Background: To improve the safety of ventilator care and decrease the risk of ventilator-induced lung injury, we designed and tested an electronic algorithm that incorporates patient characteristics and ventilator settings, allowing near-real-time notification of bedside providers about potentially injurious ventilator settings.
Methods: Electronic medical records of consecutive patients who received invasive ventilation were screened in three Mayo Clinic Rochester intensive care units. The computer system alerted bedside providers via the text paging notification about potentially injurious ventilator settings. Alert criteria included a Pao2/Fio2 ratio of <300 mm Hg, free text search for the words “edema” or “bilateral + infiltrates” on the chest radiograph report, a tidal volume of >8 mL/kg predicted body weight (based on patient gender and height), a plateau pressure of >30 cm H2O, and a peak airway pressure of >35 cm H2O. Respiratory therapists answered a brief online satisfaction survey. Ventilator-induced lung injury risk was compared before and after the introduction of ventilator-induced lung injury alert.
Findings: The prevalence of acute lung injury was 42% (n = 490) among 1,159 patients receiving >24 hrs of invasive ventilation. The system sent 111 alerts for 80 patients, with a positive predictive value of 59%. The exposure to potentially injurious ventilation decreased after the intervention from 40.6 ± 74.6 hrs to 26.9 ± 77.3 hrs (p = .004).
Interpretations: Electronic medical record surveillance of mechanically ventilated patients accurately detects potentially injurious ventilator settings and is able to influence bedside practice at moderate costs. Its implementation is associated with decreased patient exposure to potentially injurious mechanical ventilation settings.
From the Departments of Internal Medicine (VH, MT, MK, AA, RK, CV, KS, SJT, RDH, OG) and Anesthesiology (BWP), Division of Pulmonary and Critical Care Medicine, and Multidisciplinary Epidemiology and Translational Research in Intensive Care (METRIC) (VH, MT, MK, AA, RK, CV, KS, SJT, BWP, RDH, OG), Mayo Clinic College of Medicine, Rochester, MN; and School of Health Information Sciences (JZ), University of Texas Health Science Center at Houston, Houston, TX.
Supported, in part, by National Heart, Lung and Blood Institute grant K23 HL78743-01A1, National Institutes of Health (NIH) grant KL2 RR024151, and grant 1 KL2 RR024151 from the National Center for Research Resources (NCRR), a component of the NIH, and the NIH Roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH.
The authors have not disclosed any potential conflicts of interest.
For information regarding this article, E-mail: firstname.lastname@example.org