Driving Pressure, Elastance, and Outcomes in a Real-World Setting: A Bi-Center Analysis of Electronic Health Record Data.

Abstract

Observational cohort study. Fourteen ICUs in two quaternary academic medical centers. Adult patients who received mechanical ventilation for more than 48 hours and less than 30 days. None. EHR data from 4,233 ventilated patients from 2016 to 2018 were extracted, harmonized, and merged. A minority of the analytic cohort (37%) experienced a Pao2/Fio2 of less than 300. A time-weighted mean exposure was calculated for ventilatory variables including tidal volume (VT), plateau pressures (PPLAT), DP, and ERS. Lung-protective ventilation adherence was high (94% with VT < 8.5 mL/kg, time-weighted mean VT = 6. 8 mL/kg, 88% with PPLAT ≤ 30 cm H2O). Although time-weighted mean DP (12.2 cm H2O) and ERS (1.9 cm H2O/[mL/kg]) were modest, 29% and 39% of the cohort experienced a DP greater than 15 cm H2O or an ERS greater than 2 cm H2O/(mL/kg), respectively. Regression modeling with adjustment for relevant covariates determined that exposure to time-weighted mean DP (> 15 cm H2O) was associated with increased adjusted risk of mortality and reduced adjusted ventilator-free days independent of adherence to lung-protective ventilation. Similarly, exposure to time-weighted mean ERS greater than 2 cm H2O/(mL/kg) was associated with increased adjusted risk of mortality. Elevated DP and ERS are associated with increased risk of mortality among ventilated patients independent of severity of illness or oxygenation impairment. EHR data can enable assessment of time-weighted ventilator variables and their association with clinical outcomes in a multicenter real-world setting.