Breathing Without the Machine: Rethinking Oxygen Therapy in ARDS

Acute respiratory distress syndrome, or ARDS, has long been treated as a condition that demands a breathing tube. Yet a growing body of evidence, synthesized in a new review published in Intensive Care Medicine by Laveena Munshi of the University of Toronto, Massimo Antonelli of the Fondazione Policlinico Universitario A. Gemelli in Rome, and Jean-Pierre Frat of CHU de Poitiers, argues that the picture is far more nuanced. Non-invasive respiratory supports—high-flow nasal cannula, continuous positive airway pressure, and non-invasive ventilation—are increasingly used in acute hypoxemic respiratory failure, driven by better interfaces, growing clinician expertise, and the desire to avoid the complications of invasive mechanical ventilation. In the landmark LUNG SAFE study conducted before the COVID-19 pandemic, non-invasive ventilation was used in only about 15 percent of ARDS patients. Contemporary practice has evolved substantially since then, with accumulating evidence supporting high-flow nasal cannula and helmet ventilation, and with the 2024 Global Definition of ARDS formally recognizing noninvasively supported patients.

The central tension the authors describe is physiological rather than technological. Positive airway pressure can recruit collapsed alveoli and raise lung volumes, improving oxygenation through positive end-expiratory pressure, or PEEP. In patients with hydrostatic pulmonary edema and left ventricular failure, PEEP may even improve cardiac output by reducing both afterload and preload. But in ARDS, respiratory drive is often markedly increased, fueled by cortical and peripheral inputs, mechanoreceptor stimulation, and inflammatory mediators. The result can be excessive spontaneous breathing effort, generating large negative pleural pressure swings, elevated transpulmonary pressures, and amplified regional lung stress. This phenomenon, known as patient self-inflicted lung injury, or P-SILI, can promote overdistension and the same barotrauma, volutrauma, and atelectrauma classically associated with ventilator-induced lung injury—particularly in the reduced, non-uniform ‘baby lung’ that characterizes ARDS.

The evidence linking injurious effort to outcomes is concrete. In acute hypoxemic respiratory failure, persistent high inspiratory effort under non-invasive ventilation, measured by esophageal pressure swings, has been shown to predict non-invasive ventilation failure. Large tidal volumes above 9 milliliters per kilogram of predicted body weight—reflecting the combination of delivered pressure support and strong patient effort—have likewise been associated with poor outcomes. Despite this, standardized bedside assessment of inspiratory effort, particularly using esophageal pressure monitoring, remains limited and is not routinely implemented in intensive care units. The authors argue that this monitoring gap is one of the key obstacles to using non-invasive support safely, because clinicians currently lack a reliable, widely available way to distinguish patients whose spontaneous breathing is helping them from those whose breathing is injuring them.

Interface choice matters enormously in this calculus. Helmet non-invasive ventilation enables delivery of higher PEEP levels with minimal air leaks, and patient-ventilator asynchronies are generally less clinically relevant than with face masks. Even when a patient’s inspiratory effort fails to trigger pressure support, the patient continues to perform genuine inspiratory work by drawing airflow and volume from the helmet—true work of breathing rather than an isometric effort. Intriguingly, the helmet’s longer trigger delays and less efficient pressurization may induce desynchronization that limits excessive transpulmonary pressure swings and potentially enhances lung protection. The trade-off is monitoring: tracking tidal volumes through a helmet remains challenging, and the interface can provoke claustrophobia. Face masks, by contrast, struggle to maintain a seal in edentulous patients, those with beards, or individuals with nasogastric tubes.

The clinical evidence for non-invasive ventilation and CPAP in ARDS remains inconsistent and highly dependent on disease severity. Some studies and subgroup analyses suggest these modalities may be feasible in carefully selected patients with mild ARDS who lack hemodynamic instability and retain preserved mental status, potentially avoiding intubation. Helmet-delivered non-invasive ventilation was specifically found to reduce intubation compared with high-flow nasal cannula in patients with COVID-19. Ongoing randomized trials—including HENIVOT-2, HighCPAP, and HONOUR—are expected to clarify the role of these approaches further. In the meantime, the authors outline reasonable starting settings: PEEP of approximately 5 centimeters of water, up to 10, with pressure support of 5 to 8 centimeters of water above PEEP, rapidly titrated according to comfort, effort, tidal volume, and gas exchange. Patients with persistent hypoxemia may require higher PEEP of 8 to 12 centimeters of water via helmet.

High-flow nasal cannula occupies a different niche. Delivering heated, humidified oxygen at 30 to 60 liters per minute, it reduces nasopharyngeal dead space, generates a modest PEEP effect through resistance to expiratory flow, and limits entrainment of ambient air, producing more stable alveolar oxygen delivery. These effects decrease work of breathing compared with standard oxygen, reflected in reduced respiratory rate, less dyspnea, and improved comfort. Compared with non-invasive ventilation, however, it provides lower PEEP and less homogeneous tidal volume distribution, as assessed by electrical impedance tomography. Its practical advantages are considerable: it imposes no uncomfortable interface, does not require sedation, and allows patients to remain awake, eat, speak, and undergo prone positioning.

The trial evidence favors high-flow therapy as a first-line strategy for moderate disease. Randomized trials in acute hypoxemic respiratory failure report lower intubation rates compared with conventional oxygen or non-invasive ventilation, particularly in patients with a PaO2/FiO2 ratio below 200 millimeters of mercury, although the effect on mortality remains uncertain. Current guidelines, including French consensus recommendations developed by the DELPHI method, recommend high-flow nasal cannula as first-line oxygen therapy in acute hypoxemic respiratory failure at that severity threshold. Typical initiation involves flows of 40 to 60 liters per minute, commonly 50, with the fraction of inspired oxygen adjusted to maintain target saturation and flow subsequently tuned to the patient’s inspiratory demand and work of breathing. The authors conclude that high-flow therapy should be considered first-line non-invasive support for most patients with moderate ARDS, provided close physiologic reassessment identifies failure early enough to avoid delayed intubation.

Deciding between a non-invasive trial and immediate intubation rests on a defined set of red flags. Severe hypoxemia, hemodynamic instability, additional organ failure such as renal failure or delirium, impending cardiac or respiratory arrest, coma, inability to protect the airway, a large secretion burden, or poor tolerance of interfaces should all prompt consideration of upfront intubation. When none of these is present, clinicians should assess the severity of hypoxemia alongside respiratory drive and inspiratory effort. Marked accessory muscle use and persistent signs of high work of breathing signal P-SILI risk. A reasonable sequence is a trial of high-flow nasal cannula with rapid reassessment of oxygenation, respiratory rate, and accessory muscle response, escalating to non-invasive ventilation if symptoms persist—while vigilantly avoiding high tidal volumes. Hypercapnia or respiratory acidosis may favor non-invasive ventilation with pressure support, but should also lower the threshold for intubation if it worsens.

Timing of intubation is the field’s most persistent dilemma. Delayed intubation has been identified as a risk factor for mortality in several studies, yet no trial has directly compared intubation versus no intubation in acute hypoxemic respiratory failure. Instead, randomized studies have used prespecified intubation criteria without demonstrating differences in time to intubation across oxygenation strategies. The French consensus identifies cardiac or respiratory arrest from hypoxemia and persistent hypoxemia despite maximal oxygen strategy—PaO2/FiO2 below 60 millimeters of mercury or oxygen saturation below 88 percent—as major criteria for intubation, alongside shock, worsening respiratory failure with tachypnea above 30 breaths per minute, recurrent desaturation, respiratory acidosis, or declining consciousness. The ROX index, which divides the ratio of oxygen saturation to inspired oxygen fraction by respiratory rate, is the best-validated bedside tool for predicting high-flow therapy failure and should complement, rather than replace, clinical judgment.

Awake prone positioning, widely adopted during the COVID-19 pandemic as an adjunct to high-flow therapy or non-invasive ventilation, adds another physiological lever. By improving ventilation-perfusion matching, promoting dorsal recruitment, and reducing dependent atelectasis, it reduced intubation rates when adherence was successfully achieved, though no mortality benefit was demonstrated. For non-COVID ARDS, the physiological rationale is plausible but the evidence remains insufficient. The review’s overarching message is that no single device should be applied for a fixed duration; instead, matching must be individualized to a patient’s evolving physiology. Non-invasive supports are valuable early in ARDS, but their safe use depends on careful selection, close physiologic monitoring, and timely recognition of failure. The next frontier, the authors argue, lies in phenotype-driven strategies and objective markers of injurious respiratory effort that can guide escalation decisions with far greater precision than today’s clinical gestalt.

Subject of Research: Physiology-guided use of non-invasive respiratory support, including high-flow nasal cannula and non-invasive ventilation, in acute respiratory distress syndrome

Article Title: Non-invasive respiratory supports in ARDS: Physiology-guided use, pitfalls, and pathways to success

Article References: Munshi, L., Antonelli, M., & Frat, J.-P. (2026). Non-invasive respiratory supports in ARDS: Physiology-guided use, pitfalls, and pathways to success. Intensive Care Medicine. https://doi.org/10.1007/s00134-026-08616-x

Image Credits: AI Generated

DOI: 10.1007/s00134-026-08616-x

Keywords: ARDS, non-invasive ventilation, high-flow nasal cannula, CPAP, P-SILI, PEEP, helmet ventilation, intubation, ROX index, acute hypoxemic respiratory failure, awake prone positioning, critical care