key: cord-0861916-6tlzij5v
authors: Weerakkody, Sampath; Arina, Pietro; Glenister, James; Cotterell, Sam; Boscaini-Gilroy, Giacomo; Singer, Mervyn; Montgomery, Hugh E
title: Non-invasive respiratory support in the management of acute COVID-19 pneumonia: considerations for clinical practice and priorities for research
date: 2021-11-09
journal: Lancet Respir Med
DOI: 10.1016/s2213-2600(21)00414-8
sha: 4182ec76aadef796e4461f1acd143542ff13e957
doc_id: 861916
cord_uid: 6tlzij5v

Non-invasive respiratory support (NIRS) has increasingly been used in the management of COVID-19-associated acute respiratory failure, but questions remain about the utility, safety, and outcome benefit of NIRS strategies. We identified two randomised controlled trials and 83 observational studies, compromising 13 931 patients, that examined the effects of NIRS modalities—high-flow nasal oxygen, continuous positive airway pressure, and bilevel positive airway pressure—on patients with COVID-19. Of 5120 patients who were candidates for full treatment escalation, 1880 (37%) progressed to invasive mechanical ventilation and 3658 of 4669 (78%) survived to study end. Survival was 30% among the 1050 patients for whom NIRS was the stated ceiling of treatment. The two randomised controlled trials indicate superiority of non-invasive ventilation over high-flow nasal oxygen in reducing the need for intubation. Reported complication rates were low. Overall, the studies indicate that NIRS in patients with COVID-19 is safe, improves resource utilisation, and might be associated with better outcomes. To guide clinical decision making, prospective, randomised studies are needed to address timing of intervention, optimal use of NIRS modalities—alone or in combination—and validation of tools such as oxygenation indices, response to a trial of NIRS, and inflammatory markers as predictors of treatment success.

Of the 246 million people infected with the SARS-CoV-2 virus by Oct 29, 2021, almost 5 million had died. 1 Among patients admitted to hospital with COVID-19, the predominant presenting feature is hypoxaemic respiratory failure, which often requires additional respiratory support over and above standard oxygen therapy. However, best practice remains unknown for this population and will be contingent on local availability of resources and trained staff.

Non-invasive respiratory support (NIRS) is routinely used in other conditions associated with acute hypoxaemic respiratory failure. 2 Continuous positive airway pressure (CPAP) can be delivered via a hood or helmet, or a tightfitting partial or full face mask. Bilevel positive airway pressure ventilation (BiPAP), primarily used in hypercapnic (type 2) respiratory failure, uses a similar interface to deliver additional inspiratory support over a continuous positive-pressure background. High-flow nasal oxygen (HFNO) devices, which can deliver 30-60 L/min (or higher) humidified gas flow via specially adapted nasal cannulae, reduce dead-space ventilation and deliver dynamic positive airway pressure. A recent meta-analysis showed that treatment with NIRS strategies (helmet or face mask non-invasive ventilation or HFNO) was associated with a lower risk of death in adults with acute hypoxaemic respiratory failure compared with standard oxygen therapy. 2 Initial guidance from WHO cautioned against the use of NIRS, citing potential risks of health-care worker infection through aerosolisation of viral particles, and patient self-inflicted lung injury from prolonged spontaneous hyperventilation. 3 However, the evidence base was so weak that 26 separate guidelines, most published before April 2020, offered a highly conflicting range of recommendations, including the following: avoidance of HFNO but use of CPAP only as a bridge to mechanical ventilation; use of HFNO but avoidance of non-invasive ventilation; or a preference for HFNO over non-invasive ventilation. 4 Rising case numbers in China, Europe, and the USA in the spring of 2020, allied with shortages of mechanical

• NIRS techniques (HFNO, CPAP, and BiPAP) were increasingly used in patients with COVID-19 respiratory failure with the aim of avoiding the need for invasive mechanical ventilation before data on safety and efficacy were available from RCTs • Only two RCTs comparing two modalities of NIRS in patients with COVID-19 have been reported, but many retrospective and prospective observational studies of NIRS have been published; marked heterogeneity exists between studies in terms of patient populations, including age and existing comorbidities, baseline illness severity, ward settings, and techniques used, either alone or in combination • Ceiling of treatment (respiratory support) was variably reported; regardless of non-invasive modality, patients who were candidates for full treatment escalation had better outcomes (78% survival at study end vs 30% for patients in whom NIRS was a ceiling of treatment) • 37% of patients for full treatment escalation who received NIRS progressed to invasive ventilation; the two RCTs indicate that non-invasive ventilation reduces the need for intubation compared with high-flow nasal oxygen • In multivariable analyses, age, comorbidities, baseline illness severity, degree of respiratory dysfunction before initiation of NIRS, response in respiratory variables to a trial of NIRS, and baseline concentrations of inflammatory markers were commonly identified as predictors of NIRS success or failure • Well designed RCTs are needed to provide an evidence base for optimised selection and use of NIRS devices, and appropriate use of prone positioning and adjuvant therapies, in the context of an individualised approach to management ventilators and intensive care unit (ICU) beds, 5 led to NIRS being increasingly adopted outside ICUs, with guidelines altered accordingly. Nonetheless, the role and benefits of CPAP and HFNO in the management of COVID-19 remain contentious, with lively debates about the timing of intubation and the risk-benefit balance between patient self-inflicted lung injury and ventilatorinduced lung injury. [6] [7] [8] [9] In this Personal View, we aim to provide an overview of what has been learnt so far about outcomes in patients with COVID-19 who received one or more NIRS modalities. We reviewed observational studies and randomised controlled trials (RCTs) of NIRS in COVID-19 pneumonia, examining duration of use, outcomes, predictors of success (ie, survival) or failure (ie, death or need for subsequent invasive mechanical ventilation), and changes in clinical practice over the course of the COVID-19 pandemic. We present our personal opinion as to what the findings mean for clinical decision making, and outline future directions for research and clinical practice.

Search strategy and selection criteria

We searched PubMed, Embase, ScienceDirect, Scopus, Google Scholar, and medRxiv for relevant studies published in English from Jan 1, 2020, to June 1, 2021, using the search terms ("COVID" OR "COVID-19" OR "SARS nCoV") AND ("CPAP" OR "continuous positive airway pressure" OR "NIV" OR "non-invasive ventilation" OR "NIPPV" OR "non-invasive positive pressure ventilation" OR "HFNO" OR "high-flow nasal oxygen" OR "HFNOT" OR "high-flow nasal oxygen therapy" OR "HFNC" OR "high-flow nasal cannula" OR "BiPAP" OR "bilevel positive pressure ventilation"). The following terms were also included to improve the scope and relevance of the search: "ARDS" OR "acute respiratory failure" OR "ARF" OR "hypoxaemia" OR "prone". Articles that did not include an assessment of CPAP, non-invasive ventilation, or HFNO to treat COVID-19 pneumonia were excluded. Reference lists (including those from reviews) were searched for articles not otherwise identified. Review articles were excluded unless they contained data not reported elsewhere. Letters, position papers, guidelines, case reports, and case series were excluded if they did not present original data or if the evidence presented was of poor quality or relevance. Nomenclature was often inconsistent, with the term noninvasive ventilation being used to describe either BiPAP alone, or both BiPAP and CPAP. Both are included as non-invasive ventilation where the precise modality was not specified. Literature searches were conducted by SW, PA, JG, and MS on the basis of prespecified inclusion criteria, which are summarised in the appendix (p 1). SW, PA, MS, and HEM reviewed 98 full-text articles for relevance to patient outcomes with the use of NIRS, conducted additional searches to avoid omissions, and identified other relevant papers that did not appear under the search categories. MS read each identified paper to confirm the accuracy of data extraction. The PRISMA statement guided our review and reporting. 10

The following data were extracted, where stated: country in which the study was done, type of hospital (university or community), clinical setting (ward, high-dependency unit, or ICU), type of NIRS used (HFNO, CPAP, or BiPAP), number of patients receiving a particular NIRS option, rates and duration of support in patient subsets in which the outcome was deemed to be a success or a failure (ie, outcome of subsequent invasive mechanical ventilation or death), complication rates, and mortality rates (eg, in-ICU, in-hospital, 28 

Search terms yielded 84 articles relating to a total of 13 140 patients after exclusion of duplicates, and after screening titles, abstracts, and full text for relevance specifically to outcomes with the use of NIRS in patients with COVID-19 (appendix p 1). Most papers originated from western European countries (62 studies), China (nine), and the USA (seven). A report of one large RCT from the UK was published as a preprint 95 after our original search and, because of its size, is included for completeness, giving a total of 13 931 patients (figure 1). Other than two multicentre RCTs from Italy and the UK, 38,95 22 prospective and 61 retrospective observational studies were identified, reporting a median of 71 patients (IQR 40-127) per study. Of these observational studies, 34 were multicentre and 49 single-centre studies. In total, 14 studies were based in community hospitals, 35 in university hospitals, and 16 in mixed institutions; the type of hospital was not specified in 20 studies. Patients were located in non-ICU wards (24 studies), in ICUs (42) , or in both (six); the location was not specified in 13 reports.

In total, 29 papers reported outcomes with CPAP, 23 with HFNO, and 6 with BiPAP as the sole modality of treatment. Use of two modalities was reported in 17 studies, and all three in 17 studies, although the reports rarely specified whether individual patients received more 11 Wang et al 12 Bellani et al 13 Franco et al 14 Liu et al 15 Vaschetto et al 16 Perkins 19 Calligaro et al 20 Chandel et al 21 Mellado-Artigas et al 22 Ferrando et al 23 Wendel Garcia et al 24 Lawton et al 25 Ramirez et al 26 Aliberti et al 27 Demoule et al 28 Daniel et al 29 Dupuis et al 30 Tonetti et al 31 Avdeev et al 32 Baqi et al 33 Vega et al 34 Celejewska-Wójcik et al 35 Vena et al 36 Deng et al 37 Grieco et al 38 Hu et al 39 Radovanovic et al 40 Patel et al 41 Arina et al 42 González-García et al 43 Di Domenico et al 44 Carteaux et al 45 Grosgurin et al 46 McDonough et al 47 Sivaloganathan et al 48 Menzella et al 49 Carpagnano et al 50 Duca et al 51 Bonnet et al 52 Potalivo et al 53 Bradley et al 54 Hernandez-Rubio et al 55 Duan et al 56 Brusasco et al 57 Walker et al 58 Zucman et al 59 Gaulton et al 60 Sargent et al 61 Roedl et al 62 Koduri et al 63 Kofod et al 64 Noeman-Ahmed et al 65 Faraone et al 66 Hallifax et al 67 Alviset et al 68 Thompson et al 69 Delbove et al 70 Lagier et al 71 Xia et al 72 Suardi et al 73 Panadero et al 74 Mukhtar et al 75 Oranger et al 76 Duan et al 77 Garner et al 78 Burns et al 79 Vianello et al 80 Corradi et al 81 Guy et al 82 Knights et al 83 Wang et al 84 Nightingale et al 85 Sayan et al 86 Winearls et al 87 Wozniak et al 88 Ashish et al 89 Pagano et al 90 Voshaar et al 91 Sartini et al 92 Lee et al 93 Xu et al 94 Kurtz et al 11 Wang et al 12 Bellani et al 13 Franco et al 14 Liu et al 15 Vaschetto et al 16 18 Coppadoro et al 19 Calligaro et al 20 Chandel et al 21 Mellado-Artigas et al 22 Ferrando et al 23 Wendel Garcia et al 24 Lawton et al 25 Ramirez et al 26 Aliberti et al 27 Demoule et al 28 Daniel et al 29 Dupuis et al 30 Tonetti et al 31 Avdeev et al 32 Baqi et al 33 Vega et al 34 Celejewska-Wójcik et al 35 Vena et al 36 Deng et al 37 Grieco et al 38 Hu et al 39 Radovanovic et al 40 Patel et al 41 Arina et al 42 González-García et al 43 Di Domenico et al 44 Carteaux et al 45 Grosgurin et al 46 McDonough et al 47 Sivaloganathan et al 48 Menzella et al 49 Carpagnano et al 50 Duca et al 51 Bonnet et al 52 Potalivo et al 53 Bradley et al 54 Hernandez-Rubio et al 55 Duan et al 56 Brusasco et al 57 Walker et al 58 Zucman et al 59 Gaulton et al 60 Sargent et al 61 Roedl et al 62 Koduri et al 63 Kofod et al 64 Noeman-Ahmed et al 65 Faraone et al 66 Hallifax et al 67 Alviset et al 68 Thompson et al 69 Delbove et al 70 Lagier et al 71 Xia et al 72 Suardi et al 73 Panadero et al 74 Mukhtar et al 75 Oranger et al 76 Duan et al 77 Garner et al 78 Burns et al 79 Vianello et al 80 Corradi et al 81 Guy et al 82 Knights et al 83 Wang et al 84 Nightingale et al 85 Sayan et al 86 Winearls et al 87 Wozniak et al 88 Ashish et al 89 Pagano et al 90 Voshaar et al 91 Sartini et al 92 Lee et al 93 Personal View than one modality. In some cases, patients escalated from HFNO to CPAP or BiPAP to offer increased respiratory support; in others, CPAP or BiPAP were switched to HFNO due to device, mask, or hood intolerance. 16 studies reported the levels of positive end-expiratory pressure used (median 10 cm H 2 O [IQR 7·5-11·0]). The median duration of NIRS, where reported, is shown in tables 1-3. 11-95 Results were variable, but most studies revealed a shorter duration of NIRS use in those for whom the intervention was deemed to be a failure (ie, outcome of death or escalation to invasive mechanical ventilation, depending on the ceiling of treatment).

Overall, survival in patients receiving NIRS across the full 85 studies was 66·3% (8702 of 13 125 patients), with survival reported from 7 days to 60 days. 59 studies reported longer-term outcomes (hospital survival or 60-day survival), in which 65% of patients (6132 of 9388) survived. The observational studies were too heterogeneous in terms of patient demographics, disease severity, and ward location to draw conclusions about superiority of one technique over another. The Italian RCT from Grieco and colleagues 38 compared HFNO with BiPAP in 109 patients; although the difference in median days free of respiratory support was not statistically significant, the rate of endotracheal intubation was significantly lower in the BiPAP group (30%) compared with those randomised to HFNO (51%), with more days free of invasive mechanical ventilation. Hospital 95 The primary composite outcome of need for tracheal intubation or mortality within 30 days was significantly lower for those who received CPAP (36·3%) compared with conventional oxygen therapy (44·4%). There was no difference between HFNO (44·4%) and conventional oxygen therapy (45·1%). Compared with oxygen therapy, there were relative reductions in mortality of 13·9% for CPAP and 6·5% for HFNO, and relative reductions in the need for intubation of 19·1% for CPAP versus oxygen and 1·2% for HFNO. Very few papers reported complications specifically related to the use of NIRS. Rates were low for studies that did report complications. 27, 37, 46, 80, 87, 95 Outcomes of NIRS in COVID-19 

Overall survival (ranging from ICU survival to 60-day survival) was 64·1%, reflecting the mix of patients who were or were not candidates for full escalation. Survival was similar with all individual modalities: 70% (IQR 53-81) for HFNO only, 69% (63-87) for CPAP only, and 62% (28-75) for BiPAP only. For studies of patients receiving CPAP, BiPAP, or HFNO, median survival was 65% (IQR 41-83). In eight of ten studies reporting, median duration of NIRS treatment was longer in patients for whom the intervention was a success.

54 papers described risk factors for failure of NIRS, in which the outcome was need for invasive ventilation or death if not for escalation (appendix pp [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] . The larger studies performed multivariable analyses, albeit on widely differing variables. A summary of frequently reported multivariable and univariate factors is shown in panel 1. Multivariable risk factors for failure included increasing age, number of comorbidities, illness severity, degree of hypoxaemia on hospital admission, degree of respiratory failure before institution of NIRS-often described as pulse oximetry oxygen saturation to fraction of inspired oxygen (SpO 2 /FiO 2 ) ratio, partial pressure of arterial oxygen to FiO 2 (PaO 2 /FiO 2 ) ratio, or SpO 2 /FiO 2 ratio divided by respiratory rate (ROX index)-poor improvement in respiratory function following a trial of NIRS over 1-6 h, and a raised concentration of C-reactive protein.

Univariate risk factors included other inflammatory markers, male sex, and respiratory rate, but these were often not carried over as independent predictors.

Some studies addressed the question of whether time to invasive ventilation has an effect on mortality. Dupuis and colleagues 30 29 found no difference in mortality between an intubation-first group versus those intubated after a period of noninvasive ventilation; however, mortality was significantly lower in those who could be maintained on non-invasive ventilation. Likewise, Menzella and colleagues 49 found a similar mortality rate in patients who had a trial with BiPAP and then required intubation compared with those who remained on BiPAP.

Mellado-Artigas and colleagues 22 propensity matched patients receiving invasive ventilation on day 1 of hospital admission against patients initially managed with HFNO. Use of HFNO was associated with sig nificant increases in ventilator-free days and a reduction in ICU length of stay with no difference in hospital mortality. However, Vaschetto and colleagues 16 reported that delay in intubation after CPAP was associated with an increased mortality risk (hazard ratio 1·093, 95% CI 1·010-1·184). Wendel Garcia and colleagues 24 propensity matched patients in a multinational registry who received oxygen therapy, HFNO, non-invasive ventilation, or invasive ventilation on day 1 of ICU admission. Non-invasive ventilation was associated with the highest overall ICU mortality, albeit not significantly different from mortality among those who were invasively ventilated. However, mortality in patients who did not progress to intubation was 36% in the non-invasive ventilation group, suggesting that high numbers of these patients had a ceiling of treatment. By contrast, registry data from 126 Brazilian ICUs 11 showed hospital mortality rates of 4·7% (73 of 1558 patients) for those managed on NIRS (predominantly CPAP or BiPAP) alone, 53% (457 of 865) for those who required escalation from NIRS to invasive ventilation, and 59% (1042 of 1765) for those who were invasively ventilated without a prior trial of NIRS. Compared with subsets of patients treated with NIRS only, the subsets requiring invasive ventilation after NIRS or requiring direct invasive ventilation were older, included more patients with frailty or comorbidities, and had a higher median illness severity and a lower median PaO 2 /FiO 2 ratio. Patients who received invasive ventilation after a trial of NIRS were of a similar age and had similar levels of frailty, comorbidities, and illness severity, but a lower baseline PaO 2 /FiO 2 ratio, compared with invasively ventilated patients who did not receive prior NIRS.

National and network database studies from ICUs in the UK, 96, 97 in France, Belgium, and Switzerland, 98 in Germany, 99 and in Brazil, 11 and a study of total hospital population from the UK, 100 have reported changes in management and outcomes of patients with COVID-19 over the first wave of the pandemic, 11,96,98,100 and between the first and second waves. 97, 99 All found temporal decreases in mortality rates and length of stay for survivors of COVID-19, and a marked shift towards NIRS with significant reductions in the use of mechanical ventilation despite, where reported, comparable degrees of illness severity. Although modelling did suggest a relationship between mortality reduction and use of NIRS, 11,100 a causal relationship between NIRS and improved outcomes has not been confirmed.

NIRS techniques (HFNO, CPAP, and BiPAP) have been widely and increasingly used in the management of COVID-19-related hypoxaemic respiratory failure. Initial fears about transmission to health-care workers wearing personal protective equipment have abated. Notably,

A variety of predictors of NIRS success (ie, survival) or failure (ie, death or need for subsequent invasive mechanical ventilation) have been identified in multivariable and univariate analyses. Frequently reported factors associated with treatment failure are listed. APACHE=Acute Physiology and Chronic Health Evaluation. FiO 2 =fraction of inspired oxygen. NIRS=non-invasive respiratory support (continuous positive airway pressure, bilevel positive airway pressure, or high-flow nasal oxygen). PaO 2 =partial pressure of arterial oxygen. ROX index=respiratory rate and oxygenation index (SpO 2 /FiO 2 ratio divided by respiratory rate). SOFA=Sequential Organ Failure Assessment. SpO 2 =pulse oximetry oxygen saturation.

spontaneous manoeuvres such as breathing, speaking and, particularly, coughing have recently been shown to generate more aerosol emission than CPAP or HFNO. 101 The debate about optimal timing of ventilation continues, as does uncertainty regarding the relative risks of patient self-inflicted lung injury versus ventilatorinduced lung injury. [6] [7] [8] [9] 85 observational studies 96, 97 and, to date, two single RCTs 38,95 report that, overall, invasive ventilation was avoided in 7710 (61·0%) of 12633 patients with COVID-19 who received NIRS; however, the proportion does vary according to patient demographics (age and comorbidity), baseline severity of respiratory and other organ dysfunction, ceiling of treatment, degree of systemic inflammation, and response to a trial of NIRS. Overall, approximately 81% of patients with no ceiling of treatment survive to at least 30 days, whereas about 30% of patients not deemed to be suitable for mechanical ventilation survive following treatment with NIRS.

From the observational studies, there was no clear difference in outcomes between use of different modalities of NIRS. However, marked heterogeneity of the patient populations, differing locations and expertise within the hospital, and wide variation in the baseline level of respiratory failure before starting NIRS make direct comparisons problematic. The Italian RCT reported a similar hospital mortality in patients randomised to either BiPAP or HFNO, although requirement for invasive ventilation was significantly lower in the BiPAP group (30% vs 51%). 38 The Recovery-RS RCT reported superiority of CPAP, but no benefit from HFNO, over conventional oxygen therapy in terms of the composite outcome of need for intubation or death within 30 days. 95 The use of non-invasive oxygenation strategies in adults with acute hypoxemic respiratory failure has been well established for several decades, although there remains a relative dearth of RCT data. A recent systematic review and meta-analysis identified 25 RCTs, comprising 3804 participants, that tested helmet or face mask non-invasive ventilation or HFNO against standard oxygen therapy. 2 The authors reported, with moderate certainty, that all techniques lower the risks of endotracheal intubation and all-cause in-hospital mortality, although risk of bias due to lack of blinding was deemed to be high.

A significant proportion of patients will progress from NIRS to invasive ventilation, if deemed to be appropriate, or enter an end-of-life care pathway. The observational studies, unsurprisingly, highlight increasing age, underlying comorbidities, and other organ dysfunctions as risk factors (appendix pp [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] . The three-times higher survival rates in candidates for full escalation over those in whom NIRS represents a ceiling of respiratory support reflects the greater frailty and poorer physiological reserve of the latter group. No studies have been done to identify thresholds of frailty or physiological reserve beyond which patients will not benefit from NIRS, although resource limitations will have an impact on capacity and the ability to provide NIRS.

Care should be taken in overall interpretation of the presented data because, so far, only two prospective RCTs have been reported. 38, 95 Multiple factors could influence outcome data, including changes in caseload and management practices geographically or over time. Examples of variability include the following: application of NIRS to different patient groups, such as elderly cohorts, or to cohorts deemed to be unsuitable for further therapy escalation; within-hospital location; pressure settings or flow rates used for NIRS; and duration of use. Scarcity of ICU beds, invasive ventilators, specific NIRS devices, or oxygen supply might also have dictated changes in local practice. NIRS might have been delivered at times in areas (and by staff) not normally deployed for care of the critically ill, and with a restricted ability to monitor clinical progress. Such factors will probably have changed as clinical burden, experience, and organisational structures evolved.

The use of any particular modality of NIRS is generally at the physician's discretion, and thresholds for transition to NIRS or invasive mechanical ventilation might have varied greatly between countries, centres, and clinicians, and over time. In some cases, use of NIRS might have represented a ceiling of treatment, given resource limitations or perceived futility of invasive mechanical ventilation. Again, clinical experience and resource availability might have changed such evaluations over time. Use of voluntary prone positioning and adjuvant therapies such as corticosteroids might have been applied in some cases or centres and not in others, or with differing frequencies over time. Finally, mortality rates might represent underestimates, because outcomes in some papers were provided only for ICU stay or at 7, 14, or 28 days after initiation of NIRS; in other instances, a proportion of patients were still within the particular outcome threshold at the time of publication submission. Where possible, these factors have been identified.

Specific NIRS modalities are often used for specific indications (eg, HFNO for hypoxaemia and BiPAP for hypercapnic respiratory failure). The devices are often used in tandem, in a complementary manner, in routine clinical practice, escalating to CPAP or BiPAP if HFNO proves to be inadequate, reverting to HFNO if patients become mask-intolerant, or switching between modalities for breaks, sleep, and so on. Individual patients might not tolerate a particular technique-eg, due to claustrophobia or discomfort with high air flows, or use of nasal cannulae, tight-fitting mask, or helmet. The support device needs to suit the patient rather than vice versa. Clinician expertise will be an important factor in achieving better success rates through operator confidence, optimisation of device settings such as pressures and flow rates, verbal encouragement and calming, targeted use of sedation and anxiolytics, and achievement of sleep and adequate hydration.

Various clinical scales and nomograms have been proposed to identify a potential need for intubation, but have yet to be prospectively tested. For example, Apigo and colleagues 102 developed a work-of-breathing score combining respiratory rate, nasal flaring and sternocleidomastoid use during inspiration, and abdominal muscle use during expiration. Liu and colleagues 15 developed a complex nomogram based on age, Glasgow coma scale, ROX (respiratory rate and oxygenation) index, use of vasopressors, and number of comorbidities to compute a probability of NIRS failure. NIRS failure is predicted in many studies by a greater degree of respiratory failure before initiation of NIRS, as well as higher concentrations of circulating inflammatory markers, indicating a more severe underlying inflam matory disease process in the lung (appendix pp 2-14). However, even studies comprising patients with moderate-to-severe respiratory failure (as judged by baseline PaO 2 /FiO 2 ratio, SpO 2 /FiO 2 ratio, or ROX index) found that many patients survived with NIRS alone, and this survival could often be predicted by a positive response to a trial of NIRS over several hours. 14, 18, 19, 21, 27, 34, 39, 41, 44, 46, 48, [53] [54] [55] [56] 59, 65, 66, 72, 74, 75, 77, 79, 93, 94 Clearly, the population of patients who recover with the use of NIRS alone will have a shorter length of ICU stay and a high survival rate; this association has been demonstrated in UK ICU database reports both across the first wave 96 and between the first and second waves. 97 In the first report, 96 use of invasive ventilation fell from 85·0% during February-March 2020 to 61·1% in April-July 2020, despite equivalent patient demographics and illness severity; length of ICU stay in survivors fell from a mean (SD) of 16·5 (9·9) days to 14·1 (10·4) days, but remained unchanged in non-survivors, whereas 28-day in-hospital mortality fell from 43·6% to 33·6%. The second report 97 notes that severity of respiratory failure (as judged by PaO 2 /FiO 2 ratio in the first 24 h of ICU admission) was worse over the second wave of COVID-19, but use of invasive ventilation decreased from 72·1% in the first wave to 54·1% in the second wave; ICU length of stay in survivors fell from a median of 12 (IQR 5-28) days to 7 (4-14) days, although non-survivors were spending an extra 3 days in intensive care. Similar reductions in the use of invasive ventilation and overall mortality have been found in studies from Brazil, 11 Francophone countries, 98 and Germany. 99 Potential confounders include increasing use of adjunct therapies such as dexamethasone, and greater overall expertise, confidence, and organisation in the manage ment of 

critically ill patients with COVID-19. As noted in observational studies, much of the use of NIRS has occurred outside the ICU. Prospectively collected pan-hospital data from 247 UK hospitals during the first wave (March-August 2020) indicated that, after adjustment, there was a 19% reduction in the odds of mortality per 4-week period (odds ratio 0·81, 95% CI 0·79-0·83), and the greater use of non-invasive ventilation accounted for 22·2% (0·94, 0·94-0·96) of the reduction. 100

Studies indicate that the use of NIRS in patients with COVID-19 is safe, reduces the need for intubation, improves resource utilisation, and might be associated with better outcomes. Moreover, NIRS might improve outcomes in patients receiving face mask oxygen who are not deemed to be suitable for escalation to invasive ventilation. However, outcomes also depend on patient factors such as age and comorbidities, and baseline illness severity prior to commencement of NIRS. From the two RCTs performed so far, non-invasive ventilation appears to be superior to HFNO in reducing the need for intubation, but many uncertainties remain.

The observational studies generally indicate that the duration of NIRS support is shorter in patients for whom NIRS fails (ie, in those who do not survive NIRS or who require escalation to invasive ventilation); however, there remains no clear indication as to the optimal timing of intubation in relation to subsequent outcomes. The current literature is conflicting, with studies variously reporting worse outcomes with either early intubation or late intubation, or no difference. 16, 21, 22, 24, 29, 30, 49 Studies also report prolonged use (>1 week) of NIRS in a substantial number of patients who survived and did not progress to mechanical ventilation (tables 2, 3).

Clearly, the ideal evidence base would comprise the findings of well-designed, clinically relevant, prospective, randomised studies with appropriate outcomes. Such outcomes include duration of mechanical ventilation, need for ICU admission, length of stay in the ICU and in hospital, and both short-term and long-term mortality and morbidity. Unfortunately, few prospective studies in critically ill patients with COVID-19 have examined outcomes beyond 28-30 days, despite the fact that many patients remain hospitalised beyond this point. 97 Longterm consequences in all patients with COVID-19 receiving different forms of non-invasive or invasive respiratory support, and over differing durations, are also unknown. Figure 2 provides an overview of various decision points along the patient pathway, based on our distillation of the available evidence. However, many questions remain and we list in panel 2 what we consider to be priority areas for research. These include indications for the institution of NIRS (eg, identification of demographic factors, laboratory markers, and physiological and clinical characteristics that might be associated with treatment success or failure) and the question of whether early institution of NIRS can positively modify disease progression, as suggested by Lawton and colleagues. 25 Studies should aim to provide an evidence base for an individualised approach to management, with optimised selection and use of appropriate NIRS device(s), and appropriate use of prone positioning. Objective criteria that assess the likelihood of patients with frailty and chronic comorbidity benefiting from either a trial of NIRS or progression to invasive ventilation need to be identified. Likewise, reliable markers predictive of NIRS failure need to be prospectively validated, especially across different hospital settings and countries, to guide optimal timing of intubation. Whether some or all patients benefit more from earlier intubation when NIRS failure is predicted rather than when it occurs is, we feel, a crucial question. Patients should not be left to struggle unduly for prolonged periods, but shortages of invasive ventilators and ICU beds might dictate otherwise at times.

We advocate an international, collaborative, and coordinated approach to the design of future prospective, randomised studies, with prespecified methods and outcomes. Given careful selection of centres that have matched equipment in adequate supply, sufficiently powered RCTs should be possible across a broad range of centres. Whether such objective criteria for the selection of patients who are likely to benefit from a trial of NIRS can be applied during periods of resource limitation is also worth examining in future studies to gauge the impact on outcomes of rationing interventions.

• Which physiological and laboratory markers can be used to guide optimal timing of intubation and invasive mechanical ventilation in terms of patient outcomes (duration of mechanical ventilation, length of stay in the ICU and in hospital, and short-term and long-term mortality and morbidity)? • Does early institution of NIRS positively modify disease progression and outcomes? • Acknowledging the various indications and variability in patient compliance for the different modalities of NIRS, how can they best be used-singly or in combination-for individual patients (ie, personalised medicine)? • Does prone positioning during NIRS offer any outcome benefit? • Can thresholds of frailty or chronic comorbidity be identified beyond which patients will not benefit from either NIRS or invasive ventilation? • Should optimal strategies differ during periods of resource limitation, in different hospital settings and different geographical regions, and for patients deemed to be unlikely to benefit from escalation to invasive ventilation?

ICU=intensive care unit. NIRS=non-invasive respiratory support.

Coronavirus disease. Statistics: cases overview

Association of noninvasive oxygenation strategies with all-cause mortality in adults with acute hypoxemic respiratory failure: a systematic review and meta-analysis

Clinical management of severe acute respiratory infection when novel coronavirus (2019-nCoV) infection is suspected

Non-invasive respiratory support strategies in COVID-19

The impact of high-flow nasal cannula use on patient mortality and the availability of mechanical ventilators in COVID-19

The respiratory drive: an overlooked tile of COVID-19 pathophysiology

P-SILI is not justification for intubation of COVID-19 patients

Proposal for Coronavirus disease 2019 management

A physiological approach to understand the role of respiratory effort in the progression of lung injury in SARS-CoV-2 infection

The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration

Evolving changes in mortality of 13,301 critically ill adult patients with COVID-19 over 8 months

Cardiovascular disease and severe hypoxemia associated with higher rates of non-invasive respiratory support failure in COVID-19

Noninvasive ventilatory support of COVID-19 patients outside the intensive care units (WARd-COVID)

Feasibility and clinical impact of out-of-ICU noninvasive respiratory support in patients with COVID-19-related pneumonia

A simple nomogram for predicting failure of non-invasive respiratory strategies in adults with COVID-19: a retrospective multicentre study

Outcomes of COVID-19 patients treated with continuous positive airway pressure outside the intensive care unit

Non-invasive ventilation for SARS-CoV-2 acute respiratory failure: a subanalysis from the HOPE COVID-19 registry

Predictors of intubation in COVID-19 patients treated with out-of-ICU continuous positive airway pressure

Helmet CPAP to treat hypoxic pneumonia outside the ICU: an observational study during the COVID-19 outbreak

The utility of high-flow nasal oxygen for severe COVID-19 pneumonia in a resource-constrained setting: a multi-centre prospective observational study

High-Flow Nasal Cannula therapy in COVID-19: using the ROX Index to predict success

High-flow nasal oxygen in patients with COVID-19-associated acute respiratory failure

Awake prone positioning does not reduce the risk of intubation in COVID-19 treated with high-flow nasal oxygen therapy: a multicenter, adjusted cohort study

Implications of early respiratory support strategies on disease progression in critical COVID-19: a matched subanalysis of the prospective RISC-19-ICU cohort

Reduced critical care demand with early CPAP and proning in COVID-19 at Bradford: a single-centre cohort

Outcomes of non-invasive ventilation as the ceiling of treatment in patients with COVID-19

Helmet CPAP treatment in patients with COVID-19 pneumonia: a multicentre cohort study

High-flow nasal cannula in critically ill patients with severe COVID-19

Non-invasive positive pressure ventilation versus endotracheal intubation in treatment of COVID-19 patients requiring ventilatory support

Association between early invasive mechanical ventilation and day-60 mortality in acute hypoxemic respiratory failure related to coronavirus disease-2019 pneumonia

Use of critical care resources during the first 2 weeks

Noninvasive ventilation for acute hypoxemic respiratory failure in patients with COVID-19

Clinical characteristics and outcome of patients with severe COVID-19 pneumonia at a public sector hospital in Karachi

High-flow nasal oxygen therapy in the treatment of acute respiratory failure in severe COVID-19 pneumonia: a prospective observational study

Clinical characteristics, management and in-hospital mortality of patients with coronavirus disease

The outcome impact of early vs late HFNC oxygen therapy in elderly patients with COVID-19 and ARDS

Effect of helmet noninvasive ventilation vs high-flow nasal oxygen on days free of respiratory support in patients with COVID-19 and moderate to severe hypoxemic respiratory failure: The HENIVOT randomized clinical trial

Application of high-flow nasal cannula in hypoxemic patients with COVID-19: a retrospective cohort study

Characteristics and outcomes in hospitalized COVID-19 patients during the first 28 days of the spring and autumn pandemic waves in Milan: an observational prospective study

Retrospective analysis of high flow nasal therapy in COVID-19-related moderate-to-severe hypoxaemic respiratory failure

Discriminating between CPAP success and failure in COVID-19 patients with severe respiratory failure

Incidence of pulmonary embolism in patients with non-invasive respiratory support during COVID-19 outbreak

Clinical characteristics and respiratory support of 310 COVID-19 patients, diagnosed at the emergency room: a single-center retrospective study

Continuous positive airway pressure for respiratory support during COVID-19 pandemic: a frugal approach from bench to bedside

Role of intermediate care unit admission and non-invasive respiratory support during the COVID-19 pandemic: a retrospective cohort study

The use of high-flow nasal oxygen in the ICU as a first-line therapy for acute hypoxemic respiratory failure secondary to Coronavirus Disease

Noninvasive ventilation for COVID-19-associated acute hypoxaemic respiratory failure: experience from a single centre

Effectiveness of noninvasive ventilation in COVID-19 related-acute respiratory distress syndrome

Bilevel and continuous positive airway pressure and factors linked to all-cause mortality in COVID-19 patients in an intermediate respiratory intensive care unit in Italy

Severity of respiratory failure and outcome of patients needing a ventilatory support in the emergency department during Italian novel coronavirus SARS-CoV2 outbreak: preliminary data on the role of helmet CPAP and non-invasive positive pressure ventilation

High flow nasal oxygen therapy to avoid invasive mechanical ventilation in SARS-CoV-2 pneumonia: a retrospective study

Sixty-day mortality among 520 Italian hospitalized COVID-19 patients according to the adopted ventilatory strategy in the context of an integrated multidisciplinary clinical organization: a population-based cohort study

Continuous positive airway pressure (CPAP) as a ceiling of care treatment for hypoxemic respiratory failure due to COVID-19

Outcomes of an intermediate respiratory care unit in the COVID-19 pandemic

High-flow nasal cannula for COVID-19 patients: a multicenter retrospective study in China

Continuous positive airway pressure in COVID-19 patients with moderate-to-severe respiratory failure

The role of CPAP as a potential bridge to invasive ventilation and as a ceiling-of-care for patients hospitalized with Covid-19-an observational study

Prediction of outcome of nasal high flow use during COVID-19-related acute hypoxemic respiratory failure

Early clinical experience in using helmet continuous positive airway pressure and high-flow nasal cannula in overweight and obese patients with acute hypoxemic respiratory failure from Coronavirus Disease

The prognostic value of chest X-ray in patients with COVID-19 on admission and when starting CPAP

Mechanical ventilation and mortality among 223 critically ill patients with coronavirus disease 2019: a multicentric study in Germany

Clinical characteristics and outcomes of 500 patients with COVID pneumonia: results from a single center

COVID-19 and acute respiratory failure treated with CPAP

Predictors of CPAP outcome in hospitalized COVID-19 patients

Effectiveness and safety of noninvasive positive pressure ventilation in the treatment of COVID-19-associated acute hypoxemic respiratory failure: a single center, non-ICU setting experience

Successful awake proning is associated with improved clinical outcomes in patients with COVID-19: single-centre high-dependency unit experience

Continuous positive airway pressure (CPAP) face-mask ventilation is an easy and cheap option to manage a massive influx of patients presenting acute respiratory failure during the SARS-CoV-2 outbreak: a retrospective cohort study

Patient characteristics and predictors of mortality in 470 adults admitted to a district general hospital in England with Covid-19

High flow nasal cannula oxygenation in COVID-19 related acute respiratory distress syndrome: a safe way to avoid endotracheal intubation?

High-flow oxygen therapy in elderly patients infected with SARS-CoV2 with a contraindication for transfer to an intensive care unit: a preliminary report

High-flow nasal oxygen in coronavirus disease 2019 patients with acute hypoxemic respiratory failure: a multicenter, retrospective cohort study

Risk factors for non-invasive/ invasive ventilatory support in patients with COVID-19 pneumonia: a retrospective study within a multidisciplinary approach

High-flow nasal cannula for Acute Respiratory Distress Syndrome (ARDS) due to COVID-19

Outcome of non-invasive ventilation in COVID-19 critically ill patients: a retrospective observational study

Continuous positive airway pressure to avoid intubation in SARS-CoV-2 pneumonia: a two-period retrospective case-control study

Use of high-flow nasal cannula and noninvasive ventilation in patients with COVID-19: a multicenter observational study

Predictors of failure of high flow nasal cannula failure in acute hypoxemic respiratory failure due to COVID-19

Improved survival following ward-based non-invasive pressure support for severe hypoxia in a cohort of frail patients with COVID-19: retrospective analysis from a UK teaching hospital

High-flow nasal cannula oxygen therapy to treat patients with hypoxemic acute respiratory failure consequent to SARS-CoV-2 infection

Diaphragmatic thickening fraction as a potential predictor of response to continuous positive airway pressure ventilation in Covid-19 pneumonia: a single-center pilot study

High-flow nasal oxygen: a safe, efficient treatment for COVID-19 patients not in an ICU

Characteristics and outcomes of patients with COVID-19 at a district general hospital in Surrey, UK

The experience of high-flow nasal cannula in hospitalized patients with 2019 novel coronavirusinfected pneumonia in two hospitals of Chongqing

Is continuous positive airway pressure (CPAP) a new standard of care for type 1 respiratory failure in COVID-19 patients? A retrospective observational study of a dedicated COVID-19 CPAP service

Impact of HFNC application on mortality and intensive care length of stay in acute respiratory failure secondary to COVID-19 pneumonia

Early conscious prone positioning in patients with COVID-19 receiving continuous positive airway pressure: a retrospective analysis

Positive role of continuous positive airway pressure for intensive care unit patients with severe hypoxaemic respiratory failure due to COVID-19 pneumonia: a single centre experience

CPAP management of COVID-19 respiratory failure: a first quantitative analysis from an inpatient service evaluation

Non-invasive CPAP in mild and moderate ARDS secondary to SARS-CoV-2

Conservative management of COVID-19 associated hypoxaemia

Respiratory parameters in patients with COVID-19 after using noninvasive ventilation in the prone position outside the intensive care unit

Risk factors for mortality and respiratory support in elderly patients hospitalized with COVID-19 in Korea

Early awake prone position combined with high-flow nasal oxygen therapy in severe COVID-19: a case series

An adaptive randomized controlled trial of non-invasive respiratory strategies in acute respiratory failure patients with COVID-19

Trends in intensive care for patients with COVID-19 in England, Wales, and Northern Ireland

Intensive Care National Audit and Research Centre (ICNARC)

Clinical characteristics and day-90 outcomes of 4244 critically ill adults with COVID-19: a prospective cohort study

Major differences in ICU admissions during the first and second COVID-19 wave in Germany

Changes in inhospital mortality in the first wave of COVID-19: a multicentre prospective observational cohort study using the WHO Clinical Characterisation Protocol UK

Aerosol emission from the respiratory tract: an analysis of relative risks from oxygen delivery systems

Development of a work of breathing scale and monitoring need of intubation in COVID-19 pneumonia