Breathing by the numbers: bedside lung mechanics reshape ARDS ventilation

When a patient with acute respiratory distress syndrome is connected to a ventilator, every setting on the machine is a physiological wager. Too much pressure and volume can tear open the very tissue the machine is trying to support; too little and the blood stops receiving enough oxygen. A new review published in Annals of Intensive Care argues that the answer to this dilemma has been sitting at the bedside all along, in the numbers that modern ventilators and a handful of simple maneuvers can already produce: respiratory system compliance, driving pressure, airway resistance, and mechanical power.

Its central claim is pragmatic rather than revolutionary: instead of applying a single protocol to everyone, clinicians can use real-time measurements of lung mechanics to align ventilator settings with each patient’s dynamic pathophysiology, reducing stress and strain on the lung while maintaining gas exchange. The authors are careful to note that several of the physiology-guided strategies they describe have strong rationale but have not yet been validated in randomized prospective trials.

The stakes are considerable. When the Berlin definition is systematically applied, ARDS affects roughly 10 percent of all intensive care unit admissions and up to 23 percent of patients requiring invasive mechanical ventilation. Lung-protective ventilation, introduced after the syndrome was first described in 1967, has improved outcomes, yet mortality remains high, reaching 46 percent in severe cases where the ratio of arterial oxygen to inspired oxygen falls below 100 mmHg. Survivors frequently carry long-term physical, cognitive, and psychological sequelae. In 2024, a new global definition expanded diagnostic criteria to non-intubated patients on high-flow nasal oxygen and allowed pulse oximetry surrogates and lung ultrasound, but both the Berlin and global definitions remain static, relying on single oxygenation measurements despite substantial fluctuations within the first 24 to 48 hours of ventilation. Recent work identifying three longitudinal oxygenation subgroups found these were more predictive of PEEP response and prognosis than a single static ratio.

The physiological core of the review is the “baby lung” concept. In ARDS, injury is spatially heterogeneous: consolidated and non-aerated regions coexist with relatively preserved areas, and the aerated, compliant portion available for ventilation behaves like a much smaller lung. Inflammatory exudate increases lung weight and creates a gravitational gradient of superimposed pressure that compresses dependent regions, so the baby lung is not a fixed structure but shifts with body position and mechanical forces, leading one team of investigators to describe the ARDS lung as a “mechanical sponge.” Crucially, the intrinsic compliance of the aerated regions often remains near normal, which means the predominant problem is reduced functional lung size rather than increased tissue stiffness. Protective strategies must therefore be tailored to this limited and vulnerable compartment.

From this concept flows the logic of measuring respiratory system compliance at the bedside. Compliance, the reciprocal of elastance, is calculated from a tidal volume divided by the difference between plateau pressure, obtained during an end-inspiratory occlusion, and total PEEP, obtained during an end-expiratory occlusion. In healthy spontaneously breathing individuals compliance typically ranges from 100 to 200 mL/cmH₂O; in intubated patients it falls to 50 to 80 mL/cmH₂O, mainly because functional residual capacity is reduced, but also because of dynamic external pressures such as elevated intra-abdominal pressure, pleural effusions, or chest wall edema. Constant loads like obesity, the authors note, primarily shift the pressure-volume curve without necessarily reducing compliance. An approximate prediction of normal compliance in intubated patients can be derived from vital capacity, estimated with sex-specific formulas based on height and age.

Driving pressure, defined as the difference between plateau pressure and total PEEP, is mathematically equal to tidal volume divided by compliance, and it reflects the distending force applied to the baby lung. Targeting a driving pressure of 15 cmH₂O or less, ideally lower, has been associated with improved survival in ARDS, and higher driving pressures increase lung stress even when tidal volume and PEEP sit within conventional limits. Current guidelines recommend tidal volumes of 4 to 8 mL/kg of predicted body weight, but the review argues that heterogeneity, encompassing not only the size of the baby lung but also the influence of patient sex and ethnicity on normal lung volume, challenges the idea of a similar tidal volume for everyone. Ongoing research is examining whether driving pressure can serve not only as a prognostic marker but as a means of individualizing tidal volume.

Airway resistance completes the basic panel. In intubated adults it typically ranges from 5 to 10 cmH₂O/L/s and can be estimated during volume-controlled ventilation from the difference between peak and plateau airway pressure divided by inspiratory flow. Increases caused by bronchospasm, secretions, or endotracheal tube obstruction raise peak pressure while plateau pressure remains unchanged, assuming no intrinsic PEEP; under pressure-controlled ventilation the same increases lower tidal volume instead.

Because airway pressures do not always reflect true lung-distending pressures, particularly in intra-abdominal hypertension or chest wall edema, the review describes partitioning lung and chest wall mechanics with an esophageal balloon catheter. This yields transpulmonary pressure at end-inspiration and end-expiration, along with separate lung and chest wall compliances and elastances. The direct measurement, however, reflects conditions near the esophagus, corresponding to dependent lung regions, and does not account for vertical pleural pressure gradients in the supine patient. An alternative elastance-derived method better represents transpulmonary pressure in non-dependent regions, which are more prone to overdistension, but it assumes pleural pressure is zero at functional residual capacity, an assumption that is not always accurate.

Mechanical power quantifies the energy transferred from the ventilator to the respiratory system per unit time, integrating tidal volume, airflow, respiratory rate, PEEP, and respiratory system mechanics. Because complete bedside calculation is often impractical, simplified equations exist for volume- and pressure-controlled modes, and a further surrogate proposed by Costa and colleagues, calculated as four times the driving pressure plus the respiratory rate, has shown predictive performance comparable to full calculations. Using this approach, patients with low compliance may benefit from lowering driving pressure through tidal volume adjustment, which strongly reduces mechanical power, whereas patients with higher compliance may not derive the same benefit. No universally accepted threshold for ventilator-induced lung injury exists: observational studies have reported that values above 12 J/min after 24 hours were associated with increased 90-day mortality, above 17 J/min with higher hospital length of stay and mortality, and above 22 J/min with fewer ventilator-free days and higher 28-day and 3-year mortality, so no single limit can yet be recommended clinically.

For PEEP, the review emphasizes that optimization requires evaluating lung recruitability, since higher PEEP can reopen collapsed alveoli but may also promote overdistension, increase dead space, and impair hemodynamics in patients with low recruitability. In the seminal work on recruitability, patients who gained at least 9 percent of aerated lung volume at higher pressure represented only around half of those studied. Although computed tomography across PEEP levels remains the gold standard, it is rarely feasible at the bedside. The recruitment-to-inflation ratio, proposed as a simpler bedside index, requires only a rapid reduction in PEEP within a clinically appropriate range, measurement of the first exhaled volume, and some calculations. A ratio below 0.3 indicates low recruitability and suggests limited benefit from high PEEP, while a ratio above 0.6 suggests substantial recruitability; intermediate values should be interpreted cautiously alongside plateau pressure, driving pressure, and hemodynamic status. The ratio should be viewed as a continuous variable reflecting the trade-off between recruitment and hyperinflation, its clinical validation remains limited, and performing it may be challenging or even risky in the most hypoxemic patients.

The review then lays out a stepwise, physiology-guided algorithm for deeply sedated or paralyzed patients on volume assist-control ventilation, explicitly noting that it rests on strong physiological rationale but has not been validated in randomized prospective trials. It begins with a tidal volume of 6 mL/kg predicted body weight and compliance measured at a baseline PEEP of 5 cmH₂O to estimate the baby lung. Near-normal compliance suggests much of the lung is already aerated and prompts assessment for overdistension, with tidal volume reduced if needed to keep driving pressure below 14 cmH₂O; low compliance prompts PEEP titration to the level that maximizes compliance, recognizing that very low PEEP has sometimes been associated with the best compliance due to intra-tidal recruitment, and that in the ART trial a strategy focused on recruitment maneuvers and compliance maximization was associated with increased mortality. Recruitability is then assessed by CT, pressure-volume curves, the recruitment-to-inflation ratio, or electrical impedance tomography; low recruitability favors PEEP of 5 to 8 cmH₂O, while high recruitability allows higher PEEP guided by a plateau pressure of 28 cmH₂O or less or maximal compliance with driving pressure limited to 14 cmH₂O. Gas exchange and hemodynamics are then reassessed, with optimal PEEP defined as maximizing oxygen delivery rather than arterial oxygen alone, driving pressure is targeted, and respiratory rate is adjusted to keep mechanical power as low as possible, with permissive hypercapnia generally tolerated but requiring caution in right ventricular dysfunction and acute brain injury.

During assisted breathing, the same principles apply with attention to patient self-inflicted lung injury. Plateau pressure can be measured during an end-inspiratory occlusion on a patient-triggered breath, and high early plateau pressure has been associated with poor outcomes. Dynamic transpulmonary pressure can be estimated using the drop in airway pressure during an end-expiratory occlusion as a surrogate for esophageal pressure swing, with a suggested safety threshold of 22 cmH₂O, though interpretation must account for conditions that increase resistance. Comparing these values with estimated inspiratory muscle pressure helps distinguish over-assistance, treated by reducing pressure support, from under-assistance, treated by gradually increasing it, with sedation or a return to controlled ventilation if excessive effort persists. This approach, too, has not been validated prospectively.

Finally, the review situates mechanics within a broader personalization framework. Hyperinflammatory subphenotypes, characterized by elevated cytokines, vascular leak, and low compliance, may benefit from higher PEEP, whereas hypoinflammatory phenotypes with preserved compliance may suffer overdistension from aggressive PEEP; many patients transition between phenotypes during the first four days of ventilation. Focal ARDS favors lower tidal volume and moderate PEEP, while diffuse ARDS may tolerate higher PEEP and recruitment maneuvers. A pooled analysis of seven ARDS Network and PETAL trials found the relationship between driving pressure and mortality significantly amplified in patients aged 80 or older, suggesting a lower target of 11 cmH₂O or less may be more appropriate in the very elderly. Artificial intelligence and omics-based tools hold promise for subphenotype identification, outcome prediction, and real-time ventilator adjustment, but clinical implementation remains limited by data heterogeneity, model interpretability, and workflow integration, and the authors conclude that further experimental and clinical studies are required to validate these approaches.

Subject of Research: A review of how bedside respiratory mechanics measurements can guide individualized ventilator settings in acute respiratory distress syndrome

Article Title: Bedside ventilatory settings guided by respiratory mechanics in acute respiratory distress syndrome

Article References: Chiumello, D., Coppola, S., Silva, P. L., Lais, G., Rocco, P. R. M., & Piquilloud, L. (2025). Bedside ventilatory settings guided by respiratory mechanics in acute respiratory distress syndrome. Annals of Intensive Care, 15(1), Article 189. https://doi.org/10.1186/s13613-025-01606-0

Image Credits: AI Generated

DOI: 10.1186/s13613-025-01606-0

Keywords: ARDS, mechanical ventilation, respiratory mechanics, driving pressure, compliance, mechanical power, PEEP, recruitment-to-inflation ratio, ventilator-induced lung injury, baby lung, esophageal pressure, artificial intelligence