MDC) vetted nominees and invitations to join the panel were extended if there was consensus among the steering committee. A total of 25 panelists were selected from the 29 nominated individuals. The aim was to balance appropriately broad representation but maintain a manageable number of participants given the time constraints of this project and need to collect feedback in compressed time periods. Attention was paid to achieve as best possible to achieve diversity in geographic, professional background, and gender. In addition, two junior members of the profession with a specific interest in the field were invited to participate to support mentorship and development as well for age diversity. We used the GRADEpro Guideline Development Tool (GDT) online software (http://gdt.guidelinedevelopment.org) to administer WHO COI disclosure forms to participating panel members. Direct financial and industry-related COIs were not permitted and were considered disqualifying.

Question development

Coincident with the creation of the panel relevant topics was proposed by the panel members. We finalized a list of questions based upon the topics identified by discussion and consensus between panel members. Questions were formatted by panel to align with the population, intervention, comparator, outcome (PICO) format where possible. The questions were reviewed and a priority rating of 1 (highest)–5 (lowest) priority taking into consideration three factors: (1) clinical relevance of the question to the current COVID outbreak; (2) feasibility of developing consensus based upon the current body of evidence; (3) an identified gap in guidance on this topic from other reputable sources, such as the past ESICM guidelines [9]. The initial scoring was undertaken by the co-chairs then reviewed and agreed upon by the panel as a whole. Questions with a prioritization score of 3 or higher were included in this rapid guideline. A full list of the questions and scores is provided in the online supplemental material.

Literature search

Due to the time-sensitive requirement for evidence-based guidance and nature of the pandemic, we conducted a pragmatic, rapid review of the literature [10]. To facilitate rapid review of the literature, for each question, we electronically searched PubMed database [table of search terms used is available in online supplemental material]. To capture recent published and ‘preprint’ literature on COVID-19, these searches were supplemented with focused searches of Google Scholar and Dimensions (2020 Digital Science & Research Solutions, Inc). Where sufficient evidence could not be found using at least two databases; or in the case where a narrow, focused search could not be conducted; we further searched EMBASE database.

Selection of studies and evidence summary generation

A single reviewer screened study titles and abstracts retrieved from the searches, and only included if applicable to each focused question. Subsequently, the methodology team reviewed the selected list of studies and supplemented the references in the evidence tables if other relevant studies were identified. Content experts were asked to indicate and add any studies that were not captured by the search. Methodologists from the Guidelines in Intensive Care Development and Evaluation (GUIDE) group (www.guidecanada.org) created evidence summaries which synthesized the available evidence for each question. If no comparative evidence was available, the methodologists summarized the evidence narratively.

Quality of evidence

We used the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach to assess the quality of evidence (also known as confidence or certainty in the evidence) [11], i.e., our confidence in the estimate of the effect to support a recommendation [12]. The quality of evidence was rated as high, moderate, low, or very low [13]. Where sufficient evidence existed, methodologists used the guideline development tool (GDT) online software (http://gdt.guidelinedevelopment.org) to generate the evidence profiles (evidence summaries) [14].

Recommendation formulation

Evidence summary tables (including GRADE assessment of the quality of evidence) were sent to subgroups of authors who reviewed the literature and drafted preliminary recommendations. We use the wording ‘we recommend’ for strong recommendations and ‘we suggest’ for weak recommendations. Best practice statements are equivalent to a ‘strong recommendation’ as either unequivocal benefit or harm is felt to exist, and as such, we are unlikely to ever have high-quality evidence [15].

The final list of recommendations was developed by panel discussion and consensus; voting on recommendations was not required. We summarized the recommendations in Table 1.

Table 1 Recommendations and statements

Recommendations

I. Planning a crisis surge response

Ia. What is the burden of the COVID-19 on critical care?

Background:

As the COVID-19 pandemic spreads, the ICU is physically, materially, and emotionally challenged with the associated immense caseload. Knowledge of the current epidemiology, clinical course and resource utilization provides valuable information to aid the strategic and daily planning of ICUs. Therefore, an understanding of number of patients and capacity, resource utilization is essential to adequately address the ‘staff,’ ‘stuff,’ ‘space’ and ‘systems’ to mount a surge response [6, 8, 16, 17].

Recommendation:
1.	For institutions preparing ICUs during the COVID-19 pandemic:
1.1.	We suggest planning and resource allocation considering that 1 in 5 hospitalized adult COVID-19 positive patients will require ICU admission. (weak recommendation, low quality evidence)
1.2.	We suggest planning for the number of critical care resources (staff, supplies, space) required should assume 70% of ICU patients will require any type of ventilatory support, including NIV and HFNO with > 50% of ICU patients requiring invasive ventilatory support, in addition to supporting other COVID-associated organ failures including renal and cardiovascular (weak recommendation, low quality evidence)

Rationale:

Our searches identified 26 original studies [1, 2, 18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39] and 2 systematic reviews [40, 41] that described outcomes of adult COVID-19 patients in the ICU. The reporting of outcomes, clinical course as well as the definitions varied among studies, and most lacked a full description of the clinical picture and resource use of ICU patients.

The original studies included retrospective cohorts and case series with a total of 83,619 patients across all spectrums of severity. We did not perform any meta-analyses; therefore, we reported means and ranges across eligible studies. Among these 5841 were admitted to the ICU, the mean rate of ICU admission among hospitalized patients with COVID-19 pneumonia was 20.1% (range 4.6–32%; low- to very-low-quality evidence). There were only 6 studies in the upper quartile of sample size with sample sizes ranging between 138 and 2087 patients.

Overall, the median age of ICU patients was 59.7 years and 62% were male. Mean ICU and hospital length of stay were, respectively, 7.3 and 12 days. ARDS was present in 38% of the patients. In studies that reported rates of ventilator support, a mean of 35% required NIV, 73% used HFNO (in only 4 studies likely biased by limited resources of full functional mechanical ventilators), 48.8% required invasive mechanical ventilation with a mean duration of 7.8 days, and 8% used extracorporeal membrane oxygenation (ECMO). The mean proportion of renal replacement therapy and vasopressors use across studies was 13.2% and 40.8%, respectively. Based upon UK data, up to 20% of critically ill COVID patients required renal replacement therapy [42].

The mean ICU mortality rate was 34.9% (range 0–72%), and hospital mortality rate was 45% (range 5–72%). Mortality varied significantly across reports and is likely influenced by a combination of system level effects resulting from crisis surge situations and variations in quality of care as well as practice patterns and population demographics.

The current literature is limited by the lack of information on long-term outcomes on ICU patients. Other limitations are related to incomplete data from the present studies either due to a lack of information on clinical characteristics and outcomes or due to the fact that in several reports, ICU discharge or 28-day outcomes were only reported for a fraction of patients as a significant number were still hospitalized and in ICU at the time of the publication of the various reports. Lastly, few countries have published their data to date.

Ib. What is the projected number of ventilator and beds required for managing peak surge during COVID-19 in a population?

Background:

The COVID-19 epidemic revealed the vulnerability of healthcare systems and how they can rapidly be overloaded in excess of the available ICU bed and ventilator capacity. Predictive models have been proposed to support healthcare authorities in early planning of resources, personnel, ICU, and hospital bed capacity. An early estimation of the proportion of the existing hospital or ICU capacity that needs to be liberated is necessary for the planning of a partial reduction or complete cancelation of nonemergency services and surgery, and nonurgent admissions [31]. Predictions can indicate that the existing capacity is insufficient and reveal the eventual need to create of additional capacity [43].

Recommendation:
2.	We recommend healthcare systems and hospitals use mathematical modeling to support their surge capacity planning and applying the following principles: (Best practice statement)
2.1	Establish predictions as early as possible in the course of the epidemic
2.2	Models should be pragmatic and focus on the only relevant question for surge capacity: how many patients will need hospital and ICU resources on a given day?
2.3	Predictions should model a best, worse, and most likely scenario and use different statistical approaches and compare the results
2.4	Predictive models should take into account the R0 of the virus, if known; the rate of spreading in other countries and settings; the expected or observed rate of hospitalization, need for ICU, need for mechanical ventilation, need for ECMO; case fatality rate; expected duration of mechanical ventilation, ICU length of stay (LOS), hospital LOS
2.5	Models should incorporate the impact of the installation of distancing measures in society and their delay until impact on case detection, actual or theoretical
2.6	Once peak surge has been reached, models should be used to plan the surge exit strategy and to continuously monitor new data to detect a second peak as early as possible

Rationale:

For a new and unknown disease, predictions can be challenging, because known parameters of earlier epidemics are often not applicable. In addition, the different testing and reporting approaches of different countries might have consequences for the external validity of using these data as parameters for models in a different healthcare setting. Models that focus on the true proportion of infected patients in a population, or the rate of hospitalization based on data from other countries, might be using the wrong assumptions. However, at a later stage, as the disease progresses and more data become available about its behavior, these additional parameters could be incorporated.

Simple projections of exponential growth, without predictions of the peak, are of less value and cannot be used for surge capacity planning, as the curve will continue to grow. Models that only look a couple of days ahead are of limited value. The assumptions used by the models should be reported transparently. Rough estimates based on ‘gut feelings’ should not be used as they could give policy makers a number of false options between which they might choose.

An example of the model used for the Belgian surge capacity planning can be found in Fig. 1.

Fig. 1

Projected versus observed number of hospitalized patients per day, for Belgium. Model created by Lize Raes and Kareljan Raes, and available on https://drive.google.com/file/d/1_tT–cLqvRyRHBjYmkkZu6MC0leXev8p/view

Ic. What are the projected supplies and equipment required to manage an intubated ICU patient during the COVID-19 (or pandemic) surge?

Background:

In the setting of a pandemic surge, many forms of supplies are essential to provide lifesaving care to critically ill patients. While lack of ventilators and staff is key considerations, a lack of other equipment and supplies—including, but not limited to personal protective equipment, monitors, intravenous supplies, medications—is also likely to result in substantial patient morbidity and mortality, and limit the number of patients who can receive effective critical care.

Recommendation:
3.	We recommend that hospitals develop an inventory of supplies and equipment necessary to provide care to critically ill patients during a pandemic, and identify potential shortages based upon projected ICU needs. (Best practice statement)
Remarks: Using this information, hospitals can seek to replenish and stockpile necessary supplies and equipment early, before supply chains are disrupted, and work to find alternatives. Collaboration with other local organizations (other hospitals, government, corporations, nongovernment organizations) can be used to ensure optimal allocation of supplies to hospitals

Rationale:

Providing lifesaving care to critically ill patients with COVID-19 is resource-intensive, anticipating an ICU length of stay of over 7 days [1, 2, 18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41]. A comprehensive list of basic supplies and equipment required has been developed in the context of influenza pandemics, and these likely apply to COVID-19 population as well [9, 16, 44,45,46,47]. Of note, the duration of mechanical ventilation and length of stay of patients with COVID may be longer than that of influenza, and thus these earlier pandemic supply estimates are likely an underestimate. In the context of a pandemic, many supply chains are likely to be disrupted and having a clear inventory and advance understanding of which supplies are likely to run out first can allow for early replenishment of these supplies, or identification of alternatives, if replenishment is unavailable (see Table 2).

II. Crisis Surge Response Strategies

IIa. What are the available strategies for institutions to overcome shortage of mechanical ventilators?

Background:

Table 2 The projected supplies required to manage an intubated intensive care unit patient during the COVID-19 (or pandemic) surge

The large surge of COVID-19 patients with respiratory failure has led to shortages of mechanical ventilators in countries such as Italy and the USA [31, 48]. Without access to mechanical ventilation, many of these patients will not survive. In order to produce the best patient outcomes, there must be adequate supply, distribution, and timely access for patients to mechanical ventilation. Therefore, strategies are required to improvise and urgently overcome these shortfalls.

Recommendation:
4.	To mitigate a shortage of mechanical ventilators:
4.1	We suggest that hospitals develop and implement protocols for intubation as well as the use of high-flow nasal oxygen (HFNO) and noninvasive ventilation (NIV) in order to reduce the need for intubation. (Weak recommendation, low-quality evidence)
4.2	We recommend that hospitals increase the quantity of standard full-featured ventilators according to the projected number of patients who require mechanical ventilation. (Strong recommendation, moderate quality evidence)
4.3	We recommend that standard full-featured ventilators (as opposed to flow generators or basic volume control resuscitation devices) are used for COVID patients requiring invasive mechanical ventilation, in particular when requiring fully controlled ventilation. (Best practice statement)
4.4	In setting with shortage of standard full-featured ventilators, we suggest using alternative devices that provide invasive mechanical ventilation, including long-term ventilators, emergency transport ventilators, anesthesia gas machines, magnetic resonance imaging (MRI) compatible ventilators. (Weak recommendation, low-quality evidence)
4.5	In setting with shortage of standard full-featured ventilators, we suggest using repurposed devices and alternative techniques as a last option, such as prolonged manual ventilation, NIV for invasive ventilation, veterinary ventilators. (Weak recommendation, low-quality evidence)
4.6	When planning for increased mechanical ventilation capacity, we recommend considering the requirements of oxygen/medical gas supply, electrical supply, airway management and ventilation consumables, physical space, and staff necessary to effectively and safely deliver mechanical ventilation. (Best practice statement).

Rationale:

To mitigate the worldwide shortage of ventilators, manufacturers have increased the production of ventilators. The unmet needs have also prompted the development of open source and easy to produce ventilators [48,49,50]. But many hospitals are unable to provide enough ICU ventilators to manage the surge of COVID-19 patients with acute hypoxemic respiratory failure.

The first goal is to avoid intubation if medically appropriate and where possible within constraints. Patients with acute hypoxemic respiratory failure due to COVID-19 are commonly managed by HFNO or NIV with premise of avoiding intubation. A systematic review and meta-analysis of 9 RCTs in unselected critically ill patients (2093 patients) demonstrated that HFNO reduces intubation compared to conventional oxygen (RR 0.85; 95% CI 0.74–0.99) but did not affect mortality or ICU length of stay [51,52,53]. Other meta-analysis comparing HFNO to NIPPV in unselected critically ill patients has shown HFNO to decrease the need for intubation of patients compared to NIPPV without significantly decreasing mortality or ICU length of stay [51]. Therefore, the Surviving Sepsis Guidelines for COVID-19 suggested the use of HFNO over conventional oxygen and over NIV [5].

Some data suggest that NIV using face masks may improve oxygenation and delay the need for intubation in patients with COVID-19, but its effect on the rate of intubation and mortality is unclear. In addition, NIV and HFNO are potentially associated with an increased risk of viral spread and nosocomial transmission of the infection due to aerosol generation [54,55,56,57,58]. NIV delivered through devices that use double-tube circuits (which includes selected NIV machines and ICU ventilators) is preferred over devices that use single-tube circuits (only inspiratory line), because of the inability to mitigate aerosol generation associated with the latter devices. Observational studies in patients with severe influenza A (H1N1) and MERS reported high NIV failure, reaching 92% in a study of Middle East respiratory syndrome (MERS) patients, with mixed data regarding mortality [59,60,61]. Moreover, it is demonstrated that failure of noninvasive respiratory support and delayed intubation are associated with worse outcome in hypoxemic patients. Helmet continuous positive airway pressure (CPAP) or (less frequently) NIV has been commonly used in Italy to manage patients with COVID-19 [62]. It has been used in the ICUs and on hospital wards and may be associated with lower risk of nosocomial transmission than some face masks when an exhalation port is open along with a single-tube ventilatory circuit or HFNO [62, 63]. However, the risk of transmission can potentially be mitigated with face masks that utilize double-tube circuits and does not have exhalation ports or with HFNO when attention is paid to reduce leakage. Data from a single-center randomized controlled trial (RCT) in patients with unselected patients with ARDS showed that treatment with helmet NIV reduced intubation rates and 90-day mortality [64].

The limited availability relative to the demand during pandemics of full-featured ventilators has prompted the search for alternative options [65]. A survey of US hospitals published in 2010 estimated that there were 62,188 ‘full-feature’ ventilators across the USA [66]. Additionally, there were 98,738 devices other than full-feature ventilators. These devices included portable mechanical gas ventilators, ‘standby’ ventilators (no longer used for everyday patient care but maintained and available on-site), portable mechanical pneumatic, noninvasive ventilators (can be repurposed and modified for invasive positive pressure ventilation (IPPV)), neonatal pediatric and CPAP, automatic resuscitator, and basic EMS transport ventilator [66]. Other options include the use of long-term ventilators, transport ventilators, veterinary ventilators, repurpose old ventilators from warehouse, CT/MRI, and use anesthesia gas machine to ventilate patients [65].

Manual ventilation after intubation can be used for brief period only, since operator fatigue, patient hypoventilation, risk of transmission of virus, staff availability are all issues which limit this strategy.

IIb. Is ventilating multiple patients on a single ventilator a feasible strategy to address shortages of mechanical ventilation?

Background:

The concept of using modifications of ventilator circuits that permit the use of a single ventilator to support multiple patients has been suggested to address the shortage of ventilators during surge of patients with COVID-19.

Recommendation:
5.	We recommend against using one ventilator to ventilate multiple patients. (Strong recommendation, low-quality evidence)

Rationale:

A study demonstrated that four simulators of adult lung of similar mechanics can be ventilated for 12 h using one ventilator [67]. Similarly, a study demonstrated the feasibility of ventilating four sheep with similar lung compliance using one ventilator [68]. However, lung compliance and resistance are likely to vary among patients with acute respiratory failure and even in the same patient over time, which would lead to large variations in tidal volumes [69]. Models of test lungs connected to a ventilator with different combinations of compliance, airway resistances, modes of ventilation, inspiratory and end-expiratory pressure levels documented large discrepancies in delivered tidal volumes with changing lung mechanics especially with compliance differences [70]. A team from Columbia University College of Physicians and Surgeons proposed a ventilator sharing protocol that requires selection of patients with similar mechanical support needs, the use of neuromuscular blockade and transferring patient to a single ventilator for weaning [71].

The Society of Critical Care Medicine (SCCM), American Association for Respiratory Care (AARC), American Society of Anesthesiologists (ASA), Anesthesia Patient Safety Foundation (APSF), American Association of Critical Care Nurses (AACN), and American College of Chest Physicians (CHEST) issued a consensus statement suggesting that sharing mechanical ventilators should not be attempted because it cannot be done safely with current equipment [72].

The reasons for avoiding ventilating multiple patients with a single ventilator include the delivery of unpredictable volumes to the two patients according to different lung compliance, inability to individually managing positive end-expiratory pressure (PEEP) and the inability to accurately monitor ventilation, measure pulmonary mechanics, and manage alarms [72]. In addition, the wide and prolonged use of neuromuscular blockers and cross-infections may lead to prolongation of ventilator dependency, which may defeat the purpose of ventilator sharing, and therefore ventilator triage may be an overall a better option [72].

IIc. What are the available strategies for institutions to overcome shortage of intensive care staff (physicians, nurses and other staff)?

Background:

A pandemic can quickly overwhelm healthcare systems as surges of critically ill patients are admitted, forcing hospitals to adopt crisis standards of care. During contingency and crisis capacity, hospitals may be required to more than double their ICU capability. This can lead to severe shortages of critical care trained staff and requires careful advanced planning. The strategies can be categorized into methods that increase the supply, minimize the loss, and maximize the utilization of staff.

Recommendation:
6.	Where there is shortage of intensive care staff, we suggest the following actions: (Weak recommendation, low-quality evidence)
6.1	Suspending all elective medical and surgical procedures and activities once ongoing chains or community transmission of COVID-19 has been documented within a State/Province/Country, in order to conserve critical care capacity
6.2	Expediting the credentialing process to quickly approve both domestic and foreign healthcare workers to assist in areas of need
6.3	Reclaiming critical care trained staff who are in other departments and hiring retired critical care trained staff
6.4	Temporarily redeploying healthcare workers and trainees to the ICU to work in a care-team model even if the ICU is normally outside the scope of their practice
6.5	Providing just-in-time training and simulation sessions for non-ICU clinicians reassigned to work in ICU, to better prepare them for their roles
6.6	Creating and maintaining a safe working environment with the necessary supplies, personal protective equipment and education to protect staff and trainees
6.7	Employing telemedicine and other technology to increase the number of overseeing critical care providers
6.8	Restructuring ICU teams to employ a tiered staffing model (‘care team’) that augments the ability of the available experienced critical care staff to care for as many patients as possible

Rationale:

Suspending elective activities frees staff and resources that would otherwise be engaged in those activities [16, 73, 74]. This also preserves the PPE supply. State and federal credentialing boards should work with governmental agencies to expedite their processes to approve essential workers from other areas to flow rapidly to areas of need [75]. For example, the Uniform Emergency Volunteer Health Practitioner Act recognizes out of state licenses for different health practitioners during an emergency. Individuals who have critical care skills and training in other departments should be recruited [74]. With appropriate supervision and organized training, staff or trainees of all types may be redeployed to ICU roles even if outside their normal area of expertise [73, 74, 76]. Staff will need to take on responsibilities not typical of their role. Due diligence must be exercised when considering recruiting staff, such as retirees, to return to work in the ICU as they may be more likely to be from ‘at risk’ or ‘vulnerable’ group.

It is of utmost importance to protect the health and safety of healthcare workers, and trainees, both to preserve the workforce and to maintain its morale. The hospital has a responsibility to provide PPE and associated training [16]. There are a number of other space adjustments a hospital can make to prevent frequent entry into a patient’s room. Intravenous pumps and other titratable medicines can be kept outside of the room. Laboratories and other procedures such as medication administration can be batched together to prevent frequent entry. Importantly, mental health and burnout must be addressed with counseling and other wellness interventions [16].

Telemedicine is a beneficial tool that allows skilled critical care physicians and nurses to work in areas from which they are geographically remote and allows high-risk healthcare workers to safely work remotely [76]. Standard team structures and workflow must be reorganized to provide quality critical care to the most patients [74, 76, 77]. This requires placing an ICU attending and experienced ICU nurses in oversight positions with non-ICU trained staff at the bedside [74, 76]. Multiple models have been suggested (see Fig. 2 for an example), but the structure meets the needs of the individual institution and its resources as well as different models in various Countries [74, 76].

Fig. 2

Care team model for extending the capacity of ICU clinicians

IId. What strategies can be used to reduce healthcare worker exposure to COVID-19?

Background:

Healthcare workers are at increased risk of exposure to SARS-CoV-2, and there have been increasing reports of healthcare worker infection and deaths resulting from contact with infected patients [78, 79]. Guidelines regarding PPE have been published previously [5]. However, it is essential to look at other ways of reducing healthcare worker exposure to the virus, in order to protect them from the potential harm to themselves from contracting COVID-19, and to ensure they are able to safely provide critical care to patients.

Recommendation:
7.	During the COVID-19 pandemic to reduce healthcare worker exposure to SARS-CoV-2:
7.1	We recommend that staff undergo training in proper donning and doffing of PPE. (Best practice statement)
7.2	We suggest using visual aids, checklists and trained observers to assist in safely doffing PPE. (Weak recommendation, low-quality evidence)
7.3	We recommend minimizing the number of staff entering the rooms of patients with COVID-19, remote access to equipment controls and bundle care to minimize the number of exposures. (Best practice statement)
7.4	We suggest minimizing transport of COVID-19 patients off patient care units (i.e., to diagnostic radiology). (Weak recommendation, low-quality evidence)
7.5	We recommend that healthcare institutions and ICUs develop and implement response plans to clinical emergencies such as endotracheal intubation, cardiac arrest for patients with COVID-19. (Best practice statement)

Rationale:

Specific PPE components and models often differ across healthcare institutions. Training of healthcare staff in donning and doffing of personal protective equipment is intended to increase the correct use of PPE and reduce healthcare worker exposure. Small studies of different training modalities have demonstrated reduced frequency of healthcare workers contamination during doffing in experimental models. Training may consist of in-person instruction, video instruction, return demonstration, use of simulated contamination, i.e., ultraviolet fluorescing powder or gel. Insufficient data exist to recommend one training modality over another [80, 81]. Few data exist on the optimal frequency of training; however, one small study demonstrated improved practice of doffing of gloves at 3-month follow-up from training [82].

Reducing the number of staff, and the frequency and duration of times that staff enter the rooms of patients with COVID-19, while ensuring safe patient care, may reduce healthcare worker exposure. Potential strategies include the use of telemedicine to monitor patients. Other remote communication/monitoring devices (i.e., baby monitors have been employed in some centers). Care activities may be ‘bundled’ to reduce the number of times a HCW enters a patient room: laboratory draws, medication administration, patient assessment, nutrition, personal care [83]. Consideration of appropriate therapeutic alternatives with longer dosing intervals may be utilized [84]. Some centers are describing novel processes for critical care patients in individual rooms: placing IV poles and medication pumps outside of the patient room using extension tubing to allow for changing infusion rates of titrated medications or remote controls for equipment without the need to enter the patient room [85]. Risks include potential inaccurate administration of medications, decreased ability to detect occlusions, potential interference with negative airflow in airborne infection isolation rooms.

Minimizing transport of patients with COVID-19 within a healthcare facility may reduce the risk of exposure of HCW and other patients and staff. In keeping with general good medical practice, laboratories, imaging studies, and other procedures that are unlikely to change patient management should be minimized [86]. When clinically appropriate, clinicians can substitute bedside diagnostic procedures, for example through the use of point-of-care ultrasound and portable X-rays devices [87, 88].

Developing and implementing standard procedures for clinical emergencies in patients with COVID-19 may reduce the risk of healthcare worker exposure. Clinical emergencies may present increased risk of HCW exposure due to time pressure on staff to respond to a deteriorating patient with risk of errors or omission of proper donning of PPE. Defining triggers for consideration of escalation (such as triggers for endotracheal intubation) with recognition of the additional time necessary for proper donning of PPE, preparing necessary equipment and staff can allow for patient management under more controlled circumstances [89].

IIe. What are the available strategies for reprocessing FFP3/N95 or surgical masks?

Background:

Recommendations from international organizations outline the use of surgical and fitted high-filtration facepiece respirators as essential PPE during patient care of COVID-19 patients, depending on the activity being undertaken [90, 91]. However, due to the exceptional increase in demand for N95 filtering facepiece respirators as a result of the pandemic, this had led to shortages in some countries [92, 93]. Shortages in filtering facepiece respirators risk both staff and patient health due to exposure to SARS-CoV-2. HCW in some areas are required to use the same N95 respirator for 1 week of shifts with all patients and storing this biohazardous material in a paper bag. A potential method of mitigating these shortages is to reprocess filtering facepiece respirators for multiple uses; however, uncertainty remains about the various reprocessing strategies.

Recommendation:
8.	In the event of a supply shortage necessitating the reuse of PPE:
8.1	We suggest reprocessing of respirators (N95/FFP3 masks) with UVGI or VHP over ethylene oxide. (Weak recommendation, very low certainty of evidence)
8.2	We suggest not using time as a decontamination method given that virus remains in the mask for > than 7 days. (Weak recommendation, very low certainty of evidence)
8.3	We suggest not extending the use of masks across multiple patients for multiple days. (Weak recommendation, very low certainty of evidence)

Rationale:

During a contingency or crisis surge response, it may be necessary to consider the reuse of what are usually single-use medical devices under normal circumstances [8, 17]. As a result of the combination of global demand for PPE combined with the impact on the production and supply chain for PPE, many countries are facing shortages. While it is easier to adapt or substitute specific items of PPE such as eye protection and gowns, with the exception of introducing devices such as powered air purifying respirators (PAPR), there are few other adaptations or substitutions for respirators (N95/FFP3 masks) and thus more likely to necessitate reuse as an option to address shortages. A joint statement by the American Society of Anaesthesiologists (ASA), Anaesthesia Patient Safety Foundation (APSF), American Academy of Anaesthesiologist Assistants (AAAA) and American Association of Nurse Anaesthetists (AANA) Anaesthesia recommends that those who will be in the vicinity of aerosol generating procedure should use properly fitted N95 masks or PAPR [94]. The CDC recommends for those who are not N95 fit-tested, have facial hair, or fail N95 fit-testing PAPRs should be used if possible. The CDC and a recent review recommend the use of source control (i.e., masking of symptomatic patients) [95, 96].

We recommend against extended wear because it increases the risk of self-inoculation and cross-contamination [97]. In addition, fit and filtration are reported to degrade with extended use [98]. Frequent donning and doffing of the same contaminated mask increases HCW contamination risk [99] and compromise fit [100].

The reuse of single-use medical device can only be considered if reprocessing the device results in a product considered ‘safe’ and the benefits overall outweigh the risks following a formal risk assessment which considers the alternative available options. The reprocessing of respirators requires ensuring the devices are effectively decontaminated and maintain their filtration efficacy and also their structural integrity to preserve mask fit. Reprocessed masks must be returned to the original wearer to avoid cross-contamination, infection by other pathogens and to reduce sensitivities to other contaminants contained in the mask such as oils, preservatives from cosmetics or residual skin care products, such as sun protection and acne care. We could find no decontaminating process that reported testing for or eliminating these concerns by testing for these factors with human participants. A number of studies have assessed the ability to reprocess respirators [97, 101,102,103,104,105,106,107,108,109,110,111,112,113]. We recommend against time as a decontamination method given that virus remains in the mask for > than 7 days [114].

While respirator reuse appears to be a feasible option in some circumstances, it is important to note that both the technique selected and the specific masks (manufacture and model) being reprocessed impact the feasibility of this strategy [115]. Expert advice should be sought prior to undertaking reprocessing, and if at all possible, a quality assurance process implemented.

III. Triage

IIIa. Is a legal framework required to permit triage in a civilian setting?

Background:

In the setting of a crisis surge response, resource allocation can be ethically justified; however, without a legal framework in which to operate safely, clinicians and hospitals participating in triage activity may be vulnerable to legal action when withholding or withdrawing care. The lack of legal protection may prevent the clinicians’ ability to perform effective triage.

Recommendation:
9.	We recommend that each State/Province/Country develop a triage protocol, and system to support it, that is based on local practices and legislation and which is adopted by individual hospitals. (Best practice statement)
10.	When State/Province/Countries develop a triage protocol, we recommend:
10.1	That hospital leadership works closely with the government to ensure legal protections prior to instituting a triage system. (Best practice statement)
10.2	Apprising clinicians of their protections when acting in good faith and in accordance with established triage protocols to ensure consistent application of triage decision-making. (Best practice statement)
10.3	Meticulous documentation of all triage decisions. (Best practice statement)

Rationale:

The need for medical triage is triggered by public health emergencies during which health systems are overwhelmed and do not have enough resources to treat all patients. A medical triage system that allocates scarce resources represents a shift from an individual patient approach to a ‘greater good’ approach. However, current European [9, 116, 117] and USA [74, 75, 118,119,120,121,122,123,124,125,126,127] law supports the good of an individual patient. This puts clinicians participating in triage who withhold or withdraw care at risk of civil or criminal charges.

A formal declaration of an emergency, disaster or public health emergency by government must precede activation of a medical triage system. As part of a legal framework, the following issues should be addressed: governing bodies must work together to ensure rapid credentialing; healthcare workers practicing outside their normal domains as well as those acting in good faith during the crisis response must be protected; acknowledgment of adapted treatment standards during the crisis [9, 74, 75, 116,117,118,119,120,121,122,123,124,125,126,127]; fair access to treatment, protection of vulnerable populations and assurance of patients’ interests and allocation of scarce resources [126]. Although there is lack of high-quality evidence to support any specific triage protocol, advance planning of a triage protocols, and systems to deliver triage prior to an emergency that is aligned with medical societies and has input from legal and ethics experts as well as community members can help mitigate the legal risks [9, 74, 75, 116,117,118,119,120,121,122,123,124,125,126,127].

IIIb. What is an appropriate minimum time-limited trial of ventilation for patients admitted to ICU during the COVID-19 crisis?

Background:

Several triage protocols for proposed use in pandemics have included the prospect of trial of therapy prior to re-assessment to assess for evidence of patient improvement. The duration of time for a time-limited trial of ventilation should take into account the natural history of the underlying illness causing the predominant number of cases cause the surge in demand. Early reports of COVID-19 patients suggest recovery is possible after prolonged periods of intubation, so the time given to a time-limited trial of ventilation must be carefully considered.

Recommendation:
11.	For an adult COVID-19 patient, we suggest that if a time-limited ventilation trial is incorporated in a triage protocol the minimum duration of the trial should be 10-12 days. (Weak recommendation, low-quality evidence)
Remarks: Parameters must be clearly delineated and balance a patient-centered approach with system needs. A time-limited trial may be ended before 10 days if a patient’s condition is worsening significantly or extended past 12 days if a patient is showing signs of improvement and resources permit. As more outcome data is reported, this recommendation may need to be updated.

Rationale:

Previous medical triage algorithms typically recommend re-assessment periods of between 48 and 120 h at which time it is decided whether to continue critical care or to divert those scarce resources to someone else who is determined to benefit more [124, 128, 129]. The ideal duration of a re-assessment period should be related to the natural history of the underlying illness and patient values such as how long a trial or what other subsequent interventions a patient might tolerate.

We do not have robust, long-term data on patient outcomes with COVID-19. China and Europe report overall ICU mortality rates of up to 38% and median time from ICU admission to death of 7 days. One international review reports a median number of ventilator days of 9.1 days (SD 5.5 days) for all intubated patients and a UK ICU cohort [42] of 1053 patients median LOS for ICU patients requiring mechanical was 8 days (IQR 5–12) for survivors and 6 days (IQR 4–9) for non-survivors [18, 21, 24, 32, 33, 36, 130]. For this reason, a time-limited trial of 10–12 days is recommended. The trial may be ended sooner if there are clear signs that a patient is worsening and unlikely to survive. The trial may be extended if the patient is showing signs of improvement and resources are available to commit to this. Finally, as data emerge over time, this recommendation may be modified in particular as we will likely be able to incorporate markers such as lymphocyte count, troponin, or d-dimer levels into our predictive models and enhance our ability to counsel families and make decisions.

IIIc. Is the sequential organ failure assessment (SOFA) score appropriate for triaging COVID-19 patients?

Background:

The first ICU triage protocol [74, 131] for use following the SARS pandemic in 2003 proposed use of the SOFA score [132]. The SOFA score [133], originally a sepsis score, seemed attractive given its simplicity and limited laboratory data required to calculate it compared with other predictive scores. Since first proposed, the SOFA score has become the basis of many triage scores, however, increasingly a number of limitations with the SOFA score have surfaced when proposed for use in triage [124, 134, 135].

Recommendation:
12.	We recommend against the use of the SOFA score for ICU triage of patients with COVID-19. (Strong recommendation, low-quality evidence)

Rationale:

Following the 2009 H1N1 pandemic where ICU triage was not required as resources were not overwhelmed, the performance of the SOFA score in predicting outcomes in critically ill H1N1 patients was evaluated by multiple research projects. A number of these studies raised concerns about the potential performance of SOFA for triage of patients with predominately isolated respiratory failure [136,137,138,139]. Although some of these studies reported a statistically significant difference in SOFA scores between ICU survivors and non-survivors, generally the SOFA scores in both groups on admission were low, often ≤ 7, and frequent survivors were seen with SOFA scores that reached > 11 during their admission.

Tang et al. published a study comparing their experience with critically ill H1N1 patients and COVID-19 patients in which they found the median SOFA scores on admission for COVID-19 patients were even lower than those of H1N1 patients (2 vs 5) [140]. Yang reported on 52 critically ill COVID-19 patients median SOFA scores on admission of 4 (range 3–4) for survivors and 6 (range 4–8) for non-survivors [36]. Similarly, Zang reported the median SOFA score in 55 critically ill COVID-19 patients was 5 (IQR 4–8). In a cohort of COVID-19 patients meeting the Berlin definition for ARDS, Liu [141] found their median SOFA score on admission to be 4 (IQR 2–5) [37]. Given that the majority of published triage protocols use a SOFA score threshold of ≤ 6 or 7 to identify the highest priority group (those most likely to survive and benefit from ICU resources) and with admission SOFA scores for both survivors and non-survivors being typically lower than the threshold, the protocols are not helpful for triage during the COVID-19 pandemic.

IV. Supporting Families and Staff

IVa. How do we manage family communication/visits/updates during the COVID-19 crisis?

Background:

Family-centered care [142, 143] in the provision of critical care is, and should remain, best practice at all times, even during an infectious disease outbreak. In keeping with this, every effort should be made to continue bedside family visitation during the COVID-19 pandemic [144, 145]. Enabling this requires specific guidance for visitors regarding PPE, clear signage, and support to ensure that family members are not attending hospital while ill and wearing PPE correctly to ensure their safety [146]. However, delivering this is challenging during surge situations due to the rapid changes in PPE guidance, human resources required to support this process, shortages of PPE, and the risk to both visitors and staff of disease transmission [145]. Restrictions to visitation should be evidence-informed and patients and families should be informed in advance of restrictions and their rationale, when possible.

Recommendation:
13.	In the event that bedside visitation by family members is not feasible due to surge conditions or PPE shortages, we recommend the following mitigation strategies be used in order to continue to deliver family-centered care: (Best practice statement)
13.1	Using available communication technology including mobile phones, videoconferencing, and messaging to enable family members to communicate with patients and staff
13.2	Using a 24/7 manned hospital phone line to address questions, concerns, special requests of family members
13.3	Engaging family members in rounds and patient care discussions (virtually) and providing technological solutions by the hospital to enable this
13.4	Engaging chaplains/spiritual care, social workers, ethics consultants, patient advocates to provide support to patients and their families

Rationale:

There is very limited evidence describing communication strategies with families during a pandemic. Existing reports are primarily in the setting of pediatrics; however, this information should also apply to the adult setting. It is recognized that during a pandemic visitation by family members is limited, and this can be a unique source of stress [147].

Communications technology, including cell phones and videoconferencing, have advanced rapidly and allow for novel approaches to facilitating communication between ICU teams, patients, and families. However, when employing novel technologies, it is important to ensure local information governance protocols are adhered to even in the pandemic setting. Utilizing existing infrastructure and ‘bring-your-own’ technology decreases the time required for implementation and costs for the hospital [144, 145, 147, 148].

Regularly engaging family during rounds builds a sense of normality to this very abnormal situation which may be comforting to both families and clinicians [144, 145]. Finally, utilising ‘non-clinician extenders’ to support families during not only off-loads clinicians who are short staffed may also provide greater consistency in support to families as well as creating an opportunity for these professionals to engage with and support clinicians [145].

IVb. What models of staff support can be used during the COVID-19 crisis?

Background:

Experiences from past outbreaks including SARS [149,150,151], H1N1 [152, 153], and Ebola [154] have documented the psychological impact they can have on healthcare workers. Given this is an identified risk, employers, and society in general, have a duty to provide support to healthcare workers during the COVID-19 pandemic in an effort to mitigate, to the degree possible, potential harmful impacts.

Recommendation:
14.	For employers, healthcare systems, and institutions during the COVID-19 pandemic:
14.1	We suggest implementing a specific program to enhance healthcare workers’ resilience to cope with psychological stressor during the COVID-19 pandemic. (Weak recommendation, low-quality evidence)
14.2	We recommend implementing programs to provide psychological support to healthcare workers throughout the COVID-19 pandemic. (Best practice statement)
14.3	We recommend implementing strategies which aim to mitigate both primary and secondary psychological stressors associated with the pandemic. (Best practice statement)

Rationale:

Programs specifically designed to build healthcare worker resilience to the psychological stressors associated with infectious disease outbreaks have been developed and demonstrated efficacy. Various approaches have been utilized including a personalized resilience plan combined with a self-triaging system [155], workshop-based training [156], and computer-assisted resiliency training [157]. A potential limitation of these strategies, however, is that they all required pre-exposure implementation so although they may benefit areas and systems which have not yet begun to receive significant volumes or COVID-19 patients, it is unclear how useful they will be at this point specifically in systems that are already in a surge situation or generally given that most countries are already well into the community spread phase of the pandemic.

Initiatives to provide psychological support for healthcare workers during the pandemic itself include strategies such as psychological first aid [158], on-site counseling drop in centers [150], and Internet-based psychological crisis intervention [159]. In order to robustly support HCWs, organizations must address both primary stressor (direct pandemic related stress) as well as secondary stressors (related to the basic needs such as physiologic and safety needs) [160, 161]. Healthcare organizations have direct influence issues such as PPE availability, work/rest ratio, nutrition at work, access to accommodations all of which may be utilized to minimize secondary stressors faced by HCWs.