Secondary Logo

Institutional members access full text with Ovid®

Mortality Risk in Homebound Older Adults Predicted From Routinely Collected Nursing Data

Sullivan, Suzanne S.; Hewner, Sharon; Chandola, Varun; Westra, Bonnie L.

doi: 10.1097/NNR.0000000000000328

Background Newer analytic approaches for developing predictive models provide a method of creating decision support to translate findings into practice.

Objectives The aim of this study was to develop and validate a clinically interpretable predictive model for 12-month mortality risk among community-dwelling older adults. This is done by using routinely collected nursing assessment data to aide homecare nurses in identifying older adults who are at risk for decline, providing an opportunity to develop care plans that support patient and family goals for care.

Methods A retrospective secondary analysis of Medicare and Medicaid data of 635,590 Outcome and Assessment Information Set (OASIS-C) start-of-care assessments from January 1, 2012, to December 31, 2012, was linked to the Master Beneficiary Summary File (2012–2013) for date of death. The decision tree was benchmarked against gold standards for predictive modeling, logistic regression, and artificial neural network (ANN). The models underwent k-fold cross-validation and were compared using area under the curve (AUC) and other data science metrics, including Matthews correlation coefficient (MCC).

Results Decision tree variables associated with 12-month mortality risk included OASIS items: age, (M1034) overall status, (M1800–M1890) activities of daily living total score, cancer, frailty, (M1410) oxygen, and (M2020) oral medication management. The final models had good discrimination: decision tree, AUC = .71, 95% confidence interval (CI) [.705, .712], sensitivity = .73, specificity = .58, MCC = .31; ANN, AUC = .74, 95% CI [.74, .74], sensitivity = .68, specificity = .68, MCC = .35; and logistic regression, AUC = .74, 95% CI [.735, .742], sensitivity = .64, specificity = .70, MCC = .35.

Discussion The AUC and 95% CI for the decision tree are slightly less accurate than logistic regression and ANN; however, the decision tree was more accurate in detecting mortality. The OASIS data set was useful to predict 12-month mortality risk. The decision tree is an interpretable predictive model developed from routinely collected nursing data that may be incorporated into a decision support tool to identify older adults at risk for death.

Suzanne S. Sullivan, PhD, MBA, RN, CHPN, is Assistant Professor, University at Buffalo, New York. This study is part of her PhD dissertation research under the supervision of Sharon Hewner, PhD, RN, FAAN, at the University at Buffalo.

Sharon Hewner, PhD, RN, FAAN, is Associate Professor, School of Nursing, University at Buffalo, New York.

Varun Chandola, PhD, is Assistant Professor, Department of Computer Science and Engineering, University at Buffalo, New York. She was an official consultant and served on the dissertation committee of Suzanne Sullivan.

Bonnie L. Westra, PhD, RN, FAAN, FACMI, is Associate Professor, School of Nursing, University of Minnesota, Minneapolis. She was an official sponsor and collaborator on this NINR F31-funded study.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (

Accepted for publication September 21, 2018.

All authors contributed to the writing and editing of this manuscript.

The authors thank Dr. Suzanne Dickerson, DNS, RN, Chair, Presidential Review Board, Professor and Chair of Bio Behavioral Health and Clinical Science, School of Nursing, University at Buffalo; Dr. Mary Ann Meeker, DNS, RN, Assistant Dean, PhD Program, School of Nursing, University at Buffalo; Peter Elkin, MD, MACP, FACMI, FNYAM, Professor and Chair, Department of Biomedical Informatics and Professor of Internal Medicine, Department of Biomedical Informatics, University at Buffalo; Jiwei Zhao, PhD, Assistant Professor, Department of Biostatistics, University at Buffalo; Bill Wu, PhD, Associate Professor (retired), School of Nursing, University at Buffalo; Davina Porock, PhD, RN, FAAN, Assistant Dean, Research and Innovation, Sheffield Hallam University, United Kingdom; the Research Foundation, University at Buffalo; the Information Technology Department, University at Buffalo; and Diane Dempsey, MS, CRA, Grants Manager, Center for Nursing Research and School Nursing, University at Buffalo, for their sponsorship, advice, and guidance on this project.

The authors acknowledge research reported in this publication was supported by the National Institute of Nursing Research of the National Institutes of Health under Award F31NR016394 (Sullivan, PI). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

This study received institutional review board ethics committee approval from the University at Buffalo Institutional Review Board. The study was deemed to be not human subject research.

The authors have no conflicts of interest to report.

Corresponding author: Suzanne S. Sullivan, PhD, MBA, RN, CHPN, School of Nursing, University at Buffalo, 3435 Main Street, 201A Wende Hall, Buffalo, NY 14214-8013 (e-mail:

Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved