# Study protocol for the design, implementation, and evaluation of the STRATIFY clinical decision support tool for emergency department disposition of patients with heart failure

## TL;DR

This study describes the development and evaluation of a clinical decision support tool called STRATIFY to help emergency department clinicians decide whether patients with heart failure should be admitted or discharged.

## Contribution

The study introduces a stakeholder-informed, iterative design approach for implementing a tailored clinical decision support system in emergency departments.

## Key findings

- STRATIFY is a validated risk prediction model for identifying low-risk heart failure patients suitable for discharge.
- A multi-level implementation strategy was developed to address data integration challenges in real-time predictive modeling.
- The study will evaluate the effectiveness of STRATIFY across seven emergency departments using an interrupted time-series design.

## Abstract

In the emergency department (ED), clinicians often make challenging, high-pressure decisions within a short time frame. Clinical decision support (CDS) tools integrated into the electronic health record can provide evidence-based support. Yet, numerous implementation barriers limit the broad use of such tools in ED settings. CDS tools could be particularly helpful for patients presenting to the ED with an acute exacerbation of heart failure (AHF), a common and costly medical condition for which patients are typically admitted to the hospital. We developed and implemented STRATIFY, a validated risk prediction model that effectively identifies AHF patients at low risk of 30-day adverse events who could potentially be discharged home from the ED.

This article describes a multi-center study to 1) develop a stakeholder-informed CDS-based implementation process for STRATIFY, 2) use novel statistical methods to overcome data integration challenges to the real-world implementation of predictive models in the ED, and 3) evaluate the implementation and effectiveness of the newly developed STRATIFY CDS at 7 EDs to guide decision-making to admit or discharge patients with AHF. The study’s multi-level implementation strategy is tailored to each site and informed by site assessments (including pre-visit surveys, on-site ED visits, and virtual interviews), small group discussions with patients and caregivers, and iterative user-centered design to develop and refine the STRATIFY CDS. Overcoming data challenges for real-time predictive models involves accommodating missing risk factor data while still generating valid predictions of risk. In the evaluation of effectiveness, we will evaluate ED disposition (admit/discharge) for patients with AHF, as well as potential adverse outcomes, using an interrupted time-series design at 7 participating EDs. The study will evaluate implementation outcomes ranging from acceptability to sustainability using electronic health record data and surveys of clinicians and patients.

This study uses a stakeholder-informed, iterative design approach to develop a tailored CDS-based process supported by a multi-level implementation strategy to incorporate a validated risk prediction tool into the care of patients with AHF in the ED. The study will advance methods to close the evidence-practice gap in the care of emergency department patients.

The online version contains supplementary material available at 10.1186/s43058-025-00779-w.