Clinical Decision Support for Septic Shock in the Emergency Department A Cluster Randomized Trial

Scott HF, Sevick CJ, Colborn KL, et al. Pediatrics. 2025;156(1):e2024069478. doi:10.1542/peds.2024-069478.

Introduction

Septic shock is a time-critical emergency in pediatric care, where delays in diagnosis and treatment significantly contribute to preventable morbidity and mortality. Early recognition and prompt initiation of antibiotics and fluid resuscitation remain key pillars advocated by international guidelines, such as those from the Surviving Sepsis Campaign. Integrating advanced technology into clinical workflows offers a promising avenue to improve early detection, with machine-learning-based Clinical Decision Support (CDS) systems increasingly utilized to flag patients at high risk before overt shock develops.

Despite encouraging technological advances, the translation of predictive accuracy into improved clinical outcomes in real-world pediatric emergency settings is unproven. This trial by Scott et al. evaluates the impact of a machine-learning CDS on timely sepsis care and outcomes in pediatric emergency departments, addressing a critical gap in implementation science and pediatric acute care.

Study Overview

Secondary Outcomes: Time-to-antibiotic administration, progression to hypotensive septic shock, plus qualitative feasibility and acceptability assessment

Key Findings

• No significant difference in primary outcome: 39.0% timely antibiotic and bolus in intervention vs 38.9% control (aOR 1.07; 95% CI 0.61–1.88)
• No significant reduction in progression to hypotensive shock (aOR 1.12; 95% CI 0.53–2.46)
• Time to antibiotic administration showed no improvement (adjusted HR 0.85; 95% CI 0.63–1.16)
• Qualitative feedback highlighted high provider acceptance and perceived unobtrusive nature of CDS alerts
• CDS tool remained continuously in use in participating EDs for 6 months post-trial, indicating sustainable integration

Evidence Synthesis and Clinical Context

The clinical imperative for timely pediatric sepsis care is established by the Surviving Sepsis Campaign international guidelines advocating rapid antibiotics and fluid bolus (1). The real-world evidence for clinical decision support systems in pediatric sepsis is limited, with few rigorous prospective studies previously conducted (2). Scott et al.’s trial contributes essential data by demonstrating feasible integration of a machine-learning CDS in active pediatric ED workflows, yet with no measurable effect on core clinical outcomes.

A key insight from this work is the complex challenge of moving from high technical model performance to actual clinical impact in the highly variable emergency care environment. For example, Lee et al. developed and validated deep learning models with promising diagnostic accuracy in retrospective EMR data (3). However, the null clinical findings of Scott et al. stress that technical prowess does not guarantee improved patient outcomes if implementation barriers and workflow factors are not addressed.

Contrasting outcomes come from Dewan et al., who reported improved care bundle compliance using a similar sepsis CDS in a pediatric ICU setting (4). The difference underscores distinct factors in the ED environment — such as rapid patient turnover and competing priorities — that may limit CDS effectiveness and require tailored approaches.

Clinical Implications

• Machine-learning CDS tools can be feasibly integrated and accepted by pediatric emergency providers, as shown by sustained post-trial use.
• However, the absence of impact on clinical outcomes indicates further optimization in alert design, timing, and integration into clinical workflow is needed before CDS can improve septic shock care.
• Clinicians should remain guided by established sepsis protocols emphasizing rapid antibiotic and fluid administration while CDS tools evolve.

Strengths & Limitations

Future Directions

Further investigation is needed on optimizing alert frequency, user interface, and incorporation of CDS within multidisciplinary pediatric sepsis care bundles. Prospective studies in diverse clinical settings, including ICUs, could clarify where predictive CDS can most effectively improve outcomes.

Conclusion

While feasible and well-accepted, machine-learning CDS for pediatric septic shock in emergency departments did not improve timely treatment or reduce progression to shock, underscoring the challenges of effective CDS integration in complex acute care workflows.

References

1. Weiss SL, Peters MJ, Alhazzani W, et al. Surviving Sepsis Campaign International Guidelines for the Management of Septic Shock and Sepsis-Associated Organ Dysfunction in Children. Pediatr Crit Care Med. 2020;21(2):e52-e106. doi:10.1097/PCC.0000000000002198. PMID: 32032273.
2. Ackermann K, Baker J, Festa M, McMullan B, Westbrook J, Li L. Computerized Clinical Decision Support Systems for the Early Detection of Sepsis Among Pediatric, Neonatal, and Maternal Inpatients: Scoping Review. JMIR Med Inform. 2022;10(5):e35061. doi:10.2196/35061. PMID: 35522467.
3. Lee JW, Lee B, Park JD. Pediatric septic shock estimation using deep learning and electronic medical records. Acute Crit Care. 2024;39(3):400-407. doi:10.4266/acc.2024.00031. PMID: 39266275.
4. Dewan M, Vidrine R, Zackoff M, et al. Design, Implementation, and Validation of a Pediatric ICU Sepsis Prediction Tool as Clinical Decision Support. Appl Clin Inform. 2020;11(2):218-225. doi:10.1055/s-0040-1705107. PMID: 32215893.
5. Scott HF, Sevick CJ, Colborn KL, et al. Clinical Decision Support for Septic Shock in the Emergency Department A Cluster Randomized Trial. Pediatrics. 2025;156(1):e2024069478. doi:10.1542/peds.2024-069478.