Eisenberg M, Freiman E, Capraro A, Madden K, Monuteaux MC, Hudgins J, et al. Comparison of Manual and Automated Sepsis Screening Tools in a Pediatric Emergency Department. Pediatrics 2021;147(2):e2020022590. https://doi.org/10.1542/peds.2020-022590Question

Among children presenting to the emergency department (ED), what is the relative diagnostic accuracy of automated versus manual sepsis screening, each compared with a gold standard defined by standard sepsis definitions, direct admission to PICU, use of ED sepsis order set, or death within 72 hours of ED disposition?

Design

Retrospective cohort.

Setting

Boston Children's Hospital ED.

Participants

Patients screened manually (year 1) or automated (year 2) for sepsis.

Intervention

Manual vs automated sepsis screen.

Outcomes

Sepsis-detection screening-tool sensitivity between the time of ED arrival and 24 hours after ED disposition.

Main Results

Year 1, 14% of ED admissions manually screened: sensitivity, 64.6% (95% CI, 54.2%-74.1%), positive likelihood ratio, 7.2 (95% CI, 6.2-8.5), negative likelihood ratio 0.4 (95% CI, 0.3-0.5). Year 2, 100% of ED admissions automated screened: sensitivity 84.6% (95% CI, 77.4%-90.2%), positive likelihood ratio 17.1 (95% CI, 15.8-18.5), negative likelihood ratio of 0.2 (95% CI, 0.1-0.2).

Conclusions

Automated sepsis screening compared with manual screening demonstrated higher sensitivity.

Commentary

Operationalizing pediatric sepsis criteria as an electronic health record-integrated continuous real-time screening tool is a major step forward. This tool demonstrates readiness for clinical use. The automated alert is notable as a “sniffer,” identifying disease whether or not clinicians suspect it. Tools like these are important safety nets that ensure that sepsis is not missed in its most critical state: after organ failure has begun. This study has several strengths. Unlike in prior studies, outcome ascertainment was appropriately separated from actions triggered by the alerts. Severe sepsis was identified consistently across epochs. Survival analysis showed that, eventually, the automated screen detected more cases of sepsis. However, the median time to alert for the manual screen was 12 minutes versus 71 minutes for the automated screen. In 60% of patients with septic shock, the alert fired after hypotension. With false positives in 5% of all ED visits, potential consequences such as increased antibiotic use warrant evaluation. The search continues to predict sepsis. The next step in sepsis diagnosis studies is rigorous evaluation of the impact of screening tools on care quality and patient outcomes.

Article InfoIdentification

DOI: https://doi.org/10.1016/j.jpeds.2021.04.051

Copyright

© 2021 Published by Elsevier Inc.

ScienceDirectAccess this article on ScienceDirect Related Articles