# A Heart Rate Variability–Derived Decision Support Tool for Prognostication in Emergency Department Patients With Suspected Infection

## TL;DR

A new tool called Sepsis Advisor uses heart rate variability and lab data to predict deterioration in emergency patients with suspected infection.

## Contribution

The study demonstrates the feasibility of deploying a heart rate variability-based decision support tool in emergency departments.

## Key findings

- 71 patients were enrolled, with 65 having sufficient heart rate variability data for model generation.
- User feedback led to four improved versions of the Sepsis Advisor tool.
- The tool was perceived as potentially improving communication and care in emergency departments.

## Abstract

Prediction of future deterioration in emergency department patients with infection is difficult, and existing prognostic tools are inaccurate. We evaluated the feasibility of deployment of a clinical decision support tool, Sepsis Advisor, which utilizes heart rate variability and laboratory values to predict future deterioration in emergency department patients with treated infection.

This study was an observational, prospective, Pilot Phase 1 feasibility implementation study involving two sites within a single academic health sciences centre. Then, 71 patients were enrolled, all with suspected/treated infection and systemic inflammatory response. Patients underwent 30 min of electrocardiograph recording. The generated predictive model and Sepsis Advisor report were shown to physicians observationally, > 48 h after clinical encounter, while assessing perceived usability, value, barriers and drivers with using the tool through interviews with nurses and physicians.

Of the 71 patients enrolled, 65 (92%) had adequate duration of heart rate variability measurements to generate a predictive model (average recording: 25 ± 7 min); 100% had clinical data entry. Creatinine, lactate, and INR were drawn 97%, 56%, and 28% of the time and were incorporated into predictive models. Physician and nurse reported drivers for use included potential to facilitate communication, improve care, and ease of integration. Barriers included the need to understand and interpret results from the tool, time constraints, changing routines, and gaining buy‐in. User‐centered feedback informed four improved versions of the tool.

Observational deployment of a heart rate variability–based clinical decision support tool within the emergency department is feasible and perceived to have the potential to improve care.