Using AI to Improve Sepsis Quality of Care in the Emergency Department

NCT07581340 | Completed

• 1 other identifier: 1403
• Study Type: interventional
• Target: 66
• Locations: 1 country
• Sites: 1

Brief Summary

Sepsis is a life-threatening condition caused by the body's response to infection and is a leading cause of death worldwide. Hospitals use a complex quality measure called SEP-1 to track whether patients with severe sepsis or septic shock receive recommended care, such as timely antibiotics, fluids, and laboratory testing. However, evaluating SEP-1 is difficult. It requires manual review of medical records, is time-consuming and expensive, and typically provides feedback to clinicians months after care is delivered. This delay limits the ability to improve care in real time. This study tested whether artificial intelligence (AI), specifically a type of system called a large language model (LLM), could improve the quality of sepsis care by providing faster and more detailed feedback to physicians. The study was conducted at two emergency departments within a large academic health system. Sixty-six attending physicians were randomly assigned to one of two groups. In the intervention group, the AI system reviewed each patient's medical record at the time of hospital discharge and determined whether SEP-1 care standards were met. Physicians then received near real-time, individualized feedback about their performance, including specific areas for improvement. In the control group, physicians received standard feedback based on a small sample of cases reviewed months later using traditional methods.

Conditions

Sepsis

Keywords

Sepsis; Artificial intelligence; Implementation science; Large language model; Quality improvement

Outcome Measures

Primary Outcomes (1)

• SEP-1 Compliance

  SEP-1 (the Severe Sepsis and Septic Shock Early Management Bundle) is a quality measure developed by the Centers for Medicare \& Medicaid Services that evaluates whether patients with sepsis receive a set of time-sensitive interventions and diagnostic tests. It is a binary process measure (e.g., either a "pass" or "fail").

  This was assessed from time of the event (e.g., development of severe sepsis / septic shock) up to 1 month after the event.

Secondary Outcomes (1)

• In-hospital mortality

  From when the patient arrives to the hospital to their discharge, either alive or expired, up to 6 months after the event (development of severe sepsis / septic shock).

Study Arms (2)

Intervention group

EXPERIMENTAL

Participants in the intervention arm receive near-real-time, individualized feedback on SEP-1 performance generated by a large language model (LLM) that performs automated chart abstraction at the time of emergency department discharge.

Behavioral: Near-real time automated feedback on SEP-1 performance

Control group

NO INTERVENTION

Participants in the control arm receive standard sepsis quality feedback processes without real-time augmentation. This is much less than the intervention group and typically 3-4 months after a particular interaction.

Interventions

Near-real time automated feedback on SEP-1 performance BEHAVIORAL

Participants in the intervention arm receive near-real-time, individualized feedback on SEP-1 performance generated by a large language model (LLM) that performs automated chart abstraction at the time of emergency department discharge.

Intervention group

Eligibility Criteria

• Age: 18 Years+
• Sex: all
• Healthy Volunteers: No
• Age Groups: Adult (18-64), Older Adult (65+)

You may not qualify if:

• Patients who do not meet CMS SEP-1 criteria for severe sepsis or septic shock Sepsis onset occurring prior to emergency department arrival or after hospital admission Encounters without sufficient clinical documentation to assess SEP-1 compliance Patients transferred from another facility with ongoing sepsis care already initiated Cases in which the attending physician of record is not assigned to a study arm

Sponsors & Collaborators

• University of California, San Diego: lead

Study Sites (1)

UC San Diego Health

San Diego, California, 92103-1911, United States

Location

Related Publications (1)

• Boussina A, Krishnamoorthy R, Quintero K, Joshi S, Wardi G, Pour H, Hilbert N, Malhotra A, Hogarth M, Sitapati AM, VanDenBerg C, Singh K, Longhurst CA, Nemati S. Large Language Models for More Efficient Reporting of Hospital Quality Measures. NEJM AI. 2024 Oct 24;1(11):10.1056/aics2400420. doi: 10.1056/aics2400420. Epub 2024 Oct 21.

  PMID: 39703686 BACKGROUND

MeSH Terms

Conditions

Sepsis

Condition Hierarchy (Ancestors)

Infections; Systemic Inflammatory Response Syndrome; Inflammation; Pathologic Processes; Pathological Conditions, Signs and Symptoms

Study Design

• Study Type: interventional
• Phase: not applicable
• Allocation: RANDOMIZED
• Masking: NONE
• Purpose: OTHER
• Intervention Model: SINGLE GROUP
• Sponsor Type: OTHER
• Responsible Party: PRINCIPAL INVESTIGATOR
• PI Title: Associate Professor

Model Details: Randomized quality improvement study on the physician level within a group.

Study Record Dates

• First Submitted: April 13, 2026
• First Posted: May 12, 2026
• Study Start: December 1, 2024
• Primary Completion: August 1, 2025
• Study Completion: December 12, 2025
• Last Updated: May 12, 2026

Record last verified: 2024-10

Data Sharing

• IPD Sharing: Will not share

Locations

US (1)