Title

Predictors of Re-Hospitalization for Home Healthcare Patients

Date of Award

2008-04-01

Degree Name

PhD Nursing

Dissertation Committee

Cynthia D. Connelly, PhD, RN, FAAN; Jane Georges, PhD, RN; Linda D. Urden, DNSc, RN, CNA-BC, FAAN

Keywords

home healthcare, nursing, patients, re-hospitalization

Abstract

The overall purpose of this study was to examine the predictive capability of OASIS admission data for acute care re-hospitalization of home healthcare patients. Secondary data analysis using logistic regression was conducted on retrospective data from OASIS collected during the time period of July 1, 2006 to June 30, 2007. This study was conducted in a Medicare certified Home Health organization that is part of the largest public health system in California. The sample of 1802 patients with complete episodes of care was derived from a data set of 5,523 patients. All patients were included in the analysis and logistic regression model and the disease specific independent variables included patients with a primary or secondary diagnosis of diabetes and an open skin lesion or wound. The OASIS variables examined in the logistic regression model that showed significance as predictors of acute care re-hospitalization included a diagnosis of diabetes, overall prognosis, rehabilitation prognosis, existing dyspnea, existing urine and bowel incontinence, impairment in currently dressing the upper body and the ability to take own oral medications. These findings apply to all patients in the OASIS database as the logistic regression model included all patients. An interesting finding was that the presence of a lesion or open wound was not significant as a predictor of acute care re-hospitalization. Also of interest was the occurrence of re-hospitalization of 15% that is lower than that reported in the literature as well as the occurrence of diabetes of 14% which is lower than the population in the community. The study methodology related to the backwards method of logistic regression modeling was useful in being able to examine a large number of variables and their relationship to a dichotomous dependent variable. Since this design and method has not been described in the literature prior to this study it has interesting implications for future research using OASIS.

Document Type

Dissertation: Open Access

Department

Nursing

This document is currently not available here.

Share

COinS