STRATIFY Research Study

Emergency department heart failure

STRATIFY, Improving Heart Failure Risk Stratification in the ED, is an externally validated predictive model with excellent negative predictive value for identifying patients with acute heart failure at low risk of 30-day events.

Watch the video to hear Principal Investigator Dr. Sunil Kripalani of the Department of Medicine discuss the STRATIFY study.

Acute heart failure prediction tool

The STRATIFY Acute Heart Failure prediction tool was developed to identify emergency department (ED) patients with acute heart failure (AHF) at low risk of 30-day death or serious complications. 

These patients are potential candidates for safe ED discharge and outpatient management. The 30-day adverse outcome in STRATIFY focused on events of most interest to physicians when considering ED discharge, and included death, cardiopulmonary resuscitation, mechanical cardiac support, intubation or mechanical ventilation, emergent dialysis, percutaneous coronary intervention, coronary artery bypass grafting, and acute coronary syndrome. 

A total of 13 risk factors available within three hours of ED presentation were selected for the tool. Eight are collected at triage, one from EKG and four from blood testing. For identifying low-risk patients safe for discharge, the most important performance measure is the negative predictive value (NPV), which represents the proportion of patients with predicted risks less than a certain threshold that are free of 30-day adverse events. 

At the risk threshold of 3% and 5%, the NPVs were 100% and 96% in the derivation study, suggesting STRATIFY is useful at this end of the risk continuum and has utility for facilitating safe ED discharge. For STRATIFY implementation, we use threshold corresponding to 10% and 30% risk percentile for risk stratification.

Download the STRATIFY flyer for full description.

Evidence

  • The Stratification Decision Tool: Identification of Emergency Department Patients With Acute Heart Failure at Low Risk for 30-Day Adverse Events (PubMed article)
  • STRATIFY DesignRisk stratification in acute heart failure: rationale and design of the STRATIFY and DECIDE studies (PubMed article)

 

Stratify Components

  • Age Body Mass Index (BMI)
  • Diastolic Blood Pressure (DBP)
  • Oxygen Saturation
  • Respiratory Rate
  • Outpatient supplemental oxygen
  • On dialysis 
  • Use of outpatient ACEI
  • QRS duration < 120
  • Sodium
  • Troponin
  • BNP
  • Blood Urea Nitrogen (BUN)

Patient Education Materials

 

What is STRATIFY? 

It’s a tool we use to help you and your doctor make the best decisions about your care.

 

What does STRATIFY Do?

Calculates score 

STRATIFY calculates a score using data from your chart, like vital signs, medical history, and labs.

Determines Risk

This score helps your doctor see if your risk of complications in the next 30 days is low or high.

Care Plan

The score helps us decide if you need to be in the hospital or if it’s safe for you to get care at home.

If your doctor decides to do something different than what  STRATIFY suggests, they will talk with you about it.

Risk Stratification Figure in the Electronic Health Record (EHR):

VUMC team puts tool to reduce heart failure admissions to test 

 

STRATIFY principal investigators, from left, Dandan Liu, PhD, Alan Storrow, MD, and Sunil Kripalani, MD, MSc, are testing the real-world implementation of a risk stratification tool to avoid unnecessary hospital admissions of patients diagnosed with acute heart failure in the emergency department.

Developers and Principal Investigators

Alan Storrow, MD

Associate Professor of Emergency Medicine

Associate Director of the EM Research Division

Dandan Liu, MD

Associate Professor of Biostatistics

VUMC Department of Biostatistics

Executive Director, Vanderbilt Biostatistics Data Coordinating Center (VBDCC)

Sunil Kripalani, MD, MSc

Professor of Medicine

VUMC Department of Medicine, Division of General Internal Medicine and Public Health 

Director, VUMC Center for Health Services Research

Clinicians 

  • Adam Wright, Department of Biomedical Informatics
  • Vanderbilt Clinical Informatics Core (VCLIC)
  • Laurie Novak, Center for Health Services Research
  • Casey Distaso

Biostatistics

  • Cathy Jenkins
  • Tianyi Sun

Informatics/ Health IT

  • Tim Coffman, Health IT Product Development, Application Development, Architect and Technical Lead
  • Marc Beller, Health IT Product Development, Research Portfolio Management
  • Simeon Hearring, Health IT Product Development
  • Janos Mathos, Health IT Product Development, Knowledge Engineering
  • Asli Weitkamp, Health IT Product Development, Strategic CDS & Knowledge Engineering
  • Allison McCoy, Vanderbilt Clinical Informatics Core (VCLIC)
  • Dan Albert, Health IT Product Development, Application Development, Associate Director
  • Allyson Hobbie, Health IT Product Development, Principal Product Manager
  • Vikas Jain, Health IT Product Development, Senior IT Project Manager
  • Doug Wallace, Website 

User design

  • Shiloh Anders, Center for Research and Innovation in Systems Safety (CRISS)
  • Carrie Reale
  • Russ Beebe
  • Janelle Faiman
  • Janos Mathe

Implementation team

  • Deonni Stolldorf, School of Nursing
  • Laurie Novak, Center for Health Services Research
  • Anna Sachs, Project Manager, STRATIFY grant
  • Isaac Schlotterbeck, Senior Health Services Research Analyst, Center for Health Services Research
  • Rachel Hilton 

Project Management and Support

  • Anna Sachs, Project Manager, STRATIFY grant
  • Zorah Taplin, Health IT Product Development, Research Portfolio Management
  • Vikas Jain, Health IT Product Development, Senior IT Project Manager
  • Marc Beller, Health IT Product Development, Research Portfolio Management
  • Isaac Schlotterbeck, Senior Health Services Research Analyst, Center for Health Services Research

Oregon Health Sciences University Team, Portland, Oregon 

  • Bory Kea, Site PI
  • Ben Orwoll, Co-PI Informatics
  • Joy Kim, Study Coordinator 

Henry Ford Health System, Detroit

  • Joe Miller, Site PI
  • Sandeep Soman, Informatics