Automated surveillance for ventilator-associated events

Area of Science:

• Critical Care Medicine
• Health Informatics

Background:

• The US Centers for Disease Control and Prevention (CDC) introduced a new, multitiered definition for ventilator-associated events (VAEs) to supersede the previous ventilator-associated pneumonia (VAP) definition.
• The study aimed to assess the feasibility and impact of implementing the VAE definition using electronic health records (EHR).

Purpose of the Study:

• To evaluate the automated implementation of the new VAE definition using EHR.
• To determine if the VAE definition identifies different patient cohorts compared to the VAP definition.
• To assess the reliability and efficiency of an automated VAE detection algorithm.

Main Methods:

• A retrospective cohort analysis was conducted on ICU admissions from 2008 to 2013 at a single urban, tertiary-care hospital.
• An automated algorithm was developed and applied to EHR data to identify VAEs.
• The algorithm's reliability was compared against human abstractors.

Main Results:

• Out of 26,466 ICU admissions, 10,998 (42%) required mechanical ventilation, and 675 (3%) had VAEs.
• VAEs were significantly associated with an increased risk of death (OR, 1.91).
• The automated algorithm demonstrated high reliability (sensitivity 93.5%, specificity 100%) and showed minimal agreement with the former VAP definition (κ = 0.06).

Conclusions:

• A fully automated method for identifying VAEs is efficient and reliable within a healthcare institution.
• While VAEs are linked to adverse patient outcomes, further research is needed to identify effective preventive interventions.
• The VAE definition identifies a distinct patient group compared to the VAP definition.