The study included all triaged and treated patients aged 2–18 years who presented to the pediatric ED of Vanderbilt Children’s Hospital during a 4-week (1/27/06 – 2/24/06) study period. We excluded patients without electronic triage documentation, who left the ED or were transferred prior to physician evaluation, or whose final assigned diagnosis could not be established through chart review. One author independently established a reference standard diagnosis for all included patients by reviewing electronic and paper-based patient records. A diagnosis of asthma was defined as an assigned final ED diagnosis of asthma, wheezing, or reactive airway disease, or a suspicion of asthma, which was later ruled out by a trial of beta-agonist administration.
The BN was developed and trained on prospectively collected data from a previous 2-month period. The BN included 10 clinical variables from the current and previous patient encounters that had documentation in the electronic medical record, including billing codes. After a patient was triaged, the BN computed in real-time the probability of an acute asthma exacerbation. We evaluated the performance of the BN by calculating the area under the receiver operating characteristic (ROC) curve. Additionally, we determined operational characteristics at a fixed level of 85% sensitivity.