# Evaluation of sepsis CDS tool knowledge and utilization among graduate medical trainees: insights to inform redesign in an academic health system

## TL;DR

This study evaluates how well sepsis decision support tools are understood and used by medical trainees, finding that usability and alert fatigue limit their effectiveness.

## Contribution

The study provides insights from trainees on how to redesign sepsis CDS tools for better workflow integration and usability.

## Key findings

- Only 11.3% of trainees who recognized the sepsis alert had used the advisor tool.
- Alert fatigue and limited clinical utility were the top concerns among users.
- Trainees suggested non-interruptive alerts and workflow-aligned tools to improve engagement.

## Abstract

Early recognition and treatment of sepsis is critical to reducing mortality. Many hospitals have adopted Electronic Health Record (EHR)-based clinical decision support (CDS) tools to assist with timely identification and management of sepsis. However, over-alerting and poor integration into clinician workflow may limit their effectiveness.

To assess medical trainees’ awareness, usage, and perceptions of two EHR-integrated sepsis CDS tools—a sepsis alert and sepsis advisor—at a tertiary academic hospital to inform future system-wide sepsis workflows redesign.

A cross-sectional survey was distributed to 673 residents and fellows at Penn State Milton S. Hershey Medical Center in March 2022. The anonymous survey explored familiarity with sepsis tools, usage patterns, perceived usefulness, and suggestions for improvement.

Ninety-two trainees responded (13.7% response rate). Among 80 who recognized the sepsis alert, only 11.3% had used the advisor. Alert fatigue (62.5%) and limited clinical utility (46.3%) were major concerns. Only 16.3% felt both tools were helpful, while 52.5% found them unhelpful. Lack of awareness about advisor features and alert suppression mechanisms further limited engagement. Respondents who found the tools unhelpful recommended a non-interruptive, asynchronous alert format, a “patient already being treated” option, and redesigning the advisor into a streamlined, workflow-aligned tool.

While sepsis CDS tools offer potential for improving care, their impact is limited by knowledge gaps, usability issues, alert fatigue, and poor clinical fit. Addressing these issues through better design, education, and workflow integration—guided by frontline user input—should be a core strategy in CDS quality improvement efforts.