A55-54 Composite Metrics Incorporating Pressure Thresholds and Elastic Power Improve Risk Stratification in Invasive Ventilation

Rationale

While clinical trials have demonstrated the benefits of low tidal volume ventilation, the relative importance of plateau pressure, driving pressure, and more recently mechanical power in predicting patient outcomes continues to be debated. Additionally, newer concepts such as the ‘hazard ratio’ combine energetics with a pressure threshold in an attempt to more accurately assess damaging ventilation. Our retrospective study compared traditional ventilatory parameters to hazard ratio-based metrics to determine which have the strongest associations with key clinical outcomes in patients with ARDS.

Methods

We analyzed a database of 47,380 invasively ventilated patients from a Learning Health Systems (LHS) data platform spanning 12 hospitals in the Fairview/University of Minnesota health system from 2011-2025. Data were systematically extracted from electronic health records and converted into the Common Longitudinal ICU Format (CLIF). We evaluated correlations between ventilation parameters and two primary outcomes: mortality and days receiving invasive ventilation. Multiple analytical approaches were employed to describe these correlations. We examined basic ventilation features (plateau pressure, driving pressure) alongside the hazard ratio-based composite features of elastic and driving power with thresholds that exceed 15 cmH₂O per inflation cycle.

Results

Hazard ratio features demonstrated the strongest correlations with mortality and ventilator days when compared to traditional metrics. Although driving pressure and total mechanical power were useful predictors of mortality and duration of invasive ventilation, hazardous elastic or driving power exceeding 15 cmH₂O per inflation cycle showed superior predictive performance.

Conclusions

Using complementary statistical methods, we identified composite metrics that integrate pressure thresholds and elastic power as more closely associated with mortality and ventilation duration than driving pressure or total power used independently. These findings suggest that current lung protective strategies may benefit from consideration of threshold-based power metrics. However, the retrospective nature of this analysis indicates that highly ranked associations do not necessarily imply causality or guarantee therapeutic benefit from targeted interventions. Further prospective studies are needed to determine whether controlling these composite metrics will improve clinical outcomes.

This abstract is funded by: None