Early Identification of Patients With Acute Decompensated Heart Failure

Saul Blecker, David Sontag, Leora I. Horwitz, Gilad Kuperman, Hannah Park, Alex Reyentovich, Stuart D. Katz

Research output: Contribution to journalArticle

Abstract

Background: Interventions to reduce readmissions after acute heart failure hospitalization require early identification of patients. The purpose of this study was to develop and test accuracies of various approaches to identify patients with acute decompensated heart failure (ADHF) with the use of data derived from the electronic health record. Methods and Results: We included 37,229 hospitalizations of adult patients at a single hospital during 2013-2015. We developed 4 algorithms to identify hospitalization with a principal discharge diagnosis of ADHF: 1) presence of 1 of 3 clinical characteristics, 2) logistic regression of 31 structured data elements, 3) machine learning with unstructured data, and 4) machine learning with the use of both structured and unstructured data. In data validation, algorithm 1 had a sensitivity of 0.98 and positive predictive value (PPV) of 0.14 for ADHF. Algorithm 2 had an area under the receiver operating characteristic curve (AUC) of 0.96, and both machine learning algorithms had AUCs of 0.99. Based on a brief survey of 3 providers who perform chart review for ADHF, we estimated that providers spent 8.6 minutes per chart review; using this this parameter, we estimated that providers would spend 61.4, 57.3, 28.7, and 25.3 minutes on secondary chart review for each case of ADHF if initial screening were done with algorithms 1, 2, 3, and 4, respectively. Conclusions: Machine learning algorithms with unstructured notes had the best performance for identification of ADHF and can improve provider efficiency for delivery of quality improvement interventions.

Original languageEnglish (US)
JournalJournal of Cardiac Failure
DOIs
StateAccepted/In press - 2017

Fingerprint

Heart Failure
Hospitalization
Area Under Curve
Electronic Health Records
Quality Improvement
ROC Curve
Logistic Models
Machine Learning

Keywords

  • Electronic health record
  • Heart failure
  • Hospitalization
  • Phenotype

ASJC Scopus subject areas

  • Cardiology and Cardiovascular Medicine

Cite this

Blecker, S., Sontag, D., Horwitz, L. I., Kuperman, G., Park, H., Reyentovich, A., & Katz, S. D. (Accepted/In press). Early Identification of Patients With Acute Decompensated Heart Failure. Journal of Cardiac Failure. https://doi.org/10.1016/j.cardfail.2017.08.458

Early Identification of Patients With Acute Decompensated Heart Failure. / Blecker, Saul; Sontag, David; Horwitz, Leora I.; Kuperman, Gilad; Park, Hannah; Reyentovich, Alex; Katz, Stuart D.

In: Journal of Cardiac Failure, 2017.

Research output: Contribution to journalArticle

Blecker, S, Sontag, D, Horwitz, LI, Kuperman, G, Park, H, Reyentovich, A & Katz, SD 2017, 'Early Identification of Patients With Acute Decompensated Heart Failure', Journal of Cardiac Failure. https://doi.org/10.1016/j.cardfail.2017.08.458
Blecker, Saul ; Sontag, David ; Horwitz, Leora I. ; Kuperman, Gilad ; Park, Hannah ; Reyentovich, Alex ; Katz, Stuart D. / Early Identification of Patients With Acute Decompensated Heart Failure. In: Journal of Cardiac Failure. 2017.
@article{066743e3ad1941eb937b5df60266e14c,
title = "Early Identification of Patients With Acute Decompensated Heart Failure",
abstract = "Background: Interventions to reduce readmissions after acute heart failure hospitalization require early identification of patients. The purpose of this study was to develop and test accuracies of various approaches to identify patients with acute decompensated heart failure (ADHF) with the use of data derived from the electronic health record. Methods and Results: We included 37,229 hospitalizations of adult patients at a single hospital during 2013-2015. We developed 4 algorithms to identify hospitalization with a principal discharge diagnosis of ADHF: 1) presence of 1 of 3 clinical characteristics, 2) logistic regression of 31 structured data elements, 3) machine learning with unstructured data, and 4) machine learning with the use of both structured and unstructured data. In data validation, algorithm 1 had a sensitivity of 0.98 and positive predictive value (PPV) of 0.14 for ADHF. Algorithm 2 had an area under the receiver operating characteristic curve (AUC) of 0.96, and both machine learning algorithms had AUCs of 0.99. Based on a brief survey of 3 providers who perform chart review for ADHF, we estimated that providers spent 8.6 minutes per chart review; using this this parameter, we estimated that providers would spend 61.4, 57.3, 28.7, and 25.3 minutes on secondary chart review for each case of ADHF if initial screening were done with algorithms 1, 2, 3, and 4, respectively. Conclusions: Machine learning algorithms with unstructured notes had the best performance for identification of ADHF and can improve provider efficiency for delivery of quality improvement interventions.",
keywords = "Electronic health record, Heart failure, Hospitalization, Phenotype",
author = "Saul Blecker and David Sontag and Horwitz, {Leora I.} and Gilad Kuperman and Hannah Park and Alex Reyentovich and Katz, {Stuart D.}",
year = "2017",
doi = "10.1016/j.cardfail.2017.08.458",
language = "English (US)",
journal = "Journal of Cardiac Failure",
issn = "1071-9164",
publisher = "Churchill Livingstone",

}

TY - JOUR

T1 - Early Identification of Patients With Acute Decompensated Heart Failure

AU - Blecker, Saul

AU - Sontag, David

AU - Horwitz, Leora I.

AU - Kuperman, Gilad

AU - Park, Hannah

AU - Reyentovich, Alex

AU - Katz, Stuart D.

PY - 2017

Y1 - 2017

N2 - Background: Interventions to reduce readmissions after acute heart failure hospitalization require early identification of patients. The purpose of this study was to develop and test accuracies of various approaches to identify patients with acute decompensated heart failure (ADHF) with the use of data derived from the electronic health record. Methods and Results: We included 37,229 hospitalizations of adult patients at a single hospital during 2013-2015. We developed 4 algorithms to identify hospitalization with a principal discharge diagnosis of ADHF: 1) presence of 1 of 3 clinical characteristics, 2) logistic regression of 31 structured data elements, 3) machine learning with unstructured data, and 4) machine learning with the use of both structured and unstructured data. In data validation, algorithm 1 had a sensitivity of 0.98 and positive predictive value (PPV) of 0.14 for ADHF. Algorithm 2 had an area under the receiver operating characteristic curve (AUC) of 0.96, and both machine learning algorithms had AUCs of 0.99. Based on a brief survey of 3 providers who perform chart review for ADHF, we estimated that providers spent 8.6 minutes per chart review; using this this parameter, we estimated that providers would spend 61.4, 57.3, 28.7, and 25.3 minutes on secondary chart review for each case of ADHF if initial screening were done with algorithms 1, 2, 3, and 4, respectively. Conclusions: Machine learning algorithms with unstructured notes had the best performance for identification of ADHF and can improve provider efficiency for delivery of quality improvement interventions.

AB - Background: Interventions to reduce readmissions after acute heart failure hospitalization require early identification of patients. The purpose of this study was to develop and test accuracies of various approaches to identify patients with acute decompensated heart failure (ADHF) with the use of data derived from the electronic health record. Methods and Results: We included 37,229 hospitalizations of adult patients at a single hospital during 2013-2015. We developed 4 algorithms to identify hospitalization with a principal discharge diagnosis of ADHF: 1) presence of 1 of 3 clinical characteristics, 2) logistic regression of 31 structured data elements, 3) machine learning with unstructured data, and 4) machine learning with the use of both structured and unstructured data. In data validation, algorithm 1 had a sensitivity of 0.98 and positive predictive value (PPV) of 0.14 for ADHF. Algorithm 2 had an area under the receiver operating characteristic curve (AUC) of 0.96, and both machine learning algorithms had AUCs of 0.99. Based on a brief survey of 3 providers who perform chart review for ADHF, we estimated that providers spent 8.6 minutes per chart review; using this this parameter, we estimated that providers would spend 61.4, 57.3, 28.7, and 25.3 minutes on secondary chart review for each case of ADHF if initial screening were done with algorithms 1, 2, 3, and 4, respectively. Conclusions: Machine learning algorithms with unstructured notes had the best performance for identification of ADHF and can improve provider efficiency for delivery of quality improvement interventions.

KW - Electronic health record

KW - Heart failure

KW - Hospitalization

KW - Phenotype

UR - http://www.scopus.com/inward/record.url?scp=85030781558&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85030781558&partnerID=8YFLogxK

U2 - 10.1016/j.cardfail.2017.08.458

DO - 10.1016/j.cardfail.2017.08.458

M3 - Article

JO - Journal of Cardiac Failure

JF - Journal of Cardiac Failure

SN - 1071-9164

ER -