C105-13 Dynamic and Static Measurements of Lung Stress During Assisted Mechanical Ventilation in Patients With Acute Hypoxemic Respiratory Failure: A Multi-center Validation Study

Abstract Rationale Limiting driving pressure during mechanical ventilation can reduce lung stress and may improve patient outcomes. During assisted ventilation in spontaneously breathing patients, tidal lung stress can be assessed based on dynamic transpulmonary driving pressure (ΔPL,dyn,est) estimated using an expiratory occlusion pressure, but the safe upper limit for this parameter is uncertain. In contrast to quasi-static airway driving pressure measured during assisted ventilation (ΔPaw,static), ΔPL,dyn,est has the advantage of taking dynamic mechanisms of lung injury (e.g., pendelluft) into account. This study aimed to compare ΔPL,dyn,est with both airway and transpulmonary driving pressures measured under quasi-static conditions (ΔPaw,static and ΔPL,static, respectively) to identify the threshold that corresponds to ΔPaw,static above 15 cmH2O and to ΔPL,static above 12 cmH2O, signifying a potential increase in the risk of lung injury. Methods We conducted a secondary analysis of four independent patient cohorts (three adult and one pediatric) from South America, Europe, and North America where esophageal manometry was employed to monitor assisted mechanical ventilation in acute hypoxemic respiratory failure. The association between ΔPL,dyn,est and ΔPaw,static or ΔPL,static was assessed with mixed-effects linear models. Discrimination for measured ΔPaw,static 15 cmH2O or ΔPL,static 12 cmH2O was assessed with receiver operating characteristic curves. Summary diagnostic metrics pooling data from all four cohorts were computed using a random effects model and generic inverse variance. Results A total of 994 measurements from 250 patients from the 4 cohorts were analyzed. ΔPL,dyn,est was highly correlated with both ΔPaw,static (between-subjects R2 = 0.45, within-subjects R2 = 0.74; p 0.001) and ΔPL,static (between-subjects R2 = 0.31, within-subjects R2 = 0.70, p 0.001). ΔPL,dyn,est accurately detected high ΔPaw,static (AUROC = 0.84, 95% CI 0.76 - 0.91) or ΔPL,static (AUROC = 0.82, 95% CI: 0.77 - 0.87). ΔPL,dyn,est ≥21 cmH2O yielded the best compromise between sensitivity (79%) and specificity (76%) for ΔPaw,static of 15 cmH2O and for ΔPL,static 12 cmH2O (sensitivity 78% and specificity 76%). ΔPL,dyn,est ≥20 cmH2O was 85% sensitive, while a ΔPL,dyn,est ≥23 cmH2O was 85% specific to detect both ΔPaw,static 15 cmH2O and ΔPL,static 12 cmH2O (Figure 1). Conclusions ΔPL,dyn,est accurately detects excess quasi-static lung stress during assisted mechanical ventilation in both adults and children. Our findings suggest that a ΔPL,dyn,est of approximately 21 cmH2O balances between sensitivity and specificity to detect a quasi-static driving pressure greater than 15 cmH2O. However, the risks of sedation may favour a more conservative definition that prioritizes specificity over sensitivity. This abstract is funded by: Canadian Institutes of Health Research, PSI Foundation, FINEP, NHLBI