Prospective evaluation of an automated method to identify patients with severe sepsis or septic shock in the emergency department.

BACKGROUND: Sepsis is an often-fatal syndrome resulting from severe infection. Rapid identification and treatment are critical for septic patients. We therefore developed a probabilistic model to identify septic patients in the emergency department (ED). We aimed to produce a model that identifies 80 % of sepsis patients, with no more than 15 false positive alerts per day, within one hour of ED admission, using routine clinical data. METHODS: We developed the model using retrospective data for 132,748 ED encounters (549 septic), with manual chart review to confirm cases of severe sepsis or septic shock from January 2006 through December 2008. A naïve Bayes model was used to select model features, starting with clinician-proposed candidate variables, which were then used to calculate the probability of sepsis. We evaluated the accuracy of the resulting model in 93,733 ED encounters from April 2009 through June 2010. RESULTS: The final model included mean blood pressure, temperature, age, heart rate, and white blood cell count. The area under the receiver operating characteristic curve (AUC) for the continuous predictor model was 0.953. The binary alert achieved 76.4 % sensitivity with a false positive rate of 4.7 %. CONCLUSIONS: We developed and validated a probabilistic model to identify sepsis early in an ED encounter. Despite changes in process, organizational focus, and the H1N1 influenza pandemic, our model performed adequately in our validation cohort, suggesting that it will be generalizable.

Automated detection of physiologic deterioration in hospitalized patients.

OBJECTIVE: Develop and evaluate an automated case detection and response triggering system to monitor patients every 5 min and identify early signs of physiologic deterioration. MATERIALS AND METHODS: A 2-year prospective, observational study at a large level 1 trauma center. All patients admitted to a 33-bed medical and oncology floor (A) and a 33-bed non-intensive care unit (ICU) surgical trauma floor (B) were monitored. During the intervention year, pager alerts of early physiologic deterioration were automatically sent to charge nurses along with access to a graphical point-of-care web page to facilitate patient evaluation. RESULTS: Nurses reported the positive predictive value of alerts was 91-100% depending on erroneous data presence. Unit A patients were significantly older and had significantly more comorbidities than unit B patients. During the intervention year, unit A patients had a significant increase in length of stay, more transfers to ICU (p = 0.23), and significantly more medical emergency team (MET) calls (p = 0.0008), and significantly fewer died (p = 0.044) compared to the pre-intervention year. No significant differences were found on unit B. CONCLUSIONS: We monitored patients every 5 min and provided automated pages of early physiologic deterioration. This before-after study found a significant increase in MET calls and a significant decrease in mortality only in the unit with older patients with multiple comorbidities, and thus further study is warranted to detect potential confounding. Moreover, nurses reported the graphical alerts provided information needed to quickly evaluate patients, and they felt more confident about their assessment and more comfortable requesting help.