733: IMPACT OF ELECTRONIC SEPSIS ALERT ON PEDIATRIC SEPSIS OUTCOMES: AN INTERRUPTED TIME SERIES ANALYSIS

Abstract

Introduction: Pediatric sepsis is a significant cause of morbidity and mortality in children; guidelines now recommend a systematic screening tool to enhance early recognition of children with sepsis. However, it is unknown whether a screening tool that uses electronic clinical data to alert clinicians about a concern for sepsis can impact outcomes. We hypothesized that an electronic sepsis alert would decrease mortality, ICU admission, vasoactive use, and organ dysfunction at day 3 and day 7. Methods: This was a single-center, retrospective study of emergency department sepsis episodes from January 20, 2011 to May 20, 2021. Sepsis episodes were identified by a validated computational algorithm which reflects presence of infection with organ dysfunction (cardiac, respiratory, hematologic or renal) using data in the electronic health record. The sepsis alert was positive with elevated pulse rate or hypotension, concern for infection, and at least one of the following: abnormal capillary refill, abnormal mental status, or high-risk condition. We performed an interrupted time series analysis to compare outcomes before and after implementation of the electronic sepsis alert in March 2014. Interrupted time series analysis is a quasi-experimental design which can assess a level change (difference between predicted vs observed outcomes immediately post-intervention) as well as changes in trends (slope) before and after intervention. Outcomes included mortality, ICU admission, vasoactive use, and organ dysfunction at day 3 and day 7. Results: Of 3,269 total sepsis episodes, 770 occurred prior to and 2,429 occurred after sepsis alert implementation (70 missing alert status). After adjusting for season, there was no change in mortality (p=0.30), ICU admission (p=0.39) or vasoactive use (p=0.10). Sepsis alert implementation was associated with a significant change in organ dysfunction at day 3 (p=0.03) and day 7 (p< 0.0001). Specifically, there was an increasing slope of organ dysfunction prior to sepsis alert, which was stabilized after alert implementation. Conclusions: Using electronic clinical health data, implementation of a sepsis alert may impact rate of day 3 and day 7 organ dysfunction, though did not impact mortality, ICU admission or vasoactive use in our single center.