Combining O₂ High Flow Nasal or Non-Invasive Ventilation with Cooperative Sedation to Avoid Intubation in Early Diffuse Severe Respiratory Distress Syndrome, Especially in Immunocompromised or COVID Patients?

Typical ARDS [6] comprises two entities [55]: early diffuse ARDS entails alveolar epithelial dysfunction, unstable alveoli, fluid-filled alveoli (non-cardiogenic pulmonary edema), bilateral infiltrates that ultimately coalesce into compressive atelectasis. A direct, proportional, relationship exists between the amount of non-aerated tissue and lowered compliance [6]. Typical ARDS is addressed with high PEEP [56], except in the setting of “focal “ ARDS [55].

Early COVID-ARDS entails pulmonary vascular endothelial dysfunction [47, 48], pulmonary vascular abnormalities [6], loss of hypoxic vasoconstriction with hyperperfusion of non-aerated, gasless tissue at variance with areas of no-perfusion and normal aeration [6], micro- and macroemboli [47, 48, 61], well aerated lung volume [62], high compliance and low driving pressure (DP) [63]. Intrapulmonary shunt (“shunt”) is perfusion of non-aerated alveoli (low or zero VA/Q [4]). The implication is that a high shunt fraction goes to gasless tissue [62]. Micro-emboli prevent recruited alveoli to participate in gas exchange. Venous admixture is intrapulmonary shunt+VA/Q mismatch [8]. In COVID-ARDS, VA/Q mismatch is more important than shunt i.e., predominantly low perfusion of ventilated alveoli. By contrast, in typical ARDS, shunt is more important than VA/Q mismatch i.e., adequate perfusion of nonventilated alveoli [8] (COVID-ARDS: high VA/Q and dead space; diffuse typical ARDS: low VA/Q) [63]). In COVID-ARDS, profound hypoxemia [48] occurs when compared to typical ARDS with same compliance. Typical ARDS presents with a higher P/F for the same compliance [6]). Recruitment is highly variable [63]. In the COVID-ARDS setting, low Vt results in increased dead space, reabsorption atelectasis, hypoventilation, hypercarbia, high hypercapnic drive and high sedative requirement. Low Vt-high PEEP conventionally proposed in typical ARDS appears of modest benefit in COVID-ARDS [6, 62].

Hypoxic vasoconstriction is relevant given the vascular disease. In pig lung injury, PS ventilation is associated with a redistribution of blood flow toward non-dependent better aerated lung without inducing recruitment. Increased aeration and improved hypoxic vasoconstriction occur in dependent regions [64]. Furthermore, alpha-2 agonists improve hypoxic vasoconstriction [65,66,67].

The mechanisms observed in early ARDS progress toward fibrosis more rapidly in COVID-ARDS compared to typical ARDS. Consequently, starting from admission, the intensivist is essentially racing against time, contending with ventilatory failure on one front and the rapid progression towards fibrosis on the other.

Hypoxemia results from reduced O₂ diffusion (typical ARDS) or inadequate alveolar perfusion (COVID-ARDS: micro- or macroemboli [48]) and is not necessarily accompanied by muscle dysfunction and signs of ventilatory failure, e.g., during “silent hypoxemia” [73,74,75]. Isolated hypoxemia without labored breathing is addressed with HFN/VHFN. Nevertheless, prolonged silent hypoxemia may lead to clinical deterioration, continued labored breathing, and eventually intubation.

The present opinion regarding clinical evolution aligns with recent guidelines on the use of HFN/NIV in the context of COVID (Table 1 [76]). One notable difference worth discussing is the emphasis on SaO₂>92% in the consensus paper [76], while others differentiate requirements based on P/F [23, 50]. Some experts [20, 21] stress the importance of observing the evolution of clinical signs, such as hyperpnea rather than tachypnea, over relying on oxygenation parameters. Intubation decision is not based upon SaO₂ or PaO₂ values [20, 21], or segregation with P/F [23, 50]. The indices help in identifying NIV failure within 2 h of treatment initiation [2, 72, 77] (HFN: ROX index: NIV; HACOR; table 1 in [10]). Subsequent observation is performed on an hourly basis. (Figure 1). Silent hypoxemia is a contributing factor to this approach. Additionally, hypoxemia acts as a short-acting, unsustained stimulus, known as “hypoxic ventilatory decline,” which primarily heightens the response to acidosis or hypercapnia [9, 74]. For instance, even when SaO₂ is ≤70%, a PaCO₂ of approximately 32 mm Hg (end-tidal <29 mm Hg) can prevent the hypoxic response [74]. Therefore, the conventional threshold (PaO₂ > 55–60 mm Hg, with SaO₂ > 92%) provides only a rough estimate, indicating merely the flat portion of the O₂ dissociation curve. Nevertheless, being outside this flat portion does not necessarily mandates intubation. The initial line of therapy focuses on restoring oxygenation and normalizing the hypoxic drive [78], which may not immediately necessitate intubation but instead requires close observation looking for worsening failure and a multimodal approach.

Muscle dysfunction involves overly active muscles. In contrast to the acute over chronic fatigue seen in COPD, muscle function in ARDS is typically intact at baseline, i.e. prior the onset of ARDS. However, muscle failure can occur due to various factors, including a) septic dysfunction [79], b) acute cardiac failure leading to exhaustion and death [80], and c) prolonged evolution (as mentioned above).

In early ARDS, a high inspiratory activity (“respiratory drive”, “drive”, “neural demand” [42, 43]) impinges upon intact muscles. A high muscular activity of intact muscles requires transpulmonary pressure to be addressed specifically (low inspiratory assistance, low pressure support: PS using upfront helmet NIV). This contrasts with acute over chronic fatigue of muscles observed in decompensated chronic COPD with reduced CO₂ excretion: in the setting of COPD, unloading the muscles is necessary for hours or days with high PS to overcome fatigue and decompensation [68]. By contrast, high PS is inappropriate for ARDS.

Respiratory and ventilatory physiology refer to brain stem processes vs. lung and chest wall function, respectively. Located in the lower brain stem, apposed to the vasomotor center, the respiratory generator (“generator”) controls the respiratory rhythm and phrenic activity and integrates the myriads of factors leading to the drive and the activation of the ventilatory muscles: (Vt, RR)=f(temperature, agitation, cardiac output, microcirculation, inflammation, lung water-diuresis, systemic pH, PaCO₂, PaO₂; Equation 1, to be used as a check list). ARDS patients present with a high drive, ventilatory muscular activity, inspiratory peak flow (“air hunger”) [2, 72, 81, 82], transpulmonary pressure, inflammation, labored breathing and impending failure. The higher drive present in COVID-ARDS patients led away from light sedation and spontaneous breathing (SB) [83,84,85], back to deep sedation, paralysis, protective ventilation and proning. As emphasized early April 2020 physiology was at loss in a bewildered world (francais.medscape.com/voirarticle/3605845?=null&icd=login_success_email_match_fpf&form=fpf). Dissecting and normalizing the myriads of factors [35, 74] involved in the genesis of hyperpnea and tachypnea allows one to lower the drive immediately following admission (“multimodal approach” [35]). Normalized drive rests on a functional generator at variance with the suppressed activity of the generator and suppressed drive caused by general anesthesia+paralysis.

As oxygenation is not the key issue anymore in ARDS [21, 35], the rationale for using high PEEP does not rest on oxygenation. Poor oxygenation (P/F<150) is unassociated with the increased inspiratory activity [2] but with inflammation [86].

In the setting of diffuse alveolar damage, solid-like behavior leading to pendelluft, increased spontaneous ventilatory effort [35, 87], atelectrauma, WOB and SILI are to be avoided. High PEEP prevents cyclic collapse of the bronchiolar tree [88] or of alveoli (atelectrauma) [89], thus suppresses the mechanical inflammation (SILI or VILI). As observed during the first breath after birth (−70 cm H₂O [90]), the first inflation of a kid’s balloon requires very high transmural (transalveolar) pressure; further inflation requires minimal incremental pressure. Once inflated by adequate PEEP, the “baby lung” in adult ARDS operates on the highest slope of the expiratory [91] P-V curve (highest compliance). The lung switches from solid- to fluid-like behavior [2], with a reduction in esophageal pressure changes and DP. The low PEEP achieved with HFN/NIV cannot recruit all atelectatic, non-aerated, areas. The objective is only to improve low VA/Q areas [92, 93], at variance with fully reopening the lung [94, 95] and correcting entirely the intrapulmonary shunt. Such a minimalist objective requires much lower PEEP levels. If so, “protective” ventilation is not protective because of low Vt, but because of PEEP and reduced solid-like behavior, WOB, pendelluft [87] and atelectrauma. Recruitment increases resting volume and FRC, lowers DP and decreases lung deformation (strain [5], Vt/end-expiratory lung volume ratio [6]). By contrast, low PS, Vt and transpulmonary pressure minimizes stress [5, 6]. In addition, under SB, the active diaphragm keeps the alveoli open during a longer expiratory interval [96], synergistically with PEEP.

The low PEEP achieved with HFN/VHFN/NIV may suit the relatively high compliance and low PEEP requirements observed in the setting of early COVID-ARDS [61, 63] and focal ARDS. PEEP is set as a function of the considered disease, using various techniques a) immediately following admission, a “one size fits all” approach uses the ARDS network table [97, 98]. Evidently, leaks observed in the setting of mask NIV would not allow setting the highest PEEP levels. This bears little consequences in the setting of focal ARDS or COVID-ARDS as lower PEEP levels are required when compared to diffuse ARDS. b) given a CT scan, rules of thumb are useful: “focal” ARDS: ~5 cm H₂O; diffuse ARDS: ~10 cm H₂O [55]; mild vs. severe: 5–10 vs. 15–20 cm H₂O [98]; COVID-ARDS: 8–10 cm H₂O [48] c) titrated to respiratory compliance [8] (COVID: avoid overdistension, increased dead space, hypercarbia and heightened drive; typical ARDS: recruitment of perfused units: low VA/Q, and increased compliance d) as soon as possible, an esophageal balloon individualizes PEEP [99, 100] in SB patients [2, 70]. e) impedance tomography or lung echography combined to arterial and venous gases and echocardiography are another option.

ARDS is managed [26] using intubation, general anesthesia [101] (GA renamed “deep sedation” [102], analgesia-sedation), paralysis [103], proning [22] and low DP [104]. Nevertheless, this is not a treatment for ventilatory failure [21]: CMV only buys time [49] for self-healing [21, 105] of the alveolus or of the capillary. Many issues are unsettled:

despite remarkable results [103], paralysis should be used sparingly, e.g., high drive [106].

The excellent epidemiological result [22] is methodologically weak. First, the results are not segregated between P/F<100 vs. <150, mixing severe and moderate ARDS. Second, no comparison exists between supine vs. prone vs. lateral+prone+lateral positioning; multiple repositioning is presumably the key to address compressive atelectasis, but not necessarily proning itself. Third, P/F returns towards baseline after turning supine, with no comparison to supine group (figure S2 [22]). The cause of ARDS is unaltered by proning; a rescue therapy causes no miracle. Simply the number of patients requiring proning progressively decreases, improved by multiple repositioning.

Mechanically, collapse is a function of the hydrostatic pressure imposed on the alveolus (i.e. the weight of the heart on the left lung). Thus, proning opens more non-dependent dorsal zones than it collapses in dependent sternal regions [112]. Indeed, proning leads often to a small improvement in PaO₂ [62], due to better VA/Q matching in vaso-dysregulated tissue [6] or perfusion redistribution in response to pressure or gravity [62] but not to alveolar recruitment. Given high compliance, minimal P/F improvement linked to CCU proning presents minimal interest in the setting of early COVID-ARDS, imposing on limited staff resources ([6] “responders”: P/F increase >20 mm Hg in 75% of the patients; [62, 113]). In intubated patients, proning vs. upright position increases P/F to a similar extent in patients with the lowest P/F (moderate and severe ARDS, panel B, Appendix vs. table S8 [7, 22]). Proning used in the setting of intubated paralyzed patients [22] was extended to awake non-intubated patients in the setting of early COVID-ARDS. Given these limitations, proning in awake non-intubated patients may avoid intubation [8].

These weaknesses leave recommendations [26] with shaky foundation [115]: “loss of muscle tone…. caused by muscle relaxants, anesthetics, and sedatives, and the use of high oxygen concentration in inspired gas are the prerequisites to produce atelectasis in…. healthy subject during anesthesia. This…. common treatment in ARDS… adds to the collapse and consolidation caused by the disease itself”.

To our surprise, with a tightly adjusted mask, Drager ventilators (Evita XL, Infinity V500) combined to cooperative sedation allows for achieving PEEP up to 20 cm H₂O with minimal leaks, for days [136]. Despite leaks and tolerance issues, since the pathophysiology of ARDS differs only in the amplitude of the dysfunction in intubated vs. non-intubated patients and the literature is limited, parameters set under invasive ventilation are used [9, 36, 37, 40]:

PEEP is set on a patient-per-patient basis given the heterogeneity of ARDS. Leaks lower PEEP; however, the patients treated with NIV are less severe or present to the CCU earlier in the evolution of their ARDS. An esophageal balloon inserted as early as possible allows for observing reduced esophageal pressure change (∆Pes<10 cm H₂O [2]) and improved labored breathing. In some patients, NIV appears successful within minutes when high PEEP combines with low PS [136], possibly restoring fluid-like behavior.

Patients presenting with a low inspiratory effort and small esophageal change on HFN require low PS, to avoid high transpulmonary pressure [82] during helmet NIV. When a high inspiratory effort and large negative esophageal change under HFN are observed, helmet NIV is superior to HFN (P/F≤200; shortest pressurization time, PEEP~10–12, PS~8–10 cm H₂O; reduced dyspnea, intermediate discomfort) [82]. The reduction of inspiratory effort during helmet NIV was larger in patients with the largest inspiratory effort during HFN, linked to inflammation or deteriorating mechanics, but not to oxygenation [86]. Accordingly, patients presenting with low PaCO₂<35 mm Hg benefit from helmet NIV, unlike patients with a high PaCO₂>35 mm Hg [10].

Partial muscle relaxation [144] may represent an additional tool when the negative evolution of esophageal swings leads to helmet NIV combined to a multimodal approach, before a decision to intubate. In patients presenting with ARDS and a high Vt>8 mL.kg⁻¹ a rocuronium infusion (5–37 mg over 6–60 min) was titrated to reduce the Vt (~9 to ~6 mL.kg⁻¹, with increased PaCO₂) and maintained for 2 h under conventional sedation (midazolam or propofol, sufentanil). Neurally adjusted ventilatory assist (NAVA) preserved diaphragmatic activity [144]. Such an approach may be useful in intubated or non-intubated patients under the care of anesthesia personnel with appropriate end tidal CO₂, Vt, SaO₂ monitoring. Although time-consuming, it may allow for the multimodal approach to achieve the temperature, agitation, systemic and microcirculation, kidney and metabolic goals under slow alpha-2 agonist sedation. Taken together, this suggests a 2 h window to improve the patient physiologically (HFN, NIV) [2, 72], then an additional 2 h using partial muscle relaxation [144], while running the multi-modal approach from admission onwards (Figure 1). Continued or increased labored breathing despite this full-fledged treatment implies intubation, and low PS under continued multimodal approach [9].

A baby lung allows only for baby O₂ consumption (VO₂) requirements. Thus, to reduce VO₂, temperature is lowered to 36<θ<35°C i.e., the lowest temperature of human at night. In patients with reduced cardioventilatory reserve, VO₂ is lowered [157] (~8–10% per °C [158] e.g., minus ~30% from 39.5 to 35.5°C). In ARDS patients, fever control is associated with improved survival [156]. Furthermore, in healthy volunteers, adrenaline infusion increases VO₂ and Vt (respectively: +11; +17% [159]) and the inspiratory flow [159], unlike a reduced drive. As the ARDS patient is often septic and requires vasopressors, they further increase VO₂.

By contrast, alpha-2 agonists lower a) the activation threshold of cold defense effectors (“set point”) [160,161,162] b) the temperature by >1°C in healthy volunteers [163] c) energy expenditure and VO₂ by ~15–18% [163, 164] d) muscular tremor [165] and VO₂, when baseline is high [166, 167]. Upon admission, paracetamol and external cooling are immediately followed by the administration of an alpha-2 agonist.

Agitation independent of the ventilatory failure (such as anxiety, delirium, pain) is to be addressed. Dexmedetomidine or clonidine are administered as first-line sedatives to stringent quietness (−2<RASS<0; up to the “ceiling effect”: dexmedetomidine or clonidine: 1.5 or 2 µg.kg.h⁻¹ respectively; no bolus administration; starting with low doses and tirating slowly; fill them up when hypovolemia is present; open them up if microcirculation is compromised [15,16,17]). The shorter half-life of dexmedetomidine facilitates nursing care (De Kock, personal communication). Alpha-2 agonists combine cardiac and vascular sympatholytic [168] and cardiac parasympathomimetic [169] actions, thus normalizing many factors within Equation 1. They evoke also sedation [170,171,172], slow wave sleep [173], normalize respiratory drive [174] with spontaneous breathing [39, 41], indifference to pain (“analgognosia” [175]) and to psychosocial or environmental stimuli (“imperturbability of mind”: ataraxia) [172], prevent delirium [176,177,178], reduce reactivity to noxious stimulus [179], especially in addict [180], young, combative patients. Importantly, alpha-2 agonists do not depress the respiratory generator (“generator”) [174]. The generator achieves adequate SB and NIV [181] without asynchrony and respiratory depression [163, 182]: a low Vt is observed with low or normal RR, according to temperature (35<θ<36°C). Indifference is achieved allowing for physiotherapy and continuous HFN/NIV [181, 183, 184] for days, without masking failure. Alpha-2 agonists lower the activity of the vasomotor center [185] and cardiac and vascular, arterial and venous, sympathetic activity, reduce the duration of CMV [13] and CCU stay [186], improve systolic [187] and diastolic [188] functions, normalize microcirculation [189], increase lactate clearance [190,191,192], lower noradrenaline requirement [193,194,195,196,197,198]. However, alone, alpha-2 agonists are useless. Only combined respiratory, ventilatory, circulatory, and autonomic interventions yield efficacy.

Supplementation: To achieve −2<RASS<0 and HFN/NIV for days, and given a ceiling effect [199], supplementation is sometimes required (“breakthrough”: haloperidol 2.5–10 mg i.v. bolus; infusion: haloperidol 50 mg/48 mL; 0.25 to 2 mL.h⁻¹ [15,16,17]). Given the depression of the generator evoked by midazolam [174], propofol and opioids, we advise against anesthetics and opioids. They depress the drive, impose closer observation and complicate the management. In addition, hyperpnea may resume after curare or sedation withdrawal [200]. Enforcing sleep-wake cycle is crucial [173].

Pain management differ between medical and surgical patients, with medical patients typically requiring fewer analgesics compared to surgical patients. Opioid-free analgesia can be employed to avoid respiratory depression and SB suppression (e.g., ketamine 50 mg+nefopam 100 mg+tramadol 400 mg, 48 mL, 0.1–2 mL.h⁻¹ [201]). Tramadol, being a weak opioid, has minimal respiratory depression effects. Cognition in elderly (nefopam) and acute kidney injury (tramadol) patients lead to rapidly lower the doses. The need for opioid-free analgesia typically decreases within 24–72h following administration of alpha-2 agonists.

Cardio-respiratory coupling [44, 45] and sympathetic normalization: First, there is a coordination between inspiratory phrenic and cervical sympathetic activity [202] (“respiratory-cardiovascular coupling”). Partial asphyxia evokes sympathetic activity throughout inspiration and expiration [202], in line with inadequate sympathetic hyperactivity in the setting of ARDS. Second, the interaction is also from the vasomotor center to the respiratory generator. Third, volume and vasopressors normalize the hypotension and the baroreflex-mediated sympathetic vascular hyperactivity. Following normalizing brain stem cardio-respiratory activity, attention can be focused on optimizing ventilatory mechanics, using appropriate tools (HFN, high PEEP-low PS, PEEP+CMV).

Alpha-2 agonists should not be used in cases of hypovolemia, sick sinus syndrome and atrio-ventricular block [15,16,17]. Positive pressure ventilation and PEEP require volume expansion to prevent hypotension and the need for vasopressors [203] as well as to avoid a pseudo-normalized intrapulmonary shunt [4].

To achieve adequate VA/Q, first, iterative echocardiography monitors the ventilation-induced changes in vena cava diameter, the right ventricular dilation, the mitral and aortic flows, the left ventricular [LV] contractility and the presence of foramen ovale (present in ~20% of the patients [204]). Various tools as volume, vasopressor, inotrope, pulmonary vasodilator are used to achieve adequate CO, mixed venous saturation, CO₂ gap, pH and lactate. Additionally, the combination of PEEP and SB acts synergistically. SB evokes diaphragmatic compression of the hepatosplanchnic blood [205] enhancing venous return, while PEEP decreases LV afterload [206]. Second, in addition to arterial and venous gases, impedance tomography or lung echography may assist in observing adequate VA.

Sympathetic normalization and improved microcirculation: First, the heightened vascular sympathetic activity is associated with high lactate [207]. The alpha-2 agonists normalize the sympathetic activity, the microcirculation [189] and the lactate concentration [189,190,191,192]. Systemic and regional metabolic acidosis is normalized within ~3–6 h, lowering peripheral inflammation and respiratory drive. Lastly, acute kidney injury and associated metabolic acidosis are managed with renal replacement therapy. In summary, counterintuitively, the treatment approach for ARDS prioritizes circulatory intervention [4, 46, 203, 208, 209] followed by ventilatory strategy.

Patients have transitioned from young trauma patients with preserved immune system at baseline and heading into severe delayed injury-acquired immunodeficiency [210], to elderly patients presenting with chronic baseline heightened inflammation, such as those with COVID-ARDS. Acute inflammation can result from conditions like sepsis, emphasizing the importance of early source control, or systemic acidosis, or impaired ventilatory mechanics (SILI or VILI).

Reducing lung water is crucial [221, 222] when inflammation [223] play a significant role, such as in high permeability edema or large negative esophageal changes. Once CO is normalized, volume infusion should be minimized. Indeed, in the setting of SB, low Vt and compliance [222], a ~10–15% CO increase to passive leg raising does not necessarily indicate the need for further hydration. To minimize lung water, the overall response is considered, at variance with BP or CO themselves: mottling, capillary refill time [224], urine output, venous SO₂, lactate, CO₂ gap, vena cava ventilatory changes.

Additionally, a) SB facilitates better lymphatic drainage compared to CMV [225] b) sympathetic blockade reduces pulmonary vein pressure and lung edema [226] c) alpha-2 agonists evoke diuresis through an anti-ADH effect [227]. The issue is not anymore the total volume of fluids administered during early resuscitation or the first day on admission, but the overall balance of fluids and weight achieved after 24, 48, 72 h d) following organophosphate poisoning, clonidine suppresses capillary filtration, thus pulmonary edema [228]. f) following lung contusion, clonidine improves inflammation [229].

Hypocapnia is an ominous sign in the setting of early ARDS [72]. PaCO₂ is lower when NIV failure occurs [2, 72], irrespective of P/F (>200 [72]; 101–170 [2]). This hypocapnia is close to the apneic threshold (healthy volunteer: ~30–35 mm Hg; NIV failure: 32 mm Hg [72]; high inflammatory status e.g. COVID: ≤30 mm Hg). This suggests switching early to helmet NIV [82, 120].

The striking observation is the occurrence of hyperpnea and hypocapnia, below the apneic threshold [230, 231]. Indeed, the threshold is overridden by systemic acidosis or central nervous system inflammation or stimulation of lung receptors [42, 43]. Athletes enduring Vt≥3 L, minute ventilation>160 L.min⁻¹ and esophageal pressure changes≥60 cm H₂O for hours [77, 119, 232] suggest that increased Vt per se is not detrimental, but rather inflammation plays a significant role. Could this be a consequence of pH, PaCO₂ or metabolic or cortical excitatory inputs onto the respiratory generator? If so, the respiratory generator should be made refractory to psychosocial stimuli generated by the CCU environment without suppressing respiratory genesis itself [174], and SB. This approach aims to alleviate increased respiratory drive and sympathetic hyperactivity, without resorting to general anesthesia and paralysis.

In COVID-ARDS, under paralysis+CMV, micro- or macrothrombi leads to high dead space, hypercapnia and a high respiratory drive: a low normal temperature (35–36°C) will help normalizing the VCO₂ and hypercapnic drive, allowing for SB, under HFN/NIV.

Hypoxemia, silent without or with overt failure, requires immediate treatment. Nevertheless, alleviating hypoxemia is not the ultimate objective:

Improved oxygenation and reduced mortality are unrelated [233]. Thus, low SaO₂ alone is not an indication for intubation [21]; rather labored breathing and impending/overt failure are.

Rather than simultaneously lowering the FiO₂ and the PEEP [97], they are adjusted sequentially [37, 41, 236].

The supine position worsens sick human (reduced FRC, increased abdominal pressure with atelectasis next to the diaphragm) [237]. Thus, the upright position presents some rationale to improve oxygenation [7]. Nevertheless, the head up position may worsen compliance and driving pressure in late ARDS (“paradoxical” positioning [240]). Furthermore, the rationale for extended upright intervals in a healthy human does not automatically transfer to a sick biped. To our knowledge, upright has not been documented in the setting of COVID-ARDS. As VHFN/NIV may evoke gastric dilation [2], the intraabdominal pressure should be reduced early (gastric and bladder catheters, enhanced intestinal motility).