Dynamic driving pressure in acute brain injury: a warning light, not yet a steering wheel

Mechanical ventilation in patients with acute brain injury (ABI) requires a delicate balance. The lungs, right heart, and diaphragm must be protected from ventilator-induced injury, while the brain must be protected from hypoxemia, dysregulated arterial partial pressure of carbon dioxide (PaCO₂), impaired venous drainage, reduced cerebral perfusion pressure, and abrupt changes in intrathoracic pressure. Low tidal volume ventilation and plateau pressure limitation remain the foundation of lung-protective ventilation (1). Driving pressure has emerged as an attractive physiologic marker because static driving pressure [plateau pressure minus positive end-expiratory pressure (PEEP)] relates delivered tidal volume to respiratory system compliance (2). Dynamic driving pressure (ΔPdyn), by contrast, is calculated from peak inspiratory pressure minus PEEP and is easier to observe at the bedside, but it is not physiologically identical to static driving pressure. Yet in neurocritical care, every ventilator adjustment may have cerebral consequences. The relevant question is therefore not only whether a ventilator variable predicts mortality, but also whether modifying that variable improves outcomes without compromising cerebral physiology.

In this context, the secondary analysis of the VENTIBRAIN study by Taran and colleagues is an important contribution (3). The investigators evaluated time-varying ΔPdyn, defined as peak inspiratory pressure minus PEEP, among 1,555 mechanically ventilated patients with traumatic brain injury, subarachnoid hemorrhage, intracranial hemorrhage, or acute ischemic stroke. Using Bayesian joint models, they found that higher ΔPdyn over the first 14 days of ventilation was associated with higher intensive care unit (ICU) mortality [hazard ratio (HR) 1.057 per 1 cm H₂O; 95% credible interval (CrI) 1.037–1.078; posterior probability of harm >99%]. The association was observed across ABI subtypes, persisted through hospital and 6-month mortality analyses, and was directionally similar when static driving pressure was examined.

The methodological approach is a strength. Joint modeling is well suited to this question because ΔPdyn is not a fixed baseline exposure: it evolves with lung mechanics, ventilator settings, disease progression, patient effort, and clinical decisions. The central descriptive claim is convincing: in brain-injured patients, as in general ICU populations, ΔPdyn tracks mortality. The harder question is what that observation licenses. The authors appropriately call for trials testing whether reducing ΔPdyn improves outcomes after ABI. That is the right next step, but it should be framed as a question of genuine uncertainty, not as confirmation that ΔPdyn is already a therapeutic target.

An integrated model of brain-lung protection

The most clinically significant finding is that the association was strongest among patients with both severe ABI and severe hypoxemia. This is biologically plausible. The injured brain may be more vulnerable to systemic inflammation, impaired carbon dioxide regulation, and hemodynamic instability; conversely, severe ABI predisposes patients to prolonged ventilation, deeper sedation, impaired cough, aspiration, and neurogenic pulmonary edema. The brain and lungs do not fail in isolation; they interact through inflammatory, mechanical, hemodynamic, and neurohumoral pathways (4,5). A recent state-of-the-art review further emphasizes that ventilator pressures affect cerebral hemodynamics through interactions among the central nervous, respiratory, and cardiovascular systems, and that ventilation in ABI should consider lung, diaphragm, right-heart, and brain protection together (6). VENTIBRAIN therefore shifts the discussion beyond “lung-protective ventilation in brain injury” toward an integrated model of brain-lung protection (7,8).

That said, the severe hypoxemia signal should be interpreted with caution. In correspondence following publication, You and colleagues noted that the apparent amplification of the ΔPdyn-mortality association at arterial partial pressure of oxygen to fraction of inspired oxygen ratio (PaO₂/FiO₂) ≤100 was clinically compelling and consistent with a “two-hit” lung-brain model, but vulnerable to overinterpretation because the most severe oxygenation subgroup was small and imprecise (9). In reply, Taran and colleagues repeated the analysis using a broader PaO₂/FiO₂ threshold of 200. The figure shows an approximate HR of 1.06 per 1 cm H₂O ΔPdyn in patients with PaO₂/FiO₂ ≤200 and approximately 1.05 in those with PaO₂/FiO₂ >200, with overlapping 95% CrIs; the authors appropriately characterized the result as sensitive to threshold selection and hypothesis-generating (10). That exchange is important. The clinical message differs depending on whether ΔPdyn is viewed as a continuous marker of worsening physiology, a categorical risk threshold, or a trigger for protocolized intervention.

Marker or mechanism?

The interpretive difficulty is that ΔPdyn is not an independent input a clinician freely dials up or down. It is partly an output of the ventilated respiratory system. For a given tidal volume, driving pressure is inversely related to respiratory system compliance; higher values may therefore arise, by construction, in patients with greater lung elastance, chest wall elastance, or both. A high ΔPdyn and a high risk of death may both be consequences of severe respiratory-system disease, without a direct causal arrow from the former to the latter. Higher ΔPdyn may also reflect atelectasis, aspiration, pulmonary edema, dyssynchrony, airway resistance, systemic illness severity, or clinical decisions made in response to deterioration.

Why negative controls cannot settle causality. The negative control analyses in VENTIBRAIN are reassuring, but they answer a narrow question. The absence of an association between ΔPdyn and blood transfusion or a simulated outcome (3,11) argues against gross model misspecification and against generalized confounding that would make ΔPdyn appear associated with unrelated outcomes. It does not prove that ΔPdyn itself causes death. A negative control outcome is useful because it should not share the pathway between exposure and outcome. For the same reason, it cannot detect bias that arises within that pathway. Blood transfusion is not downstream of lung elastance; death from respiratory failure may be. The clean negative-control results therefore support the validity of the observed association, but they do not establish causality, particularly when neurologic severity, lung injury, and withdrawal decisions may all interact.

What the interventional evidence adds

The interventional evidence remains uncertain. The work by Urner and colleagues, cited by VENTIBRAIN as its closest comparator for the strength of association, presents a large registry-based emulation in which the apparent survival benefit of reducing dynamic driving pressure was difficult to separate from overall adherence to lung-protective ventilation, rather than from maintaining a low driving pressure specifically (12). The prognostic signal, once again, may have reflected broader protective care rather than the independent effect of driving pressure reduction.

Our own work points in the same cautionary direction but should be read as hypothesis-generating. In an exploratory target trial emulation of mechanically ventilated adults with coronavirus disease 2019 (COVID-19) acute respiratory distress syndrome (ARDS), adding a sustained static driving pressure limit (ΔP <15 cm H₂O) to conventional low-tidal-volume and plateau-pressure-based ventilation did not improve 28-day survival compared with low-tidal-volume ventilation alone (13). The cohort was small, and patients with severe acute neurologic injury were excluded; therefore, the findings cannot be directly transported wholesale into ABI. They do, however, reinforce the central point that a prognostic association is not proof that intervention will improve outcomes. But the converse is also true: no existing trial or emulation has shown that driving pressure reduction improves outcomes in brain-injured patients. VENTIBRAIN’s effect-modification signal cannot be resolved by extrapolation from general ARDS data. It requires direct testing in ABI, with a defined intervention rather than an observed exposure.

The same number does not mean the same thing

ΔPdyn is attractive because it is continuously displayed and can be obtained even when plateau pressure has not been measured, is unreliable, or is difficult to interpret because of active respiratory effort. Convenience, however, is not interpretability, and static and dynamic driving pressures should be measured and interpreted together whenever feasible. During volume-controlled ventilation, the peak-to-PEEP difference includes flow-resistive components that do not necessarily represent alveolar distending pressure. During pressure-controlled ventilation, ΔPdyn may approximate the applied pressure above PEEP under passive conditions, but that approximation weakens as patient effort emerges. During pressure support or other assisted modes, airway pressure may misrepresent transpulmonary stress because the patient contributes to lung distension without a corresponding rise in measured airway pressure (14,15).

This matters in ABI because patients often transition from deep sedation and controlled ventilation to lighter sedation and spontaneous breathing. The same ΔPdyn value may carry a different physiologic meaning on day 2 than on day 7 in the same patient. The authors adjusted for ventilator mode in sensitivity analysis, and their reply confirms that mode-stratified analyses were limited by small cell counts (10). This is not a flaw in the study; it is a warning about bedside interpretation. A related tension is that ΔPdyn is promoted as continuously available, yet VENTIBRAIN analyzed values recorded once daily near 08:00. Whether transient excursions or cumulative exposure carry different prognostic weight remains unknown.

What a rising ΔPdyn should trigger

The practical risk is converting an association into a directive. A rising ΔPdyn should not trigger reflexive ventilator manipulation; it should trigger organized reassessment. Is the tidal volume appropriate, and have static driving pressure and plateau pressure been measured when feasible? Is the patient passive, or actively contributing to airway pressure? Is there bronchospasm, secretion burden, tube obstruction, biting, or dyssynchrony? Has compliance changed because of atelectasis, aspiration, pneumonia, edema, abdominal hypertension, or increased chest wall elastance? Is PEEP too low for recruitment, or too high for hemodynamics? Is PaCO₂ appropriate for this patient’s cerebral physiology? Are intracranial pressure and cerebral perfusion pressure tolerating the strategy under consideration?

There is still no validated bedside protocol for lowering driving pressure safely (13). To be fair, VENTIBRAIN reported clinically similar PaCO₂, PaO₂, and pH trajectories across ΔPdyn tertiles, which is reassuring for the observed ventilatory care in that cohort (3). But that observation does not prove that a future protocol designed to lower ΔPdyn would preserve PaCO₂, intracranial pressure, or cerebral perfusion pressure. Attempts to lower ΔP may require reducing tidal volume, increasing PEEP, altering inspiratory flow, deepening sedation, increasing respiratory rate, or accepting a higher PaCO₂. Each carries a trade-off. Lower tidal volume may increase dead-space ventilation and hypercapnia. Higher respiratory rate may increase mechanical power; therefore, a lower ΔP achieved through tachypnea could reduce one component of stress while increasing the energy delivered to the lung (16,17). Higher PEEP may recruit in some patients while worsening hemodynamics or intracranial physiology in others. In ABI, these trade-offs are amplified because PaCO₂, cerebral blood flow, intracranial pressure, and cerebral perfusion pressure are core therapeutic targets. ΔPdyn should function as a warning light, not as a steering wheel that overrides clinical judgment.

Which outcome, and whose death?

The outcome question deserves equal caution. In neurocritical care, mortality is incomplete: a strategy that improves survival while increasing survival with severe disability is not necessarily a success. VENTIBRAIN’s supplementary analyses are instructive: ΔPdyn and static ΔP were statistically similar across dichotomized 6-month Glasgow Outcome Scale-Extended categories (3). A driving pressure signal that predicts death may say little about recovery.

Withdrawal of life-sustaining treatment (WLST) compounds this, and it is where the limits of negative controls become concrete. In ABI, death may reflect physiologic deterioration, neurologic prognostication, institutional practice, clinician-family communication, and patient values; in VENTIBRAIN, 51 of 405 ICU deaths (12.6%) followed WLST (3). The investigators modeled WLST as a competing risk, which strengthens the study. Still, neurologic severity may increase both ΔPdyn and the likelihood of withdrawal. This is not the kind of bias a negative control outcome can detect, because it arises inside the clinical pathway connecting brain injury, ventilation, prognosis, and death. Future interventional work should prospectively adjudicate cause of death, neurologic prognosis, and withdrawal decisions. Otherwise, the field risks attributing to ventilator mechanics what partly belongs to decisions made downstream of neurologic injury.

Designing the trial that would settle it

The next phase should be interventional, but it will not be simple. A trial of ΔPdyn-guided ventilation in ABI must specify what clinicians are asked to do when ΔPdyn is high: the hierarchy of ventilator adjustments, thresholds for PaCO₂ and pH, oxygenation goals, intracranial pressure and cerebral perfusion pressure safeguards, and mandatory assessment for reversible causes of elevated peak pressure before protocolized changes. It should treat ΔPdyn as mode-dependent rather than mode-agnostic, and it should report absolute risks and patient-centered outcomes, not only HRs. Because WLST is common and outcome interpretation is value-laden, such a trial would also require prospective adjudication of neurologic prognosis and withdrawal decisions. These requirements make the design demanding, but they are also what would make the result clinically interpretable.

Target trial emulation can help refine feasible thresholds, mode-specific interpretations, adherence patterns, and safety signals before a large, randomized trial is launched. But emulation should refine the question, not substitute for prospective testing. Our own experience cautions in both directions: it exposed the missing titration protocol, but its precision could not close the question it opened (13).

For now, VENTIBRAIN should change what clinicians notice more than what they reflexively do. ΔPdyn is easy to see and harder to interpret. Its value may be greatest as an early signal of deteriorating mechanics in a population where lung injury can quietly worsen brain injury. The study should bring lung mechanics into daily neurocritical care rounds and dissolve the false dichotomy between lung and brain protection. The right care is neither permissive lung injury in service of the brain nor rigid ARDS protocols applied without regard to cerebral physiology. It is individualized ventilation that recognizes that the injured brain and injured lung are biologically connected.

Taran and colleagues have delivered a carefully characterized prognostic signal. The harder question is the one their data cannot answer: when ΔPdyn is high, what intervention improves the outcome that matters to the patient? Until that is answered, ΔPdyn should be measured, respected, and investigated, but not yet used as a protocolized steering wheel.

Acknowledgments

None.

Footnote

Provenance and Peer Review: This article was commissioned by the editorial office, Journal of Emergency and Critical Care Medicine. The article has undergone external peer review.

Peer Review File: Available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-2026-0035/prf

Funding: None.

Conflicts of Interest: The author has completed the ICMJE uniform disclosure form (available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-2026-0035/coif). The author has no conflicts of interest to declare.

Ethical Statement: The author is accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Acute Respiratory Distress Syndrome Network , Brower RG, Matthay MA, et al. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000;342:1301-8. [Crossref] [PubMed]
2. Amato MB, Meade MO, Slutsky AS, et al. Driving pressure and survival in the acute respiratory distress syndrome. N Engl J Med 2015;372:747-55. [Crossref] [PubMed]
3. Taran S, Citerio G, Taccone FS, et al. Dynamic driving pressure and clinical outcomes in mechanically ventilated patients with acute brain injury: a secondary analysis of the VENTIBRAIN study. Intensive Care Med 2026;52:31-41. [Crossref] [PubMed]
4. Mrozek S, Constantin JM, Geeraerts T. Brain-lung crosstalk: Implications for neurocritical care patients. World J Crit Care Med 2015;4:163-78. [Crossref] [PubMed]
5. Pelosi P, Rocco PR. The lung and the brain: a dangerous cross-talk. Crit Care 2011;15:168. [Crossref] [PubMed]
6. Georgopoulos D, Taran S, Bolaki M, et al. Mechanical Ventilation in Patients with Acute Brain Injuries: A Pathophysiology-based Approach. Am J Respir Crit Care Med 2025;211:932-45. [Crossref] [PubMed]
7. Taran S, Wahlster S, Robba C. Ventilatory targets following brain injury. Curr Opin Crit Care 2023;29:41-9. [Crossref] [PubMed]
8. Wahlster S, Sharma M, Taran S, et al. Associations between Driving Pressure and Clinical Outcomes in Acute Brain Injury: A Subanalysis of ENIO. Am J Respir Crit Care Med 2024;209:1400-4. [Crossref] [PubMed]
9. You T, Zhang L, Yuan Y. Dynamic driving pressure in acute brain injury: interpreting signals across ventilation modes and oxygenation severity. Intensive Care Med 2026;52:841-2. [Crossref] [PubMed]
10. Taran S, Citerio G, Taccone FS, et al. Dynamic driving pressure in acute brain injury: interpreting signals across ventilation modes and oxygenation severity-author's reply. Intensive Care Med 2026;52:843-4. [Crossref] [PubMed]
11. Lipsitch M, Tchetgen Tchetgen E, Cohen T. Negative controls: a tool for detecting confounding and bias in observational studies. Epidemiology 2010;21:383-8. [Crossref] [PubMed]
12. Urner M, Jüni P, Rojas-Saunero LP, et al. Limiting Dynamic Driving Pressure in Patients Requiring Mechanical Ventilation. Crit Care Med 2023;51:861-71. [Crossref] [PubMed]
13. Tanios M, Wu TT, Nguyen HM, et al. Comparing the impact of targeting limited driving pressure to low tidal volume ventilation on mortality in mechanically ventilated adults with COVID-19 ARDS: an exploratory target trial emulation. BMJ Open Respir Res 2024;11:e002439. [Crossref] [PubMed]
14. Bianchi I, Grassi A, Pham T, et al. Reliability of plateau pressure during patient-triggered assisted ventilation. Analysis of a multicentre database. J Crit Care 2022;68:96-103. [Crossref] [PubMed]
15. Stamatopoulou V, Akoumianaki E, Vaporidi K, et al. Driving pressure of respiratory system and lung stress in mechanically ventilated patients with active breathing. Crit Care 2024;28:19. [Crossref] [PubMed]
16. Costa ELV, Slutsky AS, Brochard LJ, et al. Ventilatory Variables and Mechanical Power in Patients with Acute Respiratory Distress Syndrome. Am J Respir Crit Care Med 2021;204:303-11. [Crossref] [PubMed]
17. Gattinoni L, Tonetti T, Cressoni M, et al. Ventilator-related causes of lung injury: the mechanical power. Intensive Care Med 2016;42:1567-75. [Crossref] [PubMed]