Grassi A, Bianchi I, Teggia Droghi M, et al. Increased Driving Pressure During Assisted Ventilation for Hypoxemic Respiratory Failure Is Associated with Lower ICU Survival — The ICEBERG Study. Am J Respir Crit Care Med. 2025;211(11):2169–2179.
Driving pressure (ΔP) has gained considerable traction as a marker of lung strain and a potential target for lung-protective ventilation during controlled mechanical ventilation. Whether it carries the same prognostic weight once patients transition to assisted breathing modes has remained an open question — one that the ICEBERG study set out to address in a prospective, multicenter design.
Study Design
The ICEBERG study (NCT05203536) enrolled 298 patients with acute hypoxemic respiratory failure across 16 ICUs in 9 countries. Patients were included if they had a PaO₂/FiO₂ ratio below 300 mmHg, were on invasive mechanical ventilation with evidence of spontaneous triggering, and were within 48 hours of initiating assisted ventilation. Respiratory mechanics were measured daily using inspiratory and expiratory hold maneuvers during the first 3 days of assisted ventilation, with outcomes tracked to Day 90.
Two key ΔP measurements were derived: static driving pressure (ΔP_stat), calculated as plateau pressure minus PEEP during an inspiratory hold, and dynamic transpulmonary driving pressure (ΔP_L,dyn), estimated from peak and occlusion pressures using the formula [(Ppeak − PEEP) − 2/3 × ΔPocc]. The latter incorporates the patient’s own inspiratory effort into the calculation — a meaningful distinction when the patient is actively contributing to each breath.
Key Findings
ICU mortality was 23% (67 of 298 patients). During the first 3 days of assisted ventilation, both ΔP_stat and ΔP_L,dyn were significantly higher in nonsurvivors compared to survivors (ΔP_stat: 12.6 vs. 11.3 cm H₂O; ΔP_L,dyn: 18.7 vs. 15.7 cm H₂O; both P < 0.001). Normalized respiratory system compliance (Crs/PBW) was correspondingly lower in nonsurvivors (0.65 vs. 0.79 mL/cm H₂O/kg; P < 0.001).
Notably, tidal volume, peak pressure, plateau pressure, and PEEP did not differ between groups. All patients had tidal volumes in a conventionally “safe” range (< 8 mL/kg PBW), reinforcing the concept that V_T alone is insufficient to characterize lung stress — it is the elastic properties of the system (reflected by ΔP = V_T / compliance) that appear to matter.
When patients were dichotomized at the median ΔP_stat (12 cm H₂O) or ΔP_L,dyn (16 cm H₂O), those above the cutoff had significantly higher 30-day mortality by log-rank testing. The longitudinal analysis showed that static ΔP diverged progressively between survivors and nonsurvivors over the study period, with an increasing trajectory in nonsurvivors.
In multivariable models adjusting for age, RASS score, and PaO₂/FiO₂ ratio, both ΔP_stat (HR 1.15 [1.03–1.30]) and ΔP_L,dyn (HR 1.14 [1.06–1.22]) were independently associated with ICU mortality. The RASS score was also a strong predictor (HR 0.60–0.63), suggesting that depth of sedation — and by extension, the clinical decision to lighten sedation — tracks with better respiratory mechanics and outcomes.
Clinical Relevance
The authors highlight several points of clinical relevance from these findings.
First, the study affirms the feasibility of measuring ΔP during assisted ventilation. Inspiratory hold maneuvers yielded readable plateau pressures on 91% of study days, though approximately 10% of measurements had to be discarded due to a non-flat plateau — a practical limitation clinicians should anticipate.
Second, the study introduces ΔP_L,dyn as a prospectively validated prognostic variable during assisted breathing. Because ΔPocc is a simpler maneuver that is less dependent on patient cooperation than the inspiratory hold, the dynamic ΔP estimate may be more consistently obtainable in clinical practice. In the time-dependent Cox analysis, daily ΔP_L,dyn was significantly associated with both 30- and 60-day mortality, while daily ΔP_stat showed a consistent but nonsignificant trend.
Third, the authors emphasize the “iceberg” concept: during assisted ventilation, the total distending pressure applied to the lung is the sum of the ventilator-generated pressure and the patient’s own muscular effort. Standard airway monitoring only captures the visible tip. As the investigators note, even patients with seemingly protective ventilator settings can generate injurious total pressures through vigorous spontaneous effort.
Limitations
This is an observational study, and the authors appropriately acknowledge that they cannot determine whether elevated ΔP is a marker of disease severity, a direct contributor to lung injury, or both. The overlap in ΔP values between survivors and nonsurvivors was substantial, limiting its utility as a standalone prognostic tool in individual patients.
The population was also relatively homogeneous in terms of inspiratory effort (mean effort within normal limits), and the degree of hypoxemia was mostly moderate (median PaO₂/FiO₂ of 205 mmHg). Whether these associations hold in patients with more extreme effort or more severe ARDS remains to be tested. Additionally, expiratory muscle activity — which can falsely lower the measured static ΔP — was not systematically accounted for. The authors recommend that a cautious approach should consider only values of ΔP well below 15 cm H₂O as “protective,” noting that an airway ΔP above 12 cm H₂O in their population (which may underestimate true transpulmonary pressure in the presence of expiratory muscle activity) was already associated with worse outcomes.
Bottom Line
The authors conclude that monitoring ΔP should not stop when the ventilator mode switches from controlled to assisted. They suggest that static and dynamic driving pressures, independently associated with outcome in this setting, might become targets for future interventional trials evaluating protective ventilatory strategies during spontaneous assisted breathing.
Reference: Grassi A, Bianchi I, Teggia Droghi M, et al. Increased Driving Pressure During Assisted Ventilation for Hypoxemic Respiratory Failure Is Associated with Lower ICU Survival — The ICEBERG Study. Am J Respir Crit Care Med. 2025;211(11):2169–2179. doi:10.1164/rccm.202411-2146OC
Trial Registration: ClinicalTrials.gov Identifier: NCT05203536