Adverse ventilator settings—especially high tidal volumes—are the most important determinant of poor outcome in patients with acute lung injury (ALI) and its more severe form, acute respiratory distress syndrome (ARDS). At present there are no practical objective tools that can alert clinicians to an individual patient’s risk of ventilator-induced lung injury (VILI). Novel medical informatics methods such as continuous syndromic surveillance screening—extensively used in the prevention of bioterrorism—may be adaptable to this need, especially in the coming era of widespread electronic medical record (EMR) implementation. This masters thesis describes the application of syndromic surveillance algorithms (“sniffers”) to the EMR of intensive care unit (ICU) patients. The VILI sniffer continuously searches for lung injury diagnostic criteria and potentially injurious ventilator settings; the presence of both triggers an alert to bedside providers via the hospital paging system.

We compared the diagnostic performance of the sniffer and outcomes of mechanically ventilated ALI/ARDS patients before and after sniffer implementation. Outcome measures included adherence to lung protective ventilation, duration of mechanical ventilation, ICU and hospital mortality, ventilator- and ICU-free days, and performance of the alert system compared to expert review.

In an initial cohort of 3,795 consecutive critically ill patients admitted to nine multidisciplinary (ICUs) in academic tertiary care institution our automated algorithm detected ALI with 96% sensitivity (95% CI 94–98) and 89% specificity (95% CI 88–90). The VILI sniffer was subsequently evaluated in a prospective cohort over a 20 month period. Among 1,159 patients who met study inclusion criteria expert reviewers identified 490 cases of ALI. After sniffer implementation the exposure to potentially injurious ventilator settings decreased from 40.6 ± 74.6 hrs to 26.9 ± 77.3 hrs (p = .004). This computerized surveillance system accurately identified critically ill patients who developed ALI. We demonstrated the feasibility and preliminary effectiveness of fully automated EMR surveillance of mechanically ventilated patients at risk for VILI. Implementation of the VILI sniffer was associated with a reduction in exposure to potentially injurious mechanical ventilator settings.