Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Planning care with the patient in the room: a patient-focused approach to reducing heart failure readmissions
(USC Thesis Other)
Planning care with the patient in the room: a patient-focused approach to reducing heart failure readmissions
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
PLANNING CARE WITH THE PATIENT IN THE ROOM:
A PATIENT-FOCUSED APPROACH TO REDUCING HEART FAILURE
READMISSIONS
by
Lynn Marie Garofalo
__________________________________________________
A Dissertation Presented to the
FACULTY OF THE USC SCHOOL OF POLICY, PLANNING,
AND DEVELOPMENT
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF POLICY, PLANNING, AND DEVELOPMENT
May 2013
Copyright 2013 Lynn Marie Garofalo
ii
Epigraph
”It's far more important to know what person the disease has than what disease the person
has.”
Hippocrates
iii
Acknowledgements
I began this journey eight years ago as a single professional in the health care field
interested in further exploring research and contributing to my field. Now married with
two children and a full-time job, friends and co-workers occasionally refer to me as a
“wonder woman.” While I appreciate the praise, the truth is, it is a wonder I have reached
my destination. I owe a tremendous amount of gratitude to four groups of individuals.
First, my doctoral advisor, Robert Myrtle D.P.A. and advisory committee members: Paul
Adler Ph.D., David Belson Ph.D., Ted Eytan M.D., Lisa Schilling R.N. M.P.H., and
Shinyi Wu Ph.D. for guiding and challenging me to find a narrow topic that would
interest me and contribute to health care improvement practice.
Second, the many colleagues who gave me the opportunity to learn about heart failure
practices and the readmissions challenge, guided me through the research process, and
provided other support including Dennis Deas, Ray Hahn M.H.A., Ana Jackson Ph.D.,
Elizabeth Le, Sandra Koyama M.D., Rudy Marilla M.B.A., Estee Neuwirth Ph.D., Margi
Spies, Maria Taitano MD, Phillip Tuso MD, and Heather Watson.
Third, Claude Rubinson Ph.D. from the University of Houston-Downtown who kindly
supported me through the process of learning and using the fs/QCA software.
iv
Finally, my friends who endured not seeing me for months at a time, my parents and
sister who cheered me on from afar, my husband who integrated Saturday study day into
our family routine, and my children for understanding that I had to do my homework too.
Doctoral Committee
____________________________________________________
Robert C. Myrtle, D.P.A.
Professor, Public Administration and Director, Executive Master of Leadership
Price School of Policy, University of Southern California
Los Angeles, California
Paul Adler, Ph.D.
Professor and Harold Quinton Chair in Business Policy and Professor of Management
and Organization
Marshall School of Business, University of Southern California
Los Angeles, California
David Belson, Ph.D.
Adjunct Professor
Industrial and Systems Engineering, University of Southern California
Los Angeles, California
Ted Eytan, M.D.
Director
The Permanente Federation. LLC
Kaiser Permanente
Washington D.C.
Lisa Schilling, R.N., M.P.H.
Vice President
Healthcare Performance Improvement and Execution Strategy
Kaiser Permanente
Oakland, California
Shinyi Wu, Ph.D.
Assistant Professor
Industrial and Systems Engineering, University of Southern California
Los Angeles, California
v
Table of Contents
Epigraph ii
Acknowledgements iii
List of Tables x
List of Figures xi
Abbreviations xii
Abstract xiii
Chapter One: Introduction 1
Introduction 1
Literature Review 2
Study Overview and Methodology 4
Results 6
Discussion 8
Conclusions 9
Chapter Two: Literature Review 11
Introduction 11
Patient and Provider Characteristics Positively Correlated to
Readmissions 12
Interventions that May Reduce General Readmissions 20
Interventions that May Reduce Heart Failure Readmissions 22
Conclusion 38
Chapter Three: Study Overview and Methodology 40
Introduction 40
Heart Failure Program 41
Research 45
Data for fsQCA 49
Study Approach 53
Limitations 57
Chapter Four: Study Results 59
Introduction 59
Results of Case Studies 60
Results of Fuzzy Set Qualitative Comparative Analysis 68
Conclusion 92
vi
Chapter Five: Discussion and Extensions 93
Introduction 93
Patient-Centered Care 94
A Patient-Centered Approach to Reducing Heart Failure
Readmissions 97
Conclusion 107
Chapter Six: Conclusion 108
Program Leadership and Oversight 109
Organizational Structure 110
Process Design 111
Conclusion 112
Bibliography 113
Appendix 126
Model 1: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, in_hfprogram, older_age, white_race,
discharged_home, male_gender, english_languag) 126
Model 2: ~readmit_in_30 = f(homehealth_in_48) 129
Model 3: ~readmit_in_30 = f(rtmr_in_48hour) 129
Model 4: ~readmit_in_30 = f(md_in_7days) 130
Model 5: ~readmit_in_30 = f(call_in_7days) 132
Model 6: ~readmit_in_30 = f(in_hfprogram) 133
Model 7: ~readmit_in_30 = f(older_age) 134
Model 8: ~readmit_in_30 = f(white_race) 135
Model 9: ~readmit_in_30 = f(discharged_home) 135
Model 10: ~readmit_in_30 = f(male_gender) 136
Model 11: ~readmit_in_30 = f(english_languag) 137
Model 12: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour) 138
Model 13: ~readmit_in_30 = f(homehealth_in_4, md_in_7days) 139
Model 14: ~readmit_in_30 = f(homehealth_in_4, call_in_7days) 141
Model 15: ~readmit_in_30 = f(md_in_7days, call_in_7days) 142
Model 16: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days) 144
Model 17: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
call_in_7days) 145
Model 18: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days) 148
Model 19: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age) 152
Model 20: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, white_race) 156
vii
Model 21: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, discharged_home) 160
Model 22: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, male_gender) 164
Model 23: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, english_languag) 167
Model 24: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race) 170
Model 25: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, discharged_home) 173
Model 26: 177
Model 27: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, english_languag) 181
Model 28: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, white_race, discharged_home) 185
Model 29: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, white_race, male_gender) 189
Model 30: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, white_race, english_languag) 193
Model 31: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, discharged_home) 196
Model 32~: f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, discharged_home, english_languag) 200
Model 33: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, male_gender, english_languag) 203
Model 34: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race,
discharged_home) 206
Model 35~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race, male_gender) 209
Model 36~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race, english_languag) 213
Model 37~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, discharged_home,
male_gender) 216
Model 38~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, discharged_home,
english_languag) 220
Model 39~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, male_gender,
english_languag) 223
Model 40~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, white_race, discharged_home,
male_gender) 226
viii
Model 41~ ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, white_race, discharged_home,
english_languag) 229
Model 42: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, white_race, male_gender,
english_languag) 233
Model 43: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, discharged_home, male_gender,
english_languag) 237
Model 44: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race,
discharged_home, male_gender) 241
Model 45: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race,
discharged_home, english_languag) 245
Model 46: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race, male_gender,
english_languag) 248
Model 47: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, discharged_home,
male_gender, english_languag) 252
Model 48: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, white_race, discharged_home,
male_gender, english_languag) 254
Model 49: ~readmit_in_30 = f(homehealth_in_4, older_age,
white_race, discharged_home, male_gender, english_languag) 258
Model 50: ~readmit_in_30 = f(rtmr_in_48, older_age, white_race,
discharged_home, male_gender, english_languag) 261
Model 51: ~readmit_in_30 = f(md_in_7days, older_age, white_race,
discharged_home, male_gender, english_languag) 263
Model 52~: ~readmit_in_30 = f(call_in_7days, older_age, white_race,
discharged_home, male_gender, english_languag) 266
Model 53: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race,
discharged_home, male_gender, english_languag) 269
Model 54: ~readmit_in_30 = f(homehealth_in_48, rtmr_in_48,
md_in_7days, call_in_7days, older_age, white_race,
discharged_home, male_gender) – non-English speaking patients only
(n=833) 272
Model 55: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race, male_gender,
english_languag) – patients discharged home or to another
community setting (n=8269) 276
Model 56: Hospital A 279
ix
Model 56: Hospital B 283
Model 56: Hospital C 288
Model 56: Hospital D 293
Model 56: Hospital E 297
Model 56: Hospital F 302
Model 56: Hospital G 306
Model 56: Hospital H 307
Model 56: Hospital I 311
Model 56: Hospital J 316
Model 56: Hospital K 319
Model 54: Hospital L 323
x
List of Tables
Table 1: Personal Characteristics Found in Literature ...................................................... 12
Table 2: Review of Heart Failure Interventions in the Literature ..................................... 24
Table 3: Reliability of Key Interventions for All Patients in Heart Failure Program ....... 45
Table 4: Interventions Delivered and Patient Characteristics for 8,915 Discharges ........ 52
xi
List of Figures
Figure 1: Interventions Studied to Reduce Hospital Readmissions .................................. 21
Figure 2: Frequency Distribution of Number of 30-Day Readmissions per Patient......... 53
Figure 3: Frequency Distribution of Heart Failure Patient Ages ...................................... 56
Figure 4: Post Discharge Appointment Scheduling Process ............................................ 63
Figure 5: Qualitative Comparative Analysis Models........................................................ 73
Figure 6: Results of Models 2 – 11 ................................................................................... 75
Figure 7: Results of Models 12 – 18 ................................................................................. 78
Figure 8: Results of Models 19 – 48 ................................................................................. 80
Figure 9: Results of Models 49 – 53 ................................................................................. 84
Figure 10: Results of Models 54 – 55 ............................................................................... 87
Figure 11: Results of Model 56: Stratified by Hospital .................................................... 90
Figure 12: Hansen's Readmission Reduction Intervention Framework ............................ 98
Figure 13: Garofalo’s Patient-Centered Framework for Reducing Heart Failure
Readmissions .................................................................................................................... 99
xii
Abbreviations
Abbreviation Definition
~ Negated. Represents the absence of a causal condition in fsQCA
D/C Discharge
EHR Electronic Health Record
fs/QCA Fuzzy Set / Qualitative Comparative Analysis software
fsQCA Fuzzy-set Qualitative Comparative Analysis
HF Heart Failure
NYHA New York Heart Association
xiii
Abstract
If US health care organizations did not already deem readmissions reduction an important
undertaking, the Health Care Accountability and Affordability Act has made it so. Over
two thousand hospitals are already being penalized for having higher than average
readmission rates for acute myocardial infarction, pneumonia, and heart failure. I
conducted a two-part study across a southern California health system to understand how
health care providers can most effectively reduce heart failure readmissions with
constrained finances and found a static approach that provides every patient with the
same intervention at every discharge is neither optimal for patients nor an effective use of
resources. I propose a patient-centered framework for reducing heart failure readmissions
that begins with understanding each patient and their family, and collaborating with them
to determine the optimal care plan based on multiple potential interventions. Furthermore
I discuss the importance delivering these interventions with fidelity to achieve the best
patient outcomes, and provide guidance for designing an effective heart failure
readmission reduction program with engaged leaders, program oversight that includes
leaders and frontline staff, and well designed processes that focus on the patient and
encourage continuous improvement. I believe as we get to know our patients and plan
care with them in the room, we too will learn more about how we can provide the most
effective care.
1
Chapter One: Introduction
Introduction
The nation’s aging baby boomers, escalating health care costs, and health care reform
legislation are forcing US health care providers to improve cost structures while
increasing quality outcomes. Although it is debated whether a person’s return to the
hospital, or readmission, is a direct indicator of poor quality (van Walraven, Bennett,
Jennings, Austin, & Forster, 2011), it is not debated that hospital readmissions are
expensive. Nearly 20% of Medicare beneficiaries are readmitted within thirty days
(Jencks S. W., 2009), which is estimated to cost Medicare and Medicaid $17 billion
annually (Bakhtiari, 2010). To curb this trend, the Affordable Care Act began reducing
hospitals’ DRG reimbursement in October 2012 if readmissions exceeded an expected
ratio. This policy initially applies to three specific conditions: acute myocardial
infarction, pneumonia, and heart failure (Selected Medicare Hospital Quality Provisions
Under the ACA), but will be extended in 2013.
I conducted a process assessment of a heart failure transitional care program across a
southern California hospital system and participated in chart reviews and patient
interviews to understand opportunities for better preventing thirty day readmissions. I
observed frustrated, passionate frontline staff working through checklists of prescribed
tasks, trying to ensure every patient received the same, standard interventions designed to
2
transition them safely from the hospital to home to avoid a readmission. During this
study, care providers shared examples of patients who were frequently admitted to the
hospital and would likely return again despite diligently providing the standard
interventions. From this, I developed a question: with limited financial resources and
increasing pressure to reduce avoidable readmissions, how can health care providers most
effectively and efficiently care for people with heart failure to reduce readmissions?
I examine strategies for reducing heart failure readmissions in four parts. First, I review
the literature to understand which patient and provider characteristics are positively
correlated with a higher risk of readmission and which specific heart failure intervention
strategies have been found to help curb readmissions. Second, I provide an overview of
my study including the methodology and data analysis construct. Third, I present the
results of my study. Fourth, I discuss the relevance and potential extensions of my
findings. I conclude with guidance for health care organizations attempting to reduce
heart failure readmissions by outlining a patient-centered strategy that encompasses both
the overall program and the specific process design.
Literature Review
I studied three topics within the literature regarding readmissions: personal characteristics
that may increase readmission risk, general strategies related to reducing readmissions,
and heart failure specific interventions to reduce readmissions. The literature contains an
3
extensive amount of research these topic areas, yet it is difficult to compare them due to
variation in study design, measurement, and the usual care provided.
I observed twenty-nine different personal characteristics studied in the literature, which I
segmented into patient demographics, clinical conditions, and health care utilization. I
found considerable conflicting results regarding which patient demographics may lead to
a readmission; yet three factors appeared to indicate a higher likelihood of readmission:
discharge to a community setting, lower educational level and literacy, and lower life
satisfaction. Studies consistently also found patients with certain diseases including heart
failure, heart disease, renal failure, COPD, and cancer have higher readmission rates.
Additionally, patients with more emergency department visits and hospital stays in the
previous twelve months are also correlated with greater readmission risk (Smith, Norton,
& McDonald, 1985) (Smith, et al., 2000) (Fethke, Smith, & Johnson, 1986) (Corrigan &
Martin, 1992) (Hasan, et al., 2009).
My review of the literature related to general readmissions suggested no single
intervention or bundle of interventions consistently has a significant effect on lowering
readmissions. However, an emerging predictive tool called LACE was published in
Canada to help identify patients with more previous admissions, comorbidities, and
emergency department visits (van Walraven, et al., 2012). It does not provide guidance
on how to avoid the readmissions, but may prove useful to organizations that must focus
limited resources on the most at-risk patients.
4
I summarized the heart failure literature according to the interventions they studied and
their resulting 30-day readmission rates. I found some studies reported considerable
improvement by testing a single intervention while others did not. As I examined the
studies in detail, I found the differences were usually attributable to variation in usual
care and baseline readmission rates. Therefore, instead of trying to understand which
interventions tested contributed to lower readmission rates, I broadened my scope to
understand which care provided (both studied interventions and usual care) lead to the
lowest readmission rates. I found that heart failure patients benefit from regular care from
a cardiologist, and intensive care in the hospital and after discharge; and that any specific
interventions provided to the patient should be tailored based on the in the individual
patient’s needs rather than as a uniform practice.
Study Overview and Methodology
The literature review left me contemplating a question: with limited financial resources
and increasing pressure to reduce avoidable readmissions, how can health care providers
most effectively and efficiently reduce heart failure readmissions? I studied a southern
California health system’s heart failure program to: 1) determine if there is an empirical
relationship between interventions heart failure patients receive and if they are
readmitted, and 2) understand to what extent these interventions are or may be
personalized to yield the most optimal results.
5
My research is comprised of two parts. First, I conducted case studies that included a
process assessment in nine of the system’s hospitals to understand how to improve the
process reliability and identify effective practices that may be spread. I also participated
in select patient observations in higher performing hospitals. I compiled system-wide
findings into a final report and summarized improvement opportunities. After the process
assessment, I worked with two hospitals in readmissions reduction initiatives. As a part
of those initiatives, physicians conducted chart reviews for quality improvement purposes
to understand what more could have been done to avoid a readmission. Other care
providers also interviewed select patients to ascertain what endogenous and exogenous
factors might have contributed to their return to the hospital. Second, I analyzed data
from 8,900 heart failure patient admissions for the calendar year 2010 using fuzzy set /
Qualitative Comparative Analysis (fs/QCA) software (Ragin, Drass, & Davey, Fuzzy-
Set/Qualitative Comparative Analysis 2.0, 2006). fsQCA is rooted in the principles of
Boolean algebra and provides a construct for evaluating which causal conditions can or
may exist that lead to a desired outcome: in my evaluation, the desired outcome is a
patient who is not readmitted (Ragin C. C., 1987). I constructed over fifty models with
varying combinations of ten variables to understand which different combinations at
different levels of consistency consistently led to a patient not being readmitted. The ten
available variables in my data set were:
6
Interventions: home health visit within 48 hours of discharge, real-time
medication reconciliation during the home health visit, physician visit within
seven days of discharge, and follow-up call within seven days of discharge
Participation in the heart failure program
Personal characteristics: age, race, gender, language, and discharge disposition
My study has several limitations. First, I did not include extensive patient interviews.
Second, this study was limited to enrollees in a health system; therefore findings may not
necessarily extend to indigent persons who would not typically have access to regular
health care. Third, I was unable to access patient-specific psycho-social or clinical
information, such as literacy or number of co-morbidities to determine if outcomes may
be influenced by those variables. Fourth, the data portion of my study only measures if an
intervention was completed, not the quality of each intervention. Fifth, the real-time
medication reconciliation intervention in my data analysis was new to the program and
not being done with great fidelity. Sixth, some of the intervention data was derived from
manual documentation and may contain inaccuracies.
Results
The case studies left me with six impressions: 1) processes were designed to achieve
desired outcomes and the metrics by which people are measured may have an influence
on process reliability and outcomes, 2) individual roles were often defined to complete
7
specific tasks rather than achieve outcomes, 3) a lot of documentation existed outside the
electronic health record, including spreadsheets, databases, paper forms, and other
scheduling systems, 4) patients’ personal issues appear be a common driver to
readmissions, 5) each patient received the same bundle of interventions regardless of
need, and 6) the heart failure program was designed to treat a disease, yet people with
heart failure often have other health issues. These observations left me wondering if
dispensing the same interventions to each patient was in our patients’ best interest and, if
there might be a more patient-centered and effective means of reducing heart failure
readmissions.
I conducted the data analysis to support or refute my theory that different patients might
benefit from different interventions. I found that when examined individually, the
physician visit, follow-up call, and real-time medication reconciliation each provided an
incremental benefit toward avoiding a readmission, but a home health visit did not. These
themes held true throughout all of the models, although with different combinations of
intervention and personal characteristic variables on occasion the opposite held true.
Depending upon the variables included, I got very specific results indicating that certain
individuals had different outcomes with the same set of interventions provided. I also
stratified the data by language, discharge disposition, and hospital and found further
variation.
8
My data analysis results suggest delivering the same bundle of interventions to all heart
failure may not be either the most beneficial for patients or the most effective use of
limited resources to prevent heart failure readmissions. Personal issues appear to be
important drivers of heart failure readmissions. Therefore understanding each patient’s
background, lifestyle, culture, and individual needs including other health conditions may
be important to determining the optimal plan of care for avoiding a rehospitalization.
These results are consistent with the literature, which indicates providers who developed
personal care plans with their patients achieved the lowest readmission rates.
Discussion
There is an apparent paradigm shift underway from a paternalistic model of health care in
which the physician holds most of the power toward a more “patient-centered” approach.
While this is an increasingly common term in literature, there is not one standard
definition for what constitutes patient centeredness or the elements of it (Lauver, et al.,
2002). At its core, this concept generally refers to inclusion of a patient’s values,
objectives, and support structure when planning and delivering care. For individuals who
have chronic conditions such as heart failure, for which diet is restricted, medications are
many, and co-morbidities are common, patient-centeredness is of even greater
importance to improve quality of life (Downs & Mackenzie, 2006) (Sabat, 2006). Cowie
observed that while our national health policy is changing to reflect greater collaboration
9
between patient and provider, the evidence suggests cardiologists have been slower to
adopt patient-centeredness concepts than primary care providers (Cowie, 2011).
Chassin and Loeb published a paper regarding a hand hygiene improvement initiative
between with eight hospitals. The teams found in aggregate fifteen barriers to instituting
effective hand hygiene practices, but each participating hospital experienced a differing
subset (Chassin & Loeb, 2011). Therefore, if any one hospital had taken the prescribed
solution set that had worked at another and implemented it without understand their own
specific needs; they would not have achieved the same success. They concluded the key
to improvement is to first understand the root cause of the problem each hospital is
experiencing and then apply the appropriate subset of solutions.
I extend this concept to the delivery of health care and propose a patient-centered
framework for reducing heart failure readmissions that begins with understanding each
patient’s individual needs then administering the specific interventions that would best
benefit them rather than taking a blanket approach to delivering each intervention to each
patient.
Conclusions
Delivering care consistently is far more difficult than determining the right interventions
to provide (Pronovost, et al., 2006). Thus, even a well developed patient-centered
10
readmission reduction framework will not be effective if the intended care is not
delivered reliably. As health care leaders seek to understand what has worked in other
organizations, I suggest they should evaluate not only which interventions to provide
patients, but also the organizational construct in which those processes are delivered. I
conclude with guidance for effectively designing a heart failure readmission reduction
program with engaged leaders, program oversight that includes leaders and frontline
staff, and well designed processes that focus on the patient and encourage continuous
improvement.
11
Chapter Two: Literature Review
Introduction
There is a plethora of literature regarding hospital readmissions, so much so that several
meta-analyses review the literature to identify commonalities and differences. There are
also multiple published risk classifications to accurately predict future readmissions. My
intent is not to encapsulate all published findings for a defined time period, but rather to
understand the basic tenor of literature related to non-pediatric, medical readmissions
including opportunities to further contribute.
I review the literature regarding strategies for reducing heart failure readmissions in three
parts. First, I evaluate studies regarding patient and provider characteristics that may be
positively correlated to readmissions. Second, I assess trials related to reducing adult
medicine readmissions to understand which interventions are consistently found to reduce
readmissions. Third I examine the literature specific to interventions that may prevent
heart failure readmissions. I conclude the comparative benefit of specific interventions is
challenging to derive due to the variations in study design, measurement, and specific
patient cohort; nonetheless I believe there is an underlying theme that to achieve
considerable reductions in heart failure readmissions, patients require intensive,
personalized care that begins in the inpatient setting and continues indefinitely.
12
Patient and Provider Characteristics Positively Correlated to Readmissions
Health care professionals have conducted numerous studies to better understand which
patient or provider characteristics could help predict future rehospitalizations. In
reviewing the literature, I discovered three subsets of patient data have been studied:
demographics, clinical condition, and previous emergency department and hospital
utilization. I found research varies widely in terms of purpose, scope, and approach; thus
different studies conclude that different subsets of factors are positively correlated with a
higher likelihood of readmission. Furthermore, some findings are contradictory. Yet,
there is a common theme among the aggregate findings: who we are and where we
receive care seems to affect the likelihood of us being rehospitalized.
In total, I identified twenty-nine personal characteristics cited in the literature that may be
positively related to a higher risk of readmission as enumerated in Table 1:
Table 1: Personal Characteristics Found in Literature
Patient’s Demographics Patient’s Clinical Condition Patient’s Utilization of Health
Care Resources
1. Increased Age
2. Medicare or Medicaid
Insurance Coverage
3. Male Gender
4. Non-White Race
5. Non-English Speaking
6. Clinical Depression
1. Specific Clinical Diagnoses
2. Number of Chronic Illnesses
/ Comorbidities
3. Nutritionally Compromised
4. Greater Illness Severity
5. Other Clinic Indicators
6. Presence of Hospital
Acquired Infections
1. Higher Number of
Emergency Department
Visits
2. Satisfaction with Emergency
Department Visits
3. Greater Number of
Hospitalizations
4. Hospital Length of Stay
13
Patient’s Demographics Patient’s Clinical Condition Patient’s Utilization of Health
Care Resources
7. Lower Life Satisfaction
8. Not Married
9. Living Alone
10. Physical Disabilities
11. Lower Mental Function or
Cognitive Impairment
12. Non-Readiness for Discharge
13. Discharge Home or to a
Community Setting
14. Lower Literacy
15. Smoking and Alcohol Use
16. More Concurrent Drugs
17. Drug Non-Adherence
18. Lower Income
19. Not Having Religious Beliefs
Patient’s Demographics
The literature contains conflicting information about the relationship between a person’s
gender, age, race, and readmission risk. Four studies found gender to be a predictive
factor (Billings, Dixon, Mijanovich, & Wennberg, 2006) (Fethke, Smith, & Johnson,
1986) (Kossovsky, Perneger, Sarasin, Bolla, Borst, & Gaspoz) (Silverstein, Qin, Mercer,
Fong, & Z, 2008), while several others found it did not (Holloway & Thomas, 1989)
(Smith, Norton, & McDonald, 1985) (Hasan, et al., 2009). In a 1992 review of the
literature, Corrigan summarized that gender studies found men had a higher likelihood of
90-day readmission, but there was no difference between the sexes across a 12-month
period (Corrigan & Martin, 1992). She also found age to be positively correlated to
14
readmission risk up to a certain point, after which time the correlation ceases. However a
European study of COPD patients found neither age nor gender was associated with
readmissions (Gudmundsson, et al., 2005), and an older study by Holloway found neither
age, gender, payer status nor ethnicity had any effect on readmissions (Holloway &
Thomas, 1989).
Ottenbacher also found no correlation between age and readmission risk when he
examined persons with disabilities (Ottenbacher, Smith, Illig, Fiedler, & Granger, 2000),
nor did Hasan in a study across six academic medical centers (Hasan, et al., 2009).
However, in this latter study, the team discovered Medicare beneficiaries were more
likely to be readmitted than others. Separate studies of medical-surgical adults in a
community hospital (Corrigan & Martin, 1992) and an urban hospital also found
Medicaid beneficiaries had higher readmission rates (Allaudeen, Vidyarthi, Maselli, &
Auerbach, 2011). Five other studies specifically found older age to be positively linked
with readmission (Billings, Dixon, Mijanovich, & Wennberg, 2006) (Billups, Malone, &
Carter, 2000) (Corrigan & Martin, 1992) (Bathaei, Ashktorab, Zohari, Alavi, & Ezati,
2009). Conversely, Kossovsky found younger age was linked with rehospitalization in a
study of general medicine patients (Kossovsky, Perneger, Sarasin, Bolla, Borst, &
Gaspoz).
The influence of race on a person’s risk of readmission was somewhat more consistent,
although it is unclear if race is an independent factor. Alexander conducted a study of
15
congestive heart failure hospitalizations and found African Americans and Latinos were
more commonly readmitted than Caucasian patients (Alexander, Grumbach, Remy,
Rowell, & Massie, 1999). Jiang had similar findings in a study of diabetic patients:
African Americans had higher readmission rates than Caucasians, but only in the
Medicare cohort. The authors suggested this might have been a reflection of lower
income levels and patients’ ability to manage their diabetes (Jiang, Andrews, Stryer, &
Friedman, 2005). Philbin also found lower income to be a significant factor in predicting
risk when controlling for other variables (Philbin, Dec, Jenkins, & DiSalvoi, 2001). Other
studies similarly found race in combination with other factors influenced readmission
risk: Allaudeen’s review of general medicine patients in urban teaching hospitals found
ethnicity, insurance, and age were statistically significant factors in patient readmissions
(Allaudeen, Vidyarthi, Maselli, & Auerbach, 2011), and Joynt’s study of Medicare
beneficiaries concluded African American patients who receive care in hospitals that
serve large populations of non-whites had higher readmission rates than Caucasians or
African Americans who received care in other institutions. (Joynt, Orav, & Jha, 2011).
Several studies show an elevated readmission risk for patients discharged to the
community setting (i.e. a private home) versus a nursing home (Anderson, Helms,
Hanson, & DeVilder, 1999) (Bowles, Naylor, & Foust, 2002) (Fethke, Smith, & Johnson,
1986) (Camberg, Smith, Beaudet, Daley, Cagan, & Thibault, 1997) (Corrigan & Martin,
1992) (Chu & Pei, 1999). The same risk seems to exist for people who are single or live
alone (Evangelista, Doering, & Dracup, 2000) (DiIorio, et al., 1998). Kossovsky found a
16
correlation between a patient’s readiness for discharge and being readmitted (Kossovsky,
Perneger, Sarasin, Bolla, Borst, & Gaspoz), which is consistent with a previous study
related to poor quality of inpatient care (Ashton, Kuykendall, Johnson, Wray, & Wu,
1995).
Also consistent are findings related to education and life satisfaction or quality of life.
Three studies found patients with lower health literacy had an increased risk of returning
to the hospital in part because information is not conveyed to them in a way they can
understand (Powell & Kripalani, 2005) (Baker, Parker, Williams, & Clark, 1998)
(Bathaei, Ashktorab, Zohari Anbuhi, Alavi Majd, & Ezzati, 2009). Three studies of
elderly adults found lower life satisfaction or quality of life, including poor satisfaction
with their social conditions, had an influence on more admissions (Fethke, Smith, &
Johnson, 1986) (DiIorio, et al., 1998) (Mejhert, Kahan, Persson, & Edner, 2006). Two
European articles had contradictory results regarding the affect of clinical depression on
readmission risk: a heart failure study found clinical depression increased readmission
risk (Farisa, Purcella, Heneina, & Coatsa, 2002), but a study of COPD patients found it
had no statistical significance whatsoever (Gudmundsson, et al., 2005).
A handful of articles each cited other unique patient characteristics that may contribute to
readmission risk, such as: non-English speaking (Karliner, Kim, Meltzer, & Auerbach,
2010), those with a lower mental health function (Smith, et al., 2000), older age with
cognitive impairment (DiIorio, et al., 1998), individuals who use tobacco and alcohol
17
(Evangelista, Doering, & Dracup, 2000), people on a greater number of concurrent
medications (Billups, Malone, & Carter, 2000), those who take narcotics and
corticosteroids while inpatients (Allaudeen, Vidyarthi, Maselli, & Auerbach, 2011),
people who do not adhere to their medication regimen (Bero, Lipton, & Bird, 1991), and
individuals with a lower income (Bero, Lipton, & Bird, 1991) (Jiang, Andrews, Stryer, &
Friedman, 2005). I only found one article related to body mass index, which found BMI
was not statistically significant in a study of acute myocardial infarction patients (Wells,
Gentry, Ruiz-Arango, Dias, & Landolfo, 2006).
From this portion of my review, I conclude there still remains some question about which
individual characteristics, if any, may consistently contribute to a person’s risk of being
rehospitalized. An English study of patients over sixty-five years of age found
readmissions were entirely independent of individual demographics and social
environment, and only due to their clinical condition (Victor & Vetter, 1985). Thus I
expanded my scope to incorporate the influence of a person’s clinical condition on
readmissions.
Patient’s Clinical Condition
The literature indicates there is a strong correlation between a person’s clinical condition
and readmission risk. Studies found patients with certain diseases such as heart failure,
heart disease, renal failure, COPD, and cancer (Smith, et al., 2000) (Kossovsky, Perneger,
Sarasin, Bolla, Borst, & Gaspoz) (Evangelista, Doering, & Dracup, 2000) (Allaudeen,
18
Vidyarthi, Maselli, & Auerbach, 2011) (Chu & Pei, 1999), patients whose disease state is
more advanced (Fethke, Smith, & Johnson, 1986) (Corrigan & Martin, 1992) (DiIorio, et
al., 1998) (Gudmundsson, et al., 2005), and patients with more comorbidities (Anderson,
Helms, Hanson, & DeVilder, 1999) (Billups, Malone, & Carter, 2000) (Kossovsky,
Perneger, Sarasin, Bolla, Borst, & Gaspoz) (DiIorio, et al., 1998) (Hasan, et al., 2009), or
those who are nutritionally compromised (Friedmann, Jensen, Smiciklas-Wright, &
McCamish, 1997) (Allaudeen, Vidyarthi, Maselli, & Auerbach, 2011) have a greater
chance of being readmitted. Other studies found patients specific clinical indicators
including elevated blood urea nitrogen (BUN) levels and anemia (Smith, Norton, &
McDonald, 1985) (Smith, Norton, & McDonald, 1985), or who had a hospital acquired
infection (Sreeramoju, Montie, & Ramirez, 2010) also had a higher readmission risk.
Patient’s Utilization of Health Care Resources
Four factors regarding an individual’s health care utilization and association with
rehospitalizations were cited in the literature: number of previous emergency department
visits, satisfaction with the emergency department visits, number of previous
hospitalizations, and hospital length of stay during the index admission. Rather
consistently, a higher number of emergency department and previous hospital stays in the
past twelve months visits were found to be correlated with a higher likelihood of
readmission (Smith, Norton, & McDonald, 1985) (Smith, et al., 2000) (Fethke, Smith, &
Johnson, 1986) (Corrigan & Martin, 1992) (Hasan, et al., 2009). Smith also found
patients who experienced higher satisfaction with previous emergency department visits
19
were more likely to return for care and be readmitted (Smith, et al., 2000). Five additional
studies cited hospital length of stay as a predictor, but the precise length of stay they each
found was inconsistent. Heggestad suggested lower lengths of stay increase readmission
risk, (Heggestad, 2002), Hasan said any hospitalization more than two days increased the
risk (Hasan, et al., 2009), Kossovsky found the risk to increase if the hospitalization was
more than three days (Kossovsky, Perneger, Sarasin, Bolla, Borst, & Gaspoz), and
Corrigan had a general finding that the greater the stay, the greater the risk (Corrigan &
Martin, 1992). Reflective of this correlation between utilization and readmission risk,
Canadian researchers developed a readmission predictive tool called LACE, a pneumonic
for length of stay, acuity, comorbidities, and number of emergency department visits in
the past six months (van Walraven, et al., 2012). While this tool is proving useful in
predicting readmission risk, it does not provide direction on what to do to prevent
readmissions.
Provider Characteristics
In reviewing the literature on patient characteristics, I inadvertently found mention of
specific provider characteristics that may also affect readmissions. Hasan and Holloway
both found patients who regularly received care from any physician had lower
readmission risk (Holloway & Thomas, 1989) (Hasan, et al., 2009). In his study of a
Midwest community hospital, Corrigan found younger attending physicians’ patients had
higher readmission rates (Corrigan & Martin, 1992). Evangelista’s study of veterans with
heart failure found patients who received care from a cardiologist were less than half as
20
likely to be readmitted as those receiving care from a primary care physician
(Evangelista, Doering, & Dracup, 2000). Institutional providers may also affect a
person’s likelihood of returning to a hospital: an Australian study found patients
discharged from public hospitals had a higher readmission risk than those discharged
from private hospitals (Ansari, Collopy, & Booth, 1995) and Ashton found patients are
over fifty percent more likely to be readmitted when they receive substandard inpatient
care during the index admission (Ashton, Del Junco, Souchek, Wray, & Mansyur, 1997).
In a study of heart failure patients, Chin noted that physicians who discharge patients
before they are ready to return home also increase the likelihood of a readmission (Chin
& Goldman, 1997)
Interventions that May Reduce General Readmissions
Hansen conducted a meta-analysis of the literature and identified twelve types of
interventions which he grouped into three categories as illustrated in Figure 1:
predischarge, postdischarge, and transitional (Hansen, Young, Hinami, Leung, &
Williams, 2011) .
21
Figure 1: Interventions Studied to Reduce Hospital Readmissions
Hansen attempted to statistically and conclusively identify which specific interventions
had greater affects on lowering readmissions, but he was limited for several reasons:
variation in study and intervention design, the likely variation in intervention quality,
differences in the “usual care” that patients in the control groups received, and
differences in the baseline readmission rate from which they were trying to improve. For
example, he examined seventeen studies that included follow-up calls, but the timing and
content of the call varied, and some tested it as a single intervention, while others tested it
as a part of a bundle. Similarly, post-discharge follow-up visits ranged from one week to
four weeks after leaving the hospital. Of the forty-three articles included in his review,
only ten were both randomized and controlled. No study indicated a single or bundle of
interventions had a statistically significant effect on readmissions. Mistiaen found the
same thing: study results conflicted (Mistiaen, Francke, & Poot, 2007). The authors
suggested a bundle of interventions, rather than a single one, aimed at the transition after
22
hospital discharge is necessary to make dramatic improvements. This is consistent with
other research that has also found in some cases interventions increase readmissions
(Jencks S. , 2010) (Weinburger, Oddone, & Henderson, 1996) (Holland, et al., 2005).
Hansen concluded further patient-centered studies on strategies to reduce readmissions
are needed (Hansen, Young, Hinami, Leung, & Williams, 2011).
Interventions that May Reduce Heart Failure Readmissions
The complexity and prevalence of heart failure has led to a proliferation of interventions
and disease management programs designed to improve inpatient care, the transition to
the outpatient setting, and the post-hospital discharge period as illustrated in Hansen’s
model. I review heart failure meta-analyses to understand the current thinking about
successful interventions for reducing readmissions.
Yu, Gonseth, Roccaforte, Gwadry-Sridha, Holland, and Clark each conducted a meta-
analysis of heart failure interventions (Yu, Thompson, & Lee, 2006) (Gonseth, Guallar-
Castill, Banegas, & Rodrıguez-Artalejo, 2004) (Roccaforte, Demers, Baldassarre, Teo, &
Yusuf, 2005) (Gwadry-Sridhar, Flintoft, Lee, Lee, & Guyatt, 2004) (Holland, Battersby,
Harvey, Lenaghan, Smith, & Hay, 2005) (Clark, Inglis, McAlister, Cleland, & Stewart,
2007). Their objectives, inclusion criteria and scope varied, thus the articles included in
each varied although there was substantial overlap.
23
I summarized each article studied in these six meta-analyses according to Hansen’s
framework in Table 2 below. I modified his framework slightly to account for studies that
tested select inpatient interventions in the post-discharge setting. I also excluded a few
studies that either did not explicitly target readmissions reduction or whose original
source article I was unable to access to verify specific interventions studied and results. I
segmented the trials according to which had inpatient interventions only, outpatient
interventions only, and both inpatient and outpatient, and compared their scope according
to Hansen’s framework and their net effect on thirty-day readmissions. I faced the same
challenges as other researchers in variability of the patient cohort studied, study scope,
usual care provided, and measurement used.
24
Table 2: Review of Heart Failure Interventions in the Literature
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
INPATIENT INTERVENTIONS ONLY
(Philbin,
Rocco,
Lindenmuth,
Ulrich,
McCall, &
Jenkins, 2000)
USA; 10 Hospitals
with 2906 HF
patients NYHA III-
IV.
No significant
change
(Varma,
McElnay,
Hughes,
Passmore, &
Varma, 1999)
Northern Ireland;
83 HF patients > 65
years
Not stated. Fewer
readmissions for
the intervention
group, period not
stated
OUTPATIENT INTERVENTIONS ONLY
(Blue, et al.,
2001)
UK; 165 HF
patients; outpatient
23% (intervention)
versus 55%
(control) - period
not stated
25
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
(Holland, et
al., 2005)
UK; 872 HF
patients ≥80 years.
Home-based
pharmacy
intervention
55% (intervention)
versus 42%
(control)
(Capomolla, et
al., 2002)
Italy 234 HF
patients;
individualized HF
management
program
Not stated. 8%
(intervention)
versus 35%
(control) at one
year
(Stewart,
Marley, &
Horowitz,
1999)
Australia; 200 HF
patients
Not stated. 34%
(intervention) vs.
58% (control) at 6
months
(Krumholz, et
al., 2002)
USA; 88 HF
patients; outpatient
Not stated. 50%
(intervention)
versus 95%
(control) at one
year
26
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
(Stromberg,
Martensson,
Fridlund,
Levin,
Karlsson, &
Dahlström,
2003)
2-3
weeks
Sweden; 106 HF
patients in nurse-led
HF clinic
Not stated. 63%
(intervention)
versus 100%
(control) all cause
at 3 months
(Kasper, et al.,
2002)
USA; 200 high-risk
HF patients.
Multidisciplinary
outpatient program
Not stated. Higher
for intervention
group until 2
months, then lower
(Kimmelstiel,
et al., 2004)
USA; 200 HF
patients at six sites.
Nurse management
with Cardiologist
support
Not stated. Lower
HF readmission
rate at 90 days, but
did not sustain
after several
months
27
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
(Mejhert,
Kahan,
Persson, &
Edner, Limited
long term
effects of a
management
programme for
heart failure,
2004)
Sweden; 288 HF
patients > 60 years.
Outpatient nurse
management
No difference.
(Ekman,
Andersson,
Ehnfors,
Persson, &
Fagerberg,
1998)
Sweden; 158 HF
patients NYHA III-
IV ≥65 years of age
No difference.
Found hard for
elderly ill to come
to clinic
(R.Doughty, et
al., 2002)
New Zealand; 197
HF patients NYHA
III-IV
No difference.
Intervention group
lower after 30 days
28
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
(Stewart,
Pearson, &
Horowitz,
Effects of a
Home-Based
Intervention
Among
Patients With
Congestive
Heart Failure
Discharged
From Acute
Hospital Care,
1998)
Australia 97 HF
patients
No difference.
Intervention group
lower after 8
weeks
Outpatient Structured Telephone Support or Telemonitoring
(Goldberg,
Piette, Walsh,
& Frank, 2003)
As
needed
*
Tele-
moni-
toring
USA; 280 HF
patients NYHA III-
IV. Daily weight
telemonitoring
No difference.
29
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
(Cleland,
Louis, Rigby,
Janssens, Balk,
&
Investigators,
2005)
Monthly
plus as
needed
*
Tele-
moni-
toring
Europe; 426 HF
patients. Monthly
nurse calls or daily
weight and vitals
telemonitoring
Not stated.
Telemonitoring
group was higher
at 240 days: 41%
versus 28% nurse
calls, 39% control
(Riegel,
Carlson, &
Kopp, 2002)
USA 358 HF
patients; outpatient
telephonic case
management
Not stated. 14.6%
(intervention)
versus 22.8%
(control) at 3
months
(GESICA
Investigators,
2005)
Argentina; 1518
stable HF patients;
nurse telephonic
education and
support
No difference.
26% (intervention)
versus 31%
(control) at several
months
(Riegel B. ,
Carlson,
Glaser, &
Romero, 2006)
USA; 134 Hispanic
HF patients.
Spanish education
and telephonic case
management
No significant
difference
30
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
(Barth, 2001) USA; 34 HF
patients discharged
home
Not stated. No HF
readmissions in
either group for
two months
INPATIENT AND OUTPATIENT INTERVENTIONS
(Ledwidge, et
al., 2003)
Finland; 98 HF
patients; frequent
multidisciplinary
patient support and
education with
intensive usual care
Not stated. 3.9%
(intervention)
versus 25.5%
(control) at 3
months
(Jaarsma, et
al., 1999)
Netherlands 179 HF
NYHA III-IV
patients
10% (intervention)
vs. 12% (control)
for cardiac causes
(Atienza, et al.,
2004)
Spain: 338 patients
with
decompensated
heart failure
23% readmission
rate versus 45%
(control group);
period not stated
31
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
(Harrison,
Browne,
Roberts,
Tugwell,
Gafni, &
Graham, 2002)
Canada; 192 HF
patients. Nurse-led
transitional care for
2 weeks post
discharge
Not stated. 23%
(intervention) vs.
31% (control) at
12 weeks
(McDonald, et
al., 2002)
Ireland 98 NYHA
class IV HF patients
Not stated. Overall
7.8%
(intervention) vs.
25.5% (control) at
12 weeks
(Naylor,
Brooten,
Campbell,
Maislin,
McCauley, &
Schwartz,
2004)
USA; 239 HF
patients ≥65 years
of age; care
transition with APN
Not stated. 34%
(intervention)
versus 60% at one
year
32
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
(Rich,
Beckham,
Wittenburg,
Leven,
Freedland, &
Carney, 1995)
USA; 98 HF
patients ≥70 years
of age
Not stated. 17%
(intervention)
versus 38% (usual
care) at 90-day
(Gattis,
Hasselblad,
Whellan, &
O’Connor,
1999)
USA; 187 HF left
ventricular
dysfunction
patients; pharmacist
intervention
Not stated. Lower
readmission rate
for intervention
group
(Cline,
Israelsson,
Willenheimer,
& Broms,
1998)
Sweden; 190 HF
patients ≥65 years
of age
Intervention group
lower (not stated)
at 30 days, but no
difference at one
year
33
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
(Laramee,
Levinsky,
Sargent, Ross,
& Callas,
2003)
USA 287 HF
patients
No difference.
Subset that saw
Cardiologist for
follow up had
significantly lower
readmission rate.
(Goodyer,
Miskelly, &
Milligan,
1995)
Meds
only
UK; 100 HF
patients ≥70 years
of age
Not stated; 52%
increase in med
compliance
(DeBusk, et
al., 2004)
USA; 462 low risk
HF HMO patients.
Nurse care
management
No statistically
significant
reduction
(Weinburger,
Oddone, &
Henderson,
1996)
USA; 504 HF
patients at veterans
hospitals
Not stated.
Intervention group
had higher
readmission rate
34
Heart Failure
Readmissions
Study
Pre-Discharge
(D/C)
Pre or Post-
Discharge
(D/C)
Post-Discharge (D/C) Hospital to Home
Transition
Study Scope Effect on Heart
Failure 30-Day
Readmissions
(if reported)
Discharge
Planning
MD Appointment
Scheduled pre D/C
Patient Education
Medication
Reconciliation
Timely MD/Clinic
Follow Up
Hospitalist - PCP
Communication
Follow-Up Call
Patient Hotline
Home Visit
Transition
Coach
Post Discharge
Instructions
Provider
Continuity
(Oddone,
Weinberger,
Giobbie-
Hurder,
Landsman, &
Henderson,
1999)
USA; 443 HF
patients in 9
Veterans Affairs
hospitals. Increased
outpatient access
Intervention group
had higher
readmission rate
(not specified)
Total Number
of Studies with
Each
Intervention
5 1 30 15 12 3 22 14 10 6 0 15
= intervention studied; = intervention provided as needed
35
From this review of the heart failure readmission reduction literature, I found that, like
Hansen and Mistean, it was difficult to extrapolate which specific interventions were
most effective for specific patient groups due to the studies’ variability in design, patient
cohorts, and usual care provided, as well as over half of these studies were conducted
outside of the US which have different cultures, institutions and approaches to care. I also
discovered that some studies noted success in lessening readmissions, yet their resulting
end point was still considerably high. For example, Stromberg’s control group had a
100% readmission rate at 3 months, so although the interventions reduced readmissions
by one-third, the study group still had a 60% ninety-day readmission rate, which is far
higher than Ledwidge, Naylor, McDonalds, or several other studies’ control group
results.
Therefore, to understand which care yielded the lowest readmission rates, I focused on
studies that achieved the lowest readmission rates to understand their entire care model
(usual care plus interventions studied). I discovered these five trials share several
commonalities and varied from the others: Capomolla, Barth, Jaarsma, Ledwidge, and
McDonald.
Capomolla’s trial focused on outpatient interventions; however, the team demonstrated a
reduction to single digit readmission rates: 8% at one year (Capomolla, et al., 2002).
Usual care in this Italian hospital included: an extensive evaluation to measure each
patient’s condition and functional status, development of a customized care plan, and
36
follow-up care by both a primary care physician and a cardiologist. The usual care
(control) group had a 35% readmission rate at one year – far better than the intervention
results of many other studies. The intervention group’s individualized plans were carried
out by a collaborative, multidisciplinary team, including cardiologist, nurse, psychologist,
physiotherapist, and dietician in a day hospital setting. Patients were not just educated on
heart failure, but provided continuous counseling to promote self care in daily life, such
as a personalized exercise plan. I believe these results may be attributed to both the
quality of usual care provided and personalized interventions.
Barth’s trial of a nurse-led outpatient case management program examined the effect of a
single intervention on readmissions (Barth, 2001). The nurse scheduled specific times to
talk with patients in the study group starting at 72 hours after discharge and continuing
for two months whereas the control group only received patient education at discharge on
diet, medication, activities of daily living, and symptoms for which they should contact
their physician. The entire study only included thirty-four patients in a Midwestern US
hospital and at the end of two months, neither group had a readmission. There was
insufficient information about the usual care provided to understand the incremental
affect the intervention may have had. Nonetheless I found it interesting that no patients
returned to the hospital.
Jaarmsa’s study was designed to test personalized, intensive follow-up care, but it was
very short-term: from hospitalization to ten days post-discharge (Jaarsma, et al., 1999).
37
The intervention group achieved a 10% cardiac readmission rate at one month, but the
control group’s was only 12%. This Dutch study was limited to literate patients, who are
at a lower risk of readmission according to the literature, and the hospital average length
of stay for all patients in the study was 13.6 days, far greater than the average American
hospitalization. It might be that they allowed patients to become more stable before
sending them home, which may partially explain the favorable results.
Ledwidge achieved a remarkable 3.9% readmission rate at three months for high-risk
NYHA Class IV patients (Ledwidge, et al., 2003). In this study, they strove to provide
best possible care for both intervention and control group participants to better measure
the incremental benefit of ongoing patient education and support. Thus, usual inpatient
care included a cardiologist, medication therapy, and discharge criteria that required
patients to be more stable and at their dry weight for at least two days. After discharge,
patients in the control group were referred back to their general practitioners for ongoing
care. The intervention group received regular educational and supportive calls from a
heart failure nurse, plus ongoing care in the heart failure clinic. McDonald’s study nearly
mirrored Ledwidge’s in terms of scope, approach and results (McDonald, et al., 2002).
These studies shared several things in common. First, Laramee and others found that
people with heart failure return to the hospital less often if they are receiving care from a
cardiologist instead of a generalist (Laramee, Levinsky, Sargent, Ross, & Callas, 2003)
(Capomolla, et al., 2002) (Atienza, et al., 2004) (Yu, Thompson, & Lee, 2006)
38
(McDonald, et al., 2002). This is consistent with other studies that have found heart
failure patients have better health outcomes if there care is managed by a cardiologist
perhaps in part because they are more apt to follow prescribed care guidelines (Ansari,
Alexander, Tutar, Bello, & Massie, 2003) (Reis, Edmundowicz, McNamara, Zell, Detre,
& Feldman, 1997) (Laramee, Levinsky, Sargent, Ross, & Callas, 2003). Second, these
studies combined intensive, inpatient care with intensive, ongoing, outpatient follow-up.
This supports Jencks’ and Yu’s assertions that effective heart failure programs should be
multidimensional and encompass the entire care continuum (Jencks S. , 2010) (Yu,
Thompson, & Lee, 2006). Third, patients received personalized care plans rather than a
blanket set of interventions. For example, Ekman found the post-discharge clinic
appointment was not successful with patients over 85 years of age because they were
unable to travel to their appointment (Ekman, Andersson, Ehnfors, Persson, & Fagerberg,
1998).
Conclusion
The heart failure studies in the literature vary in construct, duration, measurement,
personnel involved, and the inherent quality of the intervention. Contributing to the
difficulty in comparing studies, organizations have different models for usual care and
overall readmission rate. I narrowed my focus to the five studies that achieved the most
remarkable reductions in heart failure readmission rates. From this I conclude patients
with heart failure benefit from regular care from a cardiologist, and intensive care in the
39
hospital and after discharge. However, the interventions provided and how they are
provided should be tailored based on the in the individual patient’s needs rather than as a
uniform practice.
40
Chapter Three: Study Overview and Methodology
Introduction
The literature review left me with a question: with limited financial resources and
increasing pressure to reduce avoidable readmissions, how can health care providers most
effectively and efficiently care for people with heart failure? I had the opportunity to
conduct research in a southern California health system’s heart failure program. I
structured my study to: 1) determine if there is an empirical relationship between
interventions heart failure patients receive and if they are readmitted, and 2) understand
to what extent these interventions are or may be personalized to yield the most optimal
results.
In this chapter, I outline my study and methodology in five parts. First I introduce a
health system’s heart failure program, which was designed to reduce 30-day readmissions
via evidence-based interventions from the inpatient setting to 180 days post-discharge.
Second I describe my research, which includes a combination of case studies and fuzzy-
set qualitative comparative analysis (fsQCA).Third I explain how my data was compiled
and what information was available for this analysis. Fourth, I outline my research
approach including the models I evaluated using the fs/QCA software. Lastly, I
enumerate the limitations of my study.
41
Heart Failure Program
A southern California health system with eleven hospitals developed a heart failure
program in 2006 to provide standard, evidence-based care that would result both in
improved patient quality of life and fewer hospital readmissions. Heart failure represents
one of the most common chronic illnesses in the US and readmission rates are often used
as an indicator of quality. The program was initially piloted in one hospital and then
spread to the rest of the system.
The program was rooted in effective practices found in the literature and leveraged the
health system’s electronic medical record. Usual care and the heart failure program
encompassed several of the components reflected in Hansen’s model and in the literature.
Discharge Planning: expected to be a component of usual care.
MD appointment scheduled before discharge: an intervention all heart failure
patients should receive, in concert with the 7-day post discharge appointment.
Patient Education: an intervention all heart failure program patients should
receive beginning with when they agree to participate in the program.
Medication Reconciliation: an intervention all heart failure program patients
should receive during the home health visit with telephonic participation by a
pharmacist.
42
Timely MD/Clinic Follow Up Appointment: an intervention all heart failure
program patients should receive within seven days of discharge.
Hospitalist - PCP Communication: not consistently included in usual care or the
program intervention.
Follow-Up Call: an intervention all heart failure program patients should receive
within seven days of discharge by an outpatient care manager.
Patient Hotline: not consistently included in usual care or the program design.
Home Visit: an intervention all heart failure program patients should receive
within forty-eight hours of discharge that should include medication
reconciliation.
Transition Coach: not consistently included in usual care or the program
intervention.
Post Discharge Instructions: all patients receive post-discharge instructions.
Provider Continuity: not consistently included in usual care or the program
intervention.
In addition to these key interventions, select medical centers and service areas also
provide standard screenings for depression, consultations with a registered dietician, heart
failure classes in a small group settings, multidisciplinary outpatient follow-up
appointments, outpatient intravenous Lasix, and end-of-life planning including hospice
and palliative care consultations.
43
Patients are offered the option of enrolling in the heart failure program. Each day, each
hospital’s heart failure program nurse reviews a report of patients admitted in the
previous twenty-four hours and identifies those with a heart failure diagnosis. The nurse
determines who has already been enrolled in the heart failure program, who has
previously declined enrollment, and who is newly diagnosed by reviewing a program log
in Excel. The nurse reviews each patient’s health record before going to the patient’s
room to introduce herself and discuss the heart failure program. Patients may opt in or opt
out of the program. If the patient opts in, the nurse begins heart failure education and
provides the patient with an education packet and her contact information. Occasionally,
patients who opt in to the program decline certain interventions, such as the home health
visit. If a patient had previously enrolled in the program, the nurse confirms their
continued participation and offers to repeat the same education. Patients sometimes
decline the repeated education.
Patients who opt out receive usual care, which varies by hospital. The heart failure nurse
usually offers education to patients who declined the program, including leaving the heart
failure written materials and the narrative education. Each hospital has varying criteria
for scheduling follow-up appointments: most offer a follow-up appointment to all
Medicare recipients within seven days of discharge with the patient’s primary care
physician instead of a cardiologist or cardiology nurse practitioner. Some hospitals
conduct post-discharge calls although they are often more oriented to address general
health questions rather than have a heart failure focus. Heart failure patients who are not
44
enrolled in the program would not receive a home health nurse visit unless they are home
bound. One hospital conducts medication reconciliation from the patient’s home that
includes the patient, the home health nurse, and a pharmacist via phone for every patient
receiving a home health visit. In all other hospitals I evaluated, only patients in the heart
failure program receive this intervention.
Each hospital’s heart failure program staffing model varies slightly. Some have a single
person fulfill this role during regular business hours, which provides consistency but
leaves a gap on weekends and holidays. Others have multiple people fulfill the role who
overlap to provide seven day-a-week coverage. One hospital delegated the responsibility
to floor nurses in a decentralized model.
The program was introduced to all system hospitals by 2008. Each site had the ability to
opt in or opt out of the entire program as well as adopt any of the recommended
interventions. Each site also had the autonomy to select the local program leadership,
oversight structure, staffing, and general rigor for monitoring program performance,
which led to variations in processes and outcomes. By December 2010, none of the
hospitals was consistently delivering all key interventions highly reliably (i.e. least 85%
performance over time). Each hospital tracked and reported process reliability for all
patients in their heart failure program as indicated in Table 3. Because inpatient education
was dependent upon and linked to the process of enrolling patients in the heart failure
45
program, it could occur more than 100% if education was offered to patients not
ultimately enrolled in the program.
Table 3: Reliability of Key Interventions for All Patients in Heart Failure Program
Since the program’s inception, the 90-day heart failure any cause readmission rate fell
from 36% to 25%. The system’s 30-day any cause readmission rate for patients in the
heart failure program was 16.5%.
Research
My research and evaluation are comprised of two parts: 1) case studies and 2) qualitative
comparative analysis of heart failure data.
Inpatient
Education
Home Health Visit
within 48 Hours
Medication
Reconciliation with
Home Health Visit
Heart Failure Care
Manager Follow-Up
Call within 7 Days
MD or Cardiology
Nurse Practitioner
Appointment within
7 Days
30-day Any Cause
Heart Failure
Readmission Rate
Hospital A 129% 95% 31% 90% 36% 16.6%
Hospital B 91% 77% 0% 52% 28% 18.5%
Hospital C 107% 74% 15% 79% 36% 15.6%
Hospital D 194% 77% 5% 69% 39% 16.8%
Hospital E 133% 86% 43% 63% 22% 21.1%
Hospital F 121% 76% 5% 81% 21% 19.0%
Hospital H 112% 68% 0% 98% 55% 13.4%
Hospital G/I 102% 74% 55% 67% 25% 12.8%
Hospital J 110% 89% 0% 100% 42% 17.0%
Hospital K 96% 56% 23% 100% 35% 15.2%
Hospital L 152% 56% 6% 83% 12% 13.0%
Average 120% 76% 16% 78% 31% 16.5%
46
1) Case Studies
I conducted a process assessment in nine of the system’s hospitals to improve the process
reliability and identify effective practices that may be spread. The heart failure program’s
leaders sought to increase the reliability of the heart failure interventions to reduce the
30-day readmission rate. For each site, I compiled a proposed assessment approach and
interview list, met with hospital leaders to confirm the approach and interview list,
conducted interviews, documented the existing processes in process maps, developed
recommended changes to the heart failure program structure and processes, and presented
the final report to the hospital’s heart failure team. I also conducted select patient
observations in higher performing hospitals. I compiled system-wide findings into a final
report and summarized improvement opportunities.
After the process assessment, I worked with two hospitals in readmissions reduction
initiatives. As a part of those initiatives, physicians conducted chart reviews for quality
improvement purposes to understand what more could have been done to avoid a
readmission. Other care providers also interviewed select patients to ascertain what
endogenous and exogenous factors might have contributed to their return to the hospital.
2) Qualitative Comparative Analysis
Multiple approaches exist for evaluating the relationship between different variables or
conditions. Social scientists have long used comparative approaches for evaluating cases
to determine how they are similar and different. Qualitative Comparative Analysis (QCA)
47
is a comparative method that provides a construct for analyzing combinations of
conditions that can or may exist and their resulting outcome to understand which are
more commonly linked with the same result. Because it is rooted in case study, the
comparative method varies from traditional statistical methods that examine the
probabilities of instances occurring. QCA may be used as a supplement to case studies to
better understand the association between different conditions (Ragin C. C., 1987).
In QCA, the variable that represents the result is called an “outcome” and the variables
studied that may lead to the outcome are “conditions.” The analysis attempts to find
“solution pathways” that most commonly and consistently lead to the outcome. A
solution’s “coverage” explains the percentage of all cases that are encapsulated in the
solution: the larger the coverage, the more important the result because it represents a
larger proportion of all cases. A solution’s “consistency” indicates the percentage of the
cases in that truth table row that have the same outcome. The greater this number, the
more consistently that row leads to the outcome. Ragin indicates researchers may have to
forego greater consistency for greater coverage as the two are somewhat inversely related
(Ragin C. , Redesigning social inquiry: fuzzy sets and beyond, 2008). He also notes the
greater the number of conditions included in the analysis, the more challenging it
becomes to get conditions with considerable coverage (Ragin C. C., 1987)
QCA is rooted in the principles of Boolean algebra (Ragin C. C., 1987). I have
summarized the components that are most relevant to my study in four points:
48
1) It uses binary data as an analytical construct to represent a variable’s presence “1” or
absence “0.” Data sets that follow this structure are defined as crisp (Ragin C. ,
Redesigning social inquiry: fuzzy sets and beyond, 2008). For example, a patient is coded
either 1 or 0 if they did see or did not see, respectively, their physician within seven days
of discharge. To complement crisp sets, social scientists developed fuzzy sets to allow the
researcher to assign partial membership to variables that are more continuous in nature,
such as patient age, rather than entire presence or absence. Because my data set includes
a fuzzy set condition, patient age, I use the fuzzy set / Qualitative Comparative Analysis
(fs/QCA) software.
2) Truth tables organize all possible combinations of variables in an experimental design
construct to seek empirical relationships. The total number of rows in a truth table will
equal 2
n
(where n = number of variables, including the outcome) since each variable has
two possible outcomes: 0 and 1. Rows that do not occur frequently enough in the studied
data set may be deleted and those that do not have great enough consistency in terms of
outcome are coded to 0. If after deleting and coding, the truth table rows contain
outcomes both equal to 0 and 1 that do not offset each other, the software yields a
solution – mathematical formulas that summarize which variables can or must exist and
in which combinations to achieve the outcome.
49
3) The absence of a variable or condition in a solution toward the outcome is indicated by
a tilde “~”. If a condition is not preceded by the tilde, it indicates its presence. Conditions
not explicitly included in a solution are deemed not essential.
4) Boolean addition and multiplication explain the logical existence and relationship
between variables to achieve an outcome. In the solution equation, addition (+) indicates
one condition OR the others must exist and multiplication (*) means one condition AND
the others must exist to achieve the solution.
5) Minimization is a mathematical approach to reduce the resulting solution formulas to
describe which conditions lead to the outcome in the fewest, simplest number of
combinations possible.
Data for fsQCA
I collected data for all health system patients discharged with a diagnosis of heart failure
between January and December 2010 to evaluate whether there was a pattern to
preventive interventions received and readmissions avoided. The data was a compilation
from one master report and manual documentation systems from all hospitals. The master
report pulled from three source systems (the electronic health record’s admission /
discharge / transfer module, an outside claims processing system, and the membership
system) and information provided manually about the heart failure transitional care
50
program’s enrollees. It provided patient demographics, admission and discharge dates,
readmission information, heart failure transitional care program enrollment, and
information about select interventions for any health plan member discharged from any
hospital (within or external to the company) with a primary or secondary diagnosis of
heart failure. The original source file contained over 50,000 patient discharges for the
calendar year 2010.
Each hospital maintained a manual Excel log regarding patients with heart failure. The
heart failure team tracked: patient name, patient medical record number, patient
admission and discharge date, if the patient chose to enroll in the heart failure program,
and select intervention data, such as when the home health nurse saw the patient, if the
pharmacist conducted the real-time medication reconciliation with the home health nurse,
and if the outpatient care manager contacted the patient within seven days of discharge.
The Excel logs varied slightly in format and content, although they all contained the key
patient and intervention data noted above. I integrated these into one master Excel log.
I integrated the master Excel log with the master report by creating a temporary lookup
field based on the medical record number and admission date and then merging each line
of data from the original master report with the master Excel log. I edited this merged
master report to exclude: 1) duplicate entries; 2) pediatric patients; 3) patients whose
identities could not be verified either by name or medical record number; 4) patients not
enrolled in the heart failure program who did not have a primary diagnosis of heart
51
failure; 5) patients admitted in 2010, but discharged in 2011; 6) patients who were
discharged from non-system hospitals (since they were often transferred to a system
hospital and thus double-counted); and 7) admissions during which the patient expired.
The resulting data set contained 10,532 heart failure patient discharges for the calendar
year 2010. I was able to obtain demographic data regarding most of the heart failure
patients in this study, including age, gender, race, primary language, discharge
disposition. Complete information was not available for 1,617 patients.
28% received the home health visit within 48 hours
6% had real-time medication reconciliation during the home health visit
31% had a physician visit with seven days of discharge
54% received a post-discharge follow-up call within seven days
61% were enrolled in the heart failure program
They had a median age of 74
I reduced the total data set to those discharges for which complete demographic data was
available. Of these 8,915 patient discharges included in the study:
90% were English speaking
93% were discharged home
38% received the home health visit within 48 hours
10% had real-time medication reconciliation during the home health visit
52
35% had a physician visit with seven days of discharge
65% received a post-discharge follow-up call within seven days
74% were enrolled in the heart failure program
They had a median age of 75
21% were readmitted for any cause within 30 days of discharge
The hospitals are geographically disbursed throughout five different counties in southern
California in neighborhoods that vary by median age, ethnicity, and socio-economic
status. A summary of the intervention reliability and personal characteristics for the 8,915
discharges by hospital is reflected in Table 4.
Table 4: Interventions Delivered and Patient Characteristics for 8,915 Discharges
There are 6,385 unique patients represented in the 8,915 discharges. Seventy-four percent
of the patients had only one 30-day readmission during the twelve month study period;
three percent had four or more readmissions, as indicated in Figure 2.
n
Home
Health
Visit
Real-Time
Med
Recon MD Visit
Follow-Up
Call
Median
Age
White
Race
Discharged
Home
Male
Gender
English
Language
Readmitted
in 30 Days
Hospital A 799 33% 25% 41% 76% 78 77% 95% 52% 93% 20%
Hospital B 769 29% 0% 33% 46% 71 33% 92% 57% 88% 17%
Hospital C 631 53% 8% 33% 71% 75 38% 92% 55% 83% 23%
Hospital D 1145 42% 5% 42% 51% 73 61% 93% 49% 93% 22%
Hospital E 706 37% 21% 30% 56% 74 30% 92% 58% 94% 25%
Hospital F 1155 36% 3% 29% 63% 70 42% 96% 64% 84% 20%
Hospital G 33 18% 0% 21% 27% 78 52% 73% 42% 100% 21%
Hospital H 595 38% 0% 46% 73% 77 63% 92% 54% 88% 21%
Hospital I 676 58% 42% 38% 77% 75 64% 92% 61% 94% 20%
Hospital J 1211 35% 0% 38% 70% 78 71% 91% 56% 93% 20%
Hospital K 613 26% 3% 25% 73% 76 21% 92% 50% 92% 21%
Hospital L 580 34% 9% 33% 72% 82 90% 93% 59% 94% 24%
53
Figure 2: Frequency Distribution of Number of 30-Day Readmissions per Patient
Study Approach
The analysis is based on the relationships of ten variables on the outcome of a patient
being readmitted within thirty days of hospital discharge based on findings from the
literature. First, I determined the inclusion criteria for the interventions. I did not include
discharge planning, physician / clinic appointment scheduled before discharge, or
discharge instructions as these were elements of usual care that all patients should have
received. I also excluded patient education which all patients were reported to receive and
0
1000
2000
3000
4000
5000
6000
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.
Number of Patients
Number of 30-Day Readmissions
Number of 30-Day Readmissions per Patient
January - December 2010
Mean 1.
Median 1.
Mode 1.
n 6385
Skewness 3.
Stdev 1.
Min 1.
Max 11.
54
for which I had no data. Therefore, I focused my study on four interventions all patients
in the heart failure program should have received to avoid a readmission.
HOMEHEALTH_IN_48HOURS – A patient received a home health visit within
48 hours of discharge.
RTMR_IN_ 48HOURS – A patient received a real-time medication reconciliation
during a home health visit within 48 hours of discharge.
MD_IN_7DAYS – The patient saw a primary care physician, cardiologist, or
cardiology nurse practitioner within seven days of discharge from the hospital.
CALL_IN_7DAYS – The patient had a follow-up phone conversation with a
heart failure care manager within seven days of discharge.
The remaining six variables represent patient-specific characteristics. In addition to
inclusion in the heart failure program, I included age, race, discharge disposition, gender,
and language as they were each studied in the literature as having a potential affect on the
success of certain interventions and/or being correlated to a readmission:
IN_HFPROGRAM – the patient is enrolled in the system’s heart failure program.
OLDER_AGE – I evaluated whether a person who is very old is more likely to be
affected by certain interventions. I used a fuzzy set calibration technique to
translate patient age into a range of values between 0 and 1 with 75 as the
55
midpoint since it is the median age of the study’s patient population (see Figure
3); thus OLDER_AGE = calibrate(age,85,75,40).
WHITE_RACE – the patient’s race is white.
DICHARGED_HOME – the patient was discharged to a community setting, i.e.
home versus a long-term care or rehabilitation facility.
MALE_GENDER – the patient is male.
ENGLISH_LANGUAGE – the patient’s primary language is English.
All variables followed a crisp set construct, except OLDER_AGE, which was a fuzzy set.
56
Figure 3: Frequency Distribution of Heart Failure Patient Ages
I use the (fs/QCA) software to conduct the analysis (Ragin, Drass, & Davey, fsQCA,
2009). I employ a crisp set analysis, the truth table algorithm, and standard rules when the
OLDER_AGE variable was not included in the Model. I deleted any rows with less than
1% of the study population as they were less likely to yield reliable results. I also coded
truth table rows that had consistency less than .75 to 0 because they are too inconsistent
(Ragin C. , User's guide to fuzzy-set/Qualitative Comparative Analysis 2.0, 2006).
Although the fs/QCA software suggests using .8 as a threshold, I conferred with a
0
50
100
150
200
250
300
350
21. 24. 27. 30. 33. 36. 39. 42. 45. 48. 51. 54. 57. 60. 63. 66. 69. 72. 75. 78. 81. 84. 87. 90. 93. 96. 99. 102. 105.
Number of Patients
Patient Age
Heart Failure Patients Age Frequency Distribution as of 12/31/2010
Mean 73.
Median 75.
Mode 80.
n 8915
Skewness -1.
Stdev 13.
Min 19.
Max 103.
57
methodologist who suggested, because my study encompassed a large “n,” I should begin
with .75 and test different consistency thresholds up to .9 or 1.0 to understand how the
results might vary; keeping in mind anything above .85 is more conservative. I employ a
similar approach for models that included OLDER_AGE, but used the fuzzy set truth
table algorithm to account for the causal condition of age.
I explore nearly every possible combination of interventions and patient characteristics to
understand the potential empirical relationship between the key heart failure interventions
provided to patients, patient characteristics, and whether patients were readmitted within
30 days. I study the resulting models at least twice in accordance with QCA protocol:
with the outcome variable set to an admission READMIT_IN_30 and a negated
admission ~READMIT_IN_30.
Limitations
My study has several limitations. First, it is based primarily on data review, provider
interviews, and limited observations and patient discussions in select hospitals. It does
not include extensive patient interviews.
Second, this study was limited to enrollees in a health system; therefore findings may not
necessarily extend to indigent persons who would not typically have access to regular
health care.
58
Third, I was unable to access patient-specific psycho-social data for the analytical portion
of my study, such as signs of depression, whether they live alone or with family, if they
have a history of not adhering to their prescribed diet or medication regimen, or need
financial aid. I also did not have access to patient-specific clinical information, such as
number of co-morbidities or NYHA heart failure scores to determine if outcomes vary by
disease progression.
Fourth, the data portion of my study only measures if an intervention was completed, not
the quality of each intervention. At the time of my study, there was no such metric for
intervention quality.
Fifth, the real-time medication reconciliation intervention in my data analysis was new to
the program and not being done with great fidelity. Only those patients who received the
entire medication reconciliation with both a home health nurse in the patient’s home and
a pharmacist on the telephone, per the program’s guidelines, were coded as having it. In
the period I studied, only 847 or the 8,915 patients received medication reconciliation
according to these parameters.
Sixth, the data I obtained was in part compiled from multiple Excel worksheets
maintained at each medical center. It is possible some data inaccuracies exist due to the
manual documentation and compilation process.
59
Chapter Four: Study Results
Introduction
I conducted a two-year study across a southern California hospital system to understand
the barriers and opportunities for reducing heart failure readmissions. The system had
adopted specific practices cited in the literature; yet after an initial period of readmission
reduction, results plateaued. The first part of my study is based on case studies of patient
care processes across nine hospitals, which includes provider interviews, process
observations, chart reviews and a limited number of patient and family interviews. The
second part entails a Fuzzy-Set / Qualitative Comparative Analysis (fsQCA) of 8,900
patient admissions related to heart failure to understand if there was an empirical
relationship between a patient’s personal characteristics, the interventions a patient
receives, and if a readmission occurs. I describe my study results in two parts. First, I
present the findings of my case studies. Second, I describe the results of the fsQCA
models. I conclude that delivering the same set of interventions to every patient does not
appear to be the optimal use of limited resources and the most efficacious delivery of
care.
60
Results of Case Studies
My case studies began with a process assessment in nine hospitals. I obtained the hospital
system’s heart failure scorecard that reported each hospital’s process reliability and 30-
day all cause readmission rates. I noticed wide disparities in both process reliability and
readmission rates (see Chapter 3 for details). The study’s sponsors sought to improve the
process consistency of each intervention that was in the defined heart failure (HF) bundle.
It was intended that each patient enrolled in the heart failure program would receive each
intervention and specific follow-up care for 180 days after a hospital admission. The
sponsors believed that if the interventions were delivered more reliably the readmission
rates would further decline. Thus, my case study was originally designed to identify
effective practices in the process design, organizational structure, and role definition to
consistently deliver the bundle elements.
I used the first hospital as a test site to confirm my study approach and tools, and improve
my knowledge of HF. This first hospital had served as the heart failure program’s pilot
site and had achieved a statistically lower 30-day readmission rate, but results were not
sustained for more than one year. I found the organizational leaders involved had shifted
their attention toward competing initiatives, so they were not vigilant about overseeing
the program metrics and continuing improvement. Individuals in key process roles
reported into different organizational departments that had different priorities which were
61
not aligned. Processes were not formally designed, but rather were sometimes carried out
to best complete the HF bundle items them quickly not to achieve the best outcome.
As I moved onto the next sites, I continued to observe similar structure, role, and process
challenges, but I also encountered a new issue: the clinicians were challenging whether
the bundle elements themselves were effective. This was clearly out of my scope, but I
was intrigued. One HF nurse reviewed her list of “frequent flier” patients with me – these
were patients who were readmitted to the hospital multiple times each year. She could
predict when these individuals would return to the hospital despite completion of the
prescribed HF bundle. One older lady lived by herself and her nearest family lived out of
town. She seemed to get lonely and would occasionally call 911 to come visit the
hospital. Another older lady who did not live alone returned regularly because she said
she was treated more kindly in the hospital than at home. A third patient did not qualify
for the palliative care program due to its strict guidelines, but was nearing end of life and
returned at least once per month. She felt helpless trying to prevent readmissions because
the bundle was not sufficient and she herself was not empowered to take the actions
necessary to help these patients. In another hospital, a clinical leader asked me if I truly
believed delivering the bundle elements would further reduce their readmission rates. I
honestly did not know. She said she felt her team was too focused on making their
numbers related to completing checklists than getting to know their patients and doing
what was needed to provide the best care.
62
I discovered similar findings in the remaining hospitals. I also observed discussions
between clinicians and patients or family members to gain a better appreciation for the
unique needs each patient may have and studied the results of extensive chart reviews
conducted by physicians, social workers and pharmacists of patients readmitted. From
these case studies, I developed six impressions that frame my study.
1. Processes were not consistently designed to achieve the desired outcomes. Each
of the nine hospitals followed different processes to achieve the same outcome. I
discovered certain hospitals had clearly defined critical to quality process steps,
which helped improve process reliability. For example, a reliable post discharge
appointment scheduling process started with the attending physician and patient
discussing the appointment purpose and agreeing to its importance before the
ward clerk would use the electronic health record (EHR) in the patient’s room to
discuss optimal days and times before scheduling (see Figure 4). At discharge, the
patient’s nurse would print the discharge instructions, which included all
appointments, and review them with the patient and family member(s) to confirm
they understood future appointments. I found that when providers did not discuss
the appointment with the patient and involve them in the scheduling process,
patients either canceled or did not show for their appointments 20-30% of the
time.
63
Figure 4: Post Discharge Appointment Scheduling Process
In concert with the process design, I found role design and the metrics by which
people were measured also may have an influence on process reliability and
perhaps patient outcomes. In most hospitals, the ward clerk who scheduled the
post-discharge appointment was only accountable for making sure an appointment
was scheduled not that the patient kept the appointment. It was unclear to me if
any one person other than the overall program sponsor was accountable for that
outcome. Conversely, one hospital designed a process by which each day they
would review which patients should have had a home health visit with medication
Ward Clerk
schedules MD
appointment via
EHR in patient’s
room
Ward Clerk
adds appt
to EHR
discharge
instructions
Ward clerk asks
patient for
appointment
preferred date and
time
Hospital MD writes
7-day MD follow-
up appt orders
MD explains
importance and
purpose of follow-
up appointment to
patient
Upon discharge, floor
nurse confirms
appointment time with
patient and family
During Home
Health visit, nurse
confirms patient
can make MD
follow-up
appointment
Includes order for
patient PCP
appointment within
7 days
1 – 2 days later
Schedule for time
and day that patient
is able to make
appointment
Discuss alternatives
in real-time with
patient if desired
date and time are
not available
Outpatient HF
nurse calls patient
4-5 days post
discharge;
confirms patient
can make appt
Can patient
make
appointment?
Can patient
make
appointment?
Y
Outpatient HF
nurse reviews
patient chart at 8
days post
discharge
Did patient
keep
appointment?
Y
Call patient to
understand reason
for not keeping
appointment and
offer to reschedule
Outpatient heart
failure nurse
continues
monitoring patient
health
Inform Pharmacist
during medication
reconciliation so
appointment can
be rescheduled
Outpatient heart
failure nurse
reschedules
appointment in
EHR
N N
Outpatient heart
failure nurse
reschedules
appointment in
EHR
End Y
N
Outpatient heart
failure nurse
reschedules
appointment in
EHR
Use EHR to directly book appointment and
rather than making patient call for appointment
Helps influence
patient to keep
appointment
Patient is
discharged; takes
discharge
instructions
= key process step
# = observed failure point
1 2 3 4 5 6
7 8 9 10
Post Discharge Appointment Scheduling Process
64
reconciliation. The home health supervisor and pharmacist would review the list
daily to ensure the pharmacist had received a call for each scheduled patient. The
home health supervisor would talk to each nurse who did not complete the
medication reconciliation to understand the barriers to completing the call to
remove them from occurring again.
2. Individual roles were often task-oriented. Each hospital designed a steering
committee team that as a whole was accountable for reducing readmissions, but
the individuals involved were typically each accountable for only very specific
tasks. For example, the inpatient heart failure nurse was accountable for enrolling
patients and providing their education, the ward clerk was accountable for
scheduling the physician appointment, the pharmacist was accountable for
conducting the real-time medication reconciliation, and the outpatient heart failure
nurse was accountable for contacting the patients within seven days of discharge.
Each provider usually had a checklist that reminded them of their specific task
accountability and helped tracked task. I observed that the level of specificity in
their work and the volume of tasks seemed to give rise to a checklist mentality to
get through their work. For example, I learned some hospitals screened for
depression because they learned through the literature it was a common driver of
readmissions. I asked one about their process and the outpatient heart failure
nurse explained that she mailed a depression screening form to all patients with
instructions to fill it out and return it. I thought perhaps patients suffering from
depression might not reliably respond to questionnaires and return them, so I
65
asked what percentage of screening forms was returned. She did not know
because she said she was only accountable for mailing them. Conversely, another
hospital offered patients they deemed high risk a multidisciplinary post-discharge
visit that included a social worker and a depression screening. They found six out
of 10 patients were suffering from depression; they referred those individuals for
behavioral health assistance.
3. A lot of documentation existed outside the electronic health record. The system I
studied had an integrated EHR across most of the care continuum. Providers
agreed it was a very beneficial communication and knowledge sharing vehicle
between hospital-based physicians and ambulatory physicians, and inpatient and
outpatient case managers. Yet I was surprised to discover how much information
and documentation resided outside the EHR. For example, nurses were supposed
to use a heart failure flow sheet within the EHR to track and compare educational
scores and completion of interventions over time. The quiz results were more
reliably found in progress notes or just on the paper quizzes directly. Some
inpatient heart failure case managers used an external pharmacy system to
manage the heart failure program patient follow-up schedules and to be
automatically notified when a patient enrolled in the program was rehospitalized
or presented in the Emergency Department. Other hospitals created their own
access database, Excel file, or paper binders to track when to follow up with each
patient and to set reminders about the topics on which to focus. I believe these
shadow documentation systems existed because the EHR was designed to replace
66
a paper-based charting system not to support disease management programs and
fell short in that regard.
4. Patients’ personal issues appear be a common driver to readmissions. Throughout
my assessment, I heard similar stories and anecdotes from physicians, nurses,
pharmacists and patient families – very specific issues related to the patient’s life,
belief system, and family structure were contributing to readmissions. For
example:
o I participated in a family meeting for an elderly heart failure patient. Her
well-educated, paramedic son explained that his mother was home at that
time receiving intravenous Lasix with a home health nurse while his sister
was in the kitchen cooking a pot of chicken soup: her mother’s favorite
meal. The family often prepared her meals, but was never educated on the
specific fluid and sodium restrictions. The son requested a physician direct
his mother to meet with a dietician because, he explained, she would
respect and adhere to a plan ordered by a physician, but not by her
children.
o A nursing director told me it was not until the seventeenth admission in
twelve months that they discovered a psycho-social issue was underlying
one of her patient’s repeated return.
o A physician leader conducted a chart review of over 2,000 patients that
were readmitted within 30 days. He found what he called “personal
67
issues” to be on par with clinical issues as the most common causes of
readmission.
This left me with the impression that perhaps we did not know our patients well
enough and did not perhaps tailor our care appropriately.
5. Each patient received the same bundle of interventions regardless of need. The
program was designed to deliver the same interventions each time a patient with
heart failure was admitted to the hospital: inpatient education, a home health visit
within 48 hours of discharge, real time medication reconciliation during the home
health visit, a follow-up call within seven days, and a physician visit within seven
days. The clinicians and heart failure team were measured and evaluated on the
percentage of patients who received each intervention in their scorecard, so they
had to complete them; but several clinicians asked me – if this bundle did not
work for this patient last time, why do we think it will this time? I developed a
theory that perhaps the same bundle should not be used for every patient and that
it was missing elements necessary to address more complex clinical needs, such
as chronic kidney disease, as well as psycho-social issues, such as patients who
live alone or who cannot care for themselves.
6. The program was designed to treat a disease. The heart failure program was
designed to reduce readmissions for a specific cohort of patients with heart
failure; it was not intended to treat other conditions. Yet heart failure patients tend
to have other health issues, such as chronic kidney disease or depression. For
68
example, I learned some patients were receiving conflicting guidance from their
cardiologist and nephrologist about medications, fluid intake, and under which
conditions to return to the emergency department. To counter this, one hospital
redesigned its outpatient case manager role from each specializing in a specific
disease and managing all patients, they now each support a cohort of patients and
help coordinate care for all of their chronic conditions across the care continuum.
Results of Fuzzy Set Qualitative Comparative Analysis
I utilized the fs/QCA software to illuminate which causal conditions would lead to a
person not being readmitted. I began my data analysis anticipating that in aggregate, the
interventions noted in the literature would result in a person being less likely to be
readmitted. I also anticipated that once I added additional personal characteristics, such as
age or the venue to which a person was discharged, the analysis would yield further
clarification about which combinations of personal factors must exist for these
interventions to be effective.
Model 1
I tested my hypothesis with an initial model containing all ten conditions available in my
data set and the outcome set to a positive readmission to understand if any combinations
of conditions consistently led to a readmission. I deleted truth table rows that did not
69
occur at least 1% (80 cases) and examined the resulting truth table. I found four things of
interest:
1. The consistency of each row ranged from 14% to 29%, which indicated I would
not likely find any combinations of conditions that commonly resulted in a
readmission.
2. Every truth table row contained a “0” for the real-time medication reconciliation.
I knew only 10% of the patients in my dataset had received this intervention as it
was newer and hospitals were still struggling to perform consistently. Yet I was
surprised the cases that included real time medication reconciliation did not occur
frequently enough in this truth table to avoid deletion. I wondered if the results
might be different if it had been performed with greater fidelity.
3. All truth table rows except two contained a “1” for a patient being discharged
home. As with the medication reconciliation, patients who were not discharged
home represent a small percentage of my dataset; therefore I may need to simplify
or stratify the data to get meaningful results about how differing interventions
may benefit patients discharged to a community versus other settings.
4. Every truth table row also contained a “1” for the English as a primary language
variable, again suggesting I might need to stratify the data to better understand
how results vary by patient’s primary language.
70
The truth table rows all contained an outcome of “0” indicating they did not consistently
result in a readmission. Therefore, I re-ran this model with the outcome set to a negated
readmission.
The QCA software generates three solutions: complex, parsimonious, and intermediate.
The complex solution does not include counterfactual cases, meaning any possible
combinations of the truth table that did not exist in the actual data set are set to a 0
outcome. The parsimonious solution is based on the complex solution, but uses Boolean
techniques to simplify it to make it easier to understand which conditions are necessary to
achieve the outcome. The intermediate solution varies from the first two in that it will
include counterfactual cases in the solution set, if an outcome may be easily derived from
cases that do exist in the truth table (Ragin C. , User's guide to fuzzy-set/Qualitative
Comparative Analysis 2.0, 2006). Therefore it is a means of predicting what might occur
in hypothetical cases. I focus my analysis primarily on the parsimonious solution, but I
also review and consider the other solution types in accordance with the methodologist’s
guidance.
Again I deleted truth table rows that contained fewer than 80 patients. All rows had
consistency of 70% or higher, indicating most combinations of variables consistently
resulted in a patient not being readmitted. A methodologist I consulted suggested I
experiment with different consistency thresholds to determine if it would yield different
results. I re-ran the analysis with the consistency set to 75%, 80%, and 85%.
71
With 75% consistency, the model indicated the following combinations of conditions and
yielded 91% coverage:
md_in_7days = a physician visit
older_age*~homehealth_in_48 = older patients not having a home health visit
~older_age*call_in_7days+in_hfprogram = younger patients either receiving a
follow-up call or being enrolled in the heart failure program
male_gender*~white_race = non=white males
~male_gender*(homehealth_in_48+older_age+white_race+call_in_7days
in_hfprogram) = women either having a home health visit or older or white race
or receiving a follow-up call
I re-ran the model with 80% consistency and was surprised to discover differing solutions
with greater specificity. Furthermore as Ragin had indicated would occur, with this
higher consistency threshold, the coverage fell to 55% of all patients:
~homehealth_in_48*md_in_7days = no home health visit and a physician visit
male_gender*(~discharged_home+md_in_7days) = men either not discharged
home or with a physician visit
~white_race*male_gender*homehealth_in_48 = non-white men with a home
health visit
72
white_race*~older_age*~male_gender*homehealth_in_48 = younger, white,
women with a home health visit
(white_race+older_age)*~male_gender*discharged_home*(~in_hfprogram+~call
_in_7days) = white or older women discharged home and either not in the heart
failure program or not receiving a follow-up call
older_age*~white_race*male_gender*(call_in_7days+in_hfprogram) = older,
non-white men either receiving a follow-up call or enrolled in the heart failure
program
~older_age*white_race*male_gender*~homehealth_in_48*(call_in_7days+in_hf
program) = younger, white men who did not have a home health visit but either
received a follow-up call or were enrolled in the heart failure program
With an 85% threshold, only one truth table row consistently resulted in a non-
readmission: non-white women who saw a physician. This solution accounted for less
than 7% of all patients.
In discussing these results with a methodologist, I validated Ragin’s point that QCA
becomes more complex with a larger number of variables (Ragin C. C., 1987). I thought
as a second step I should pare down the number of variables to determine if I would find
more meaningful results with less complexity. I began constructing a table of models (see
Figure 5) and analyzed each with the outcome set to a readmission not occurring. I used
73
three consistency thresholds to assess each model, 75%, 80%, and 85%, to understand if
the results would vary by constancy (see the Appendix for detailed truth table output).
Figure 5: Qualitative Comparative Analysis Models
MODEL # HOMEHEALT
H_IN_48
RTMR_IN_
48HOUR
MD_IN
_7DAYS
CALL_IN
_7DAYS
IN_
HFPROGRAM
OLDER_
AGE
WHITE_
RACE
DISCHARGED_
HOME
MALE_
GENDER
ENGLISH_
LANGUAGE
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
VARIABLES PER SCENARIO
74
Models 2 - 11
I continued by examining each of the ten conditions independently against the non-
readmission outcomes. I started with the four intervention conditions: home health, real
time medication reconciliation, MD visit, and follow-up call and summarized results in
Figure 6. Although these results do not indicate a direct cause and effect relationship
between the intervention or individual characteristic and the non-readmission, I wanted to
explore why the literature sometimes conflicted.
MODEL # HOMEHEALT RTMR_IN_ MD_IN CALL_IN IN_ OLDER_ WHITE_ DISCHARGED_ MALE_ ENGLISH_
42
43
44
45
46
47
48
49
50
51
52
53
54 (non-
English only) X
55
(discharged
home only)
X
56 (by
hospital)
VARIABLES PER SCENARIO
75
Figure 6: Results of Models 2 – 11
At this highest level, I expected that patients might be less likely to return to the hospital
if they had these interventions, because intuitively it seems if people get more care they
should be healthier; this was not always the case. For example, patients who did not
receive a home health visit were 1% less likely to be readmitted. Ragin explains this does
not necessarily indicate a cause and effect, but rather just the relationship between the
lack of a home health visit and a non-readmission exists. Such results require further
75% CONSISTENCY 80% CONSISTENCY 85-90% CONSISTENCY
#2: Home Health only none No home health visit
(only at 79% consistency, 62% coverage)
none
#3: Medication
Reconciliation (RTMR) only
none RTMR
(only at 81% consistency, 10% coverage)
none
#4: MD visit only none An MD visit
(36% coverage)
none
#5: Call only none A follow-up call
(66% coverage)
none
#6: In heart failure (HF)
program only
none In HF program
(75% coverage)
none
#7: Older age only none Younger age
(only at 79% consistency, 43% coverage)
none
#8: White race only none none none
#9: Discharged home only none Not discharged home
(7.5% coverage)
none
#9: Male gender only none Male gender
(only at 79% consistency, 56% coverage)
none
#10: English language only none English speaking
(only at 79% consistency, 91% coverage)
none
MODEL # AND CONDITIONS
INCLUDED
PARSIMONIOUS RESULTS: SOLUTION PATHWAYS LEADING TO A NON-READMISSION
76
exploration across multiple cases to determine if the lack of a home health visit is truly
connected to a non-readmission or merely a coincidence (Ragin C. , Redesigning social
inquiry: fuzzy sets and beyond, 2008). Conversely, patients who had the medication
reconciliation, a physician visit, or a follow-up call were 2.5%, 3.5%, and 2.5% less
likely to be readmitted, respectively, although the coverage for medication reconciliation
and the physician visit was relatively small.
I next examined patients who participated in the heart failure program and found they
were 3.5% less likely to be readmitted at 80% consistency, which covered 75% of the
patient population. This suggests participation in the program has a stronger connection
with a non-readmission than each of the individual intervention conditions. I found this to
be a logical result because patients in the heart failure program are supposed to get a
greater number of the interventions. I consulted one of my doctoral committee members
about these initial results and he suggested I exclude the heart failure program condition
from future models since in essence it was somewhat redundant.
Next I evaluated the five personal conditions: age, race, discharge location, gender and
language. Again none of these conditions yielded results at 75% or 85% consistency. The
condition that had the greatest association with a non-readmission was not being
discharged home: patients not discharged home were 3.5% less likely to be readmitted.
However, this population accounted for only 7% of my entire dataset. Three conditions:
younger age, male, and English speaking patients were each less than one percent less
77
likely to be readmitted than other patients, and race had virtually no difference. I was
excited by these preliminary results and the differing outcomes when compared with
model 1. It supported my initial thinking that every intervention may not be beneficial to
every patient. To further explore my theory, I studied all combinations of the four
intervention conditions.
Models 12 - 18
I continued my analysis by evaluating models that contained two or more of the four
intervention variables to understand if patients who received the interventions had
differing outcomes. These results were somewhat similar with the first set of models as
indicated in Figure 7.
The results again indicated the relative importance of a physician appointment,
medication reconciliation, and follow-up call to avoiding readmissions as well as the
potential unimportance of a home health visit to this population. However, in model 18
which included all four interventions, the presence of a home health visit with the
medication reconciliation and physician visit emerged as a solution with 85%
consistency, but only for a very limited population. These results made me wonder if
perhaps home health visits as well as the other interventions might benefit some patients
so I further examined all four interventions in concert with all possible combinations of
the five personal conditions.
78
Figure 7: Results of Models 12 – 18
Models 19 - 49
These models include all four interventions, plus all possible combinations of patient
characteristics. I structured these to determine according to which patient conditions each
intervention may most favorably lead to a non-readmission. My intent was to clarify the
consistency with which previous cases led to solution. Because a majority of the patients
in my study were enrolled in the heart failure program, I excluded that variable to
simplify the analysis.
75% CONSISTENCY 80% CONSISTENCY 85-90% CONSISTENCY
#12: Home Health and RTMR none RTMR
(10% coverage)
none
#13: Home Health and MD
visit
none An MD visit
(36% coverage)
none
#14: Home Health and Call No home health visit
Having a follow-up call
(93% coverage)
No home health visit and a follow-up call
(35% coverage)
none
#15: MD visit and call No MD visit
A follow-up call
(92% coverage)
An MD visit
(36% coverage)
none
#16: Home Health, RTMR, MD
visit
none No home health and RTMR or MD visit
RTMR and MD visit
(25% coverage)
none
#17: Home Health, RTMR,
call
No home health visit
Having a follow-up call
(93% coverage)
RTMR
No home health visit and a follow-up call
(42% coverage)
none
#18: Home Health, RTMR, MD
visit, call
No home health visit
MD visit
Call
(95% coverage)
No home health and RTMR or MD visit
RTMR and MD visit
(25% coverage)
Home health and RTMR
and MD visit (3%
coverage)
MODEL # AND CONDITIONS
INCLUDED
PARSIMONIOUS RESULTS: SOLUTION PATHWAYS LEADING TO A NON-READMISSION
79
I again observed a reduction in the solution coverage with greater consistency. In each
model at 75% consistency, the coverage ranged from 85 – 96%, which indicated the
solutions had greater importance. With 80% consistency, the coverage fell to a range of
27 – 50% and with 85% consistency coverage was only 2 – 6%. Therefore, I focused my
analysis on solutions with 75% - 80% consistency.
As indicated in Figure 8, the first ten models each include five conditions and yield
similar results at 75% and 85% consistency. At 75%, no home health visit, an MD visit,
and a call appear are conditions in many solution sets. At 85%, real-time medication
reconciliation in conjunction with an MD visit is the sole solution. In some models, home
health is another necessary condition and in others the solution is narrowed to only men
or younger patients. At 80% consistency, the solutions become more complicated. As I
added and changed conditions to each model, the granularity increased and morphed. I
continued to observe similar themes with the association of the MD visit, the call, the real
time medication, no home health visit, and not being discharged home with a non-
readmission, but the personal conditions associated with these interventions varied from
model to model, and in some instances the converse of these associations was true. I also
noted younger age became a more common component of certain solutions, as did race
and gender. I found this interesting since they did not prove to be important conditions
when I examined them individually.
80
Figure 8: Results of Models 19 – 48
75% CONSISTENCY 80% CONSISTENCY 85-90% CONSISTENCY
#19: Home Health, RTMR,
MD visit, call, age
No home health
MD visit
Call
(95% coverage)
MD visit and RTMR or no home health
MD visit and call and older age
Younger age and home health and call and
no RTMR and no MD visit
(36% coverage)
RTMR and MD visit
#20: Home Health, RTMR,
MD visit, call, white race
No home health
MD visit
Call
(95% coverage)
RTMR and MD visit
No home health and MD visit and (non-
white or call)
Non-white and home health and call
and no RTMR
(32% coverage)
RTMR and MD visit
(4.5% coverage)
#21: Home Health, RTMR,
MD visit, call, discharged
home
No home health
MD visit
Call
(95% coverage)
Discharged home
No home health and RTMR or MD visit
RTMR and MD visit
(82% coverage)
RTMR and MD visit and home
health visit
(3% consistency)
#22: Home Health, RTMR,
MD visit, call, male gender
No home health
MD visit
Men
No RTMR and a call
(96% coverage)
MD visit and no home health or call or
RTMR
(27% coverage)
Men with RTMR and MD visit
(3% coverage)
#23: Home Health, RTMR,
MD visit, call, english
language
MD visit
Call
No home health and English
language
(93% coverage)
No home health and RTMR
MD and (Non-English or no home health or
RTMR)
(27% coverage)
RTMR and MD visit and home
health visit
(3% coverage)
#24: Home Health, RTMR,
MD visit, call, age, race
No home health
Older non white
Call and older age or white or
no RTMR
RTMR and white or older
(93% coverage)
Non-white and MD visit
Younger and MD visit and no home health
Older and MD visit and call
Younger non-white and home health and
call and no RTMR
(32% coverage)
none
#25: Home Health, RTMR,
MD visit, age, discharged
home
No home health
MD visit
Men
(95% coverage)
Not discharged home
MD visit and (no home health or RTMR)
Older age and MD visit and call
Younger and home health and call and no
RTMR and no MD visit
(43% coverage)
Younger age with RTMR and MD
visit
(2% coverage)
#26: Home Health, RTMR,
MD visit, age, male gender
No home health
No RTMR and a call
Younger and RTMR or call
Men and RTMR or call or older
age
(93% coverage)
Younger or women with no home health and
MD visit
Men with MD visit and (call or younger age)
Younger men with a call and no RTMR
Younger and MD visit and no call
(36% coverage)
RTMR and MD visit
(5% coverage)
#27: Home Health, RTMR,
MD visit, age, english
language
Call
No home health and English
language
(91% coverage)
MD visit and (RTMR or call or older age)
Younger age with no home health and MD
visit
Younger age with home health and no RTMR
and no MD visit and a call
(30% coverage)
Younger age with RTMR and MD
visit
(2% coverage)
#28: Home Health, RTMR,
MD visit, white race,
discharged home
Not discharged_home
MD visit
Call
No home health and white race
(87% coverage)
MD visit and (no home health or RTMR)
White and not discharged home
Non-white and home health and no RTMR
and call
(39% coverage)
RTMR and MD visit
(5% coverage)
MODEL # AND CONDITIONS
INCLUDED
PARSIMONIOUS RESULTS: SOLUTION PATHWAYS LEADING TO A NON-READMISSION
81
75% CONSISTENCY 80% CONSISTENCY 85-90% CONSISTENCY
#29: Home Health, RTMR,
MD visit, white race, male
No home health
MD visit
Call and non-white race
Men with RTMR or white with
no call
Women with a call and no
RTMR
(92% coverage)
Men and RTMR
Non-white women with MD visit
Men with call and non-white or MD visit
White women with a call and no home
health
MD visit and no home health and (women
or a call)
(44% coverage)
RTMR and MD visit
(5% coverage)
#30: Home Health, RTMR,
MD visit, white race, english
language
MD visit
RTMR and call
No home health and English
language
White race and RTMR or call
(92% coverage)
Home health and non-English language
MD visit and no home health or RTMR
Non-white English language and no RTMR
and no MD visit and call
(40% coverage)
RTMR and MD visit
(4% coverage)
#31: Home Health, RTMR,
MD visit, call, discharge
home, male
No home health
MD visit
Men
(96% coverage)
Men not discharged home
RTMR and MD visit
Women with MD visit and no home health
Men with MD visit and call
(31% coverage)
Men with RTMR and MD visit
(3% coverage)
#32: Home Health, RTMR,
MD visit, call, discharged
home, english language
MD visit
Call
No home health and English
language
(93% coverage)
Not discharged home
No home health and RTMR
MD visit and non-English language
No home health and MD visit
(33% coverage)
RTMR and MD visit and home
health visit
(3% coverage)
#33: Home Health, RTMR,
MD visit, call, male gender,
english language
MD visit
Women with no home health
No RTMR and a call
Men and English language
(95% coverage)
RTMR and MD visit
Women and no home health and MD visit
Men and MD visit and call
(27% coverage)
RTMR and MD visit
(5% coverage)
#34: Home Health, RTMR,
MD visit, call, older age,
white race, discharged
home
Non-white race
Younger age
No home health
RTMR
MD visit
(93% coverage)
Not discharged home and younger or white
MD and non-white
Younger and MD visit and no home health
Older and MD visit and call
Younger non-white and home health and no
RTMR and a call
(38% coverage)
none
#35: Home Health, RTMR,
MD visit, call, older age,
white race, male gender
Younger or white women with
no home health
Non-white and call
White men with no call
Older men and an MD visit
Older women with no MD visit
and a call
Older women and younger men
with a call
(89% coverage)
MD and no home health or men
Non-white men and call
Younger white women and no call
Younger men with a call and no home
health
Younger white women with home health adn
no MD visit
(45% coverage)
Younger, non-white women with
no home health and an MD visit
(2% coverage)
#36: Home Health, RTMR,
MD visit, call, older age,
white race, english
language
No home health
Call
(93% coverage)
MD visit and (younger and no home health)
or (older and home health)
No MD and call and non-white and younger
or no home health
White race and no home health and MD
visit and call
(36% coverage)
none
#37: Home Health, RTMR,
MD visit, call, older age,
discharged home, male
gender
No home health
No RTMR and call
Younger and RTMR or call
Men of older age or RTMR or
call
(93% coverage)
Older or men not discharged home
Women with MD visit or no home health
Younger men and home health and no RTMR
and a call
Men and MD visit and call
Younger age and MD visit and (no call or no
home health or men)
(37% coverage)
RTMR and MD visit
(5% coverage)
MODEL # AND CONDITIONS
INCLUDED
PARSIMONIOUS RESULTS: SOLUTION PATHWAYS LEADING TO A NON-READMISSION
82
75% CONSISTENCY 80% CONSISTENCY 85-90% CONSISTENCY
#38: Home Health, RTMR,
MD visit, call, older age,
discharged home, english
language
No home health
Call
(93% coverage)
Not discharged home
MD visit and (RTMR or call and older
age)
Younger age and no home health and
MD visit
Younger age and home health and no
RTMR and no MD visit and call
(37% coverage)
Younger age with RTMR and MD visit
(2% coverage)
#39: Home Health, RTMR,
MD visit, call, older age,
male gender, english
speaking
No home health
Men and call or older age
RTMR and call and no MD
visit or older age
(91% coverage)
Men and call and younger age or MD
visit
Md visit and younger age and no call or
men
No home health and MD visit and
younger or women
Home health and no MD visit and
younger and women or call
(41% coverage)
none
#40: Home Health, RTMR,
MD visit, call, white race,
discharged home, male
gender
Not discharged home
MD visit
home health and women or
call or white race
Call and non-white race or
women with no RTMR
Men with RTMR or no call
and white race
(87% coverage)
Not discharged home and (white or
male)
Men and RTMR
Women and no home health and MD
visit
Men and call and (non-white or an MD
visit )
Non-white race and MD visit and
(women or home health or call)
(45% coverage)
RTMR and MD visit
(5% coverage)
#41: Home Health, RTMR,
MD visit, call, white race,
discharged home, english
language
MD visit
No RTMR and call
No home health and
English language
White race and RTMR or
call
(91% coverage)
MD and no home health or RTMR
White and not discharged home
Home health and no RTMR and no MD
visit and call and non-white
Non-English and home health or MD
visit
(36% coverage)
RTMR and MD visit
(6% coverage)
#42: Home Health, RTMR,
MD visit, call, white race,
male gender, english
language
No home health
No RTMR and call
White men
(90% coverage)
RTMR and male
Non-white women and MD visit
MD and call and no home health
White women with no call and no home
health
Men and call and (no MD and non-
white) or (MD visit and white race)
(42% coverage)
RTMR and MD visit
(5% coverage)
#43: Home Health, RTMR,
MD visit, call, discharged
home, male gender, english
language
No home health
MD visit
Men
Call and no RTMR
(96% coverage
Not discharged home
RTMR and MD visit
MD and women and (no home health or
call)
(34% coverage)
RTMR and MD visit
(5% coverage)
#44: Home Health, RTMR,
MD visit, call, older age,
white race, discharged
home, male gender
Older women
White and no call
Non-white and call
Men and MD visit
Younger and no MD visit
and (call or white race)
No home health and call or
white race
Younger age and home
health and no MD visit
(85% coverage)
Not discharged home
Non-white male and MD visit
No home health and MD visit
Non-white men and home health
Younger white women with no call
Older non-white men with a call
Younger white women and a home
health visit and no MD visit
(50% coverage)
Younger, non-white women with no home
health and a MD visit
(2% coverage)
MODEL # AND CONDITIONS
INCLUDED
PARSIMONIOUS RESULTS: SOLUTION PATHWAYS LEADING TO A NON-READMISSION
83
75% CONSISTENCY 80% CONSISTENCY 85-90% CONSISTENCY
#45: Home Health, RTMR,
MD visit, call, older age,
white race, discharged
home, english language
Call
Younger or white race and
no home health
(86% coverage)
Not discharged hone
Younger and no home health and MD
visit
White and no home health and MD and
call
Older and home health and MD visit
Younger, non-white and home health
and call and no MD visit
Older white race with MD and call
(35%)
none
#46: Home Health, RTMR,
MD visit, call, older age,
white race, male gender,
english language
Women or white race with
no home health
Call or non-white race
Younger with MD visit or
men
MD visit and older age or
men
Women with call and no
MD visit or older age
(90% coverage)
MD and no call
Female and no home health and MD
visit
Non-white men and no home health and
call and no MD visit
Non-white younger men and home
health
White men and (no home health and
call and younger) or MD visit
White younger women and no call or
(home health and MD visit)
(40% coverage)
Non-white women with an MD visit
(7% coverage)
#47: Home Health, RTMR,
MD visit, call, older age,
discharged home, male
gender, english language
No home health
Men and call or older age
No RTMR and call and (no
MD visit or older age)
(90% coverage)
Not discharged home
Younger women and MD visit and no
home health
Younger age and home health and call
and no MD visit
Younger age and MD visit and no call
Men and MD visit and younger age or
call
(82% coverage)
none
#48: Home Health, RTMR,
MD visit, call, white race,
discharged home, male
gender, english language
No home health visit
MD visit
Call and non-white race
Men and RTMR
Women and no RTMR and a
call
White men with no call
(92% coverage)
Men not discharged home and RTMR
Men and call and (MD visit and white
race) or )(no MD visit and non-white)
Non-white women and MD visit
MD visit and call and no home health
White women and no home health and
no call
(46% coverage)
RTMR and MD visit
(4% coverage)
MODEL # AND CONDITIONS
INCLUDED
PARSIMONIOUS RESULTS: SOLUTION PATHWAYS LEADING TO A NON-READMISSION
84
Models 49 - 53
I conducted a final examination of the intervention conditions by independently
examining each against all five personal characteristics. Common themes continued to
emerge, including not being discharged home, having a physician appointment and not
having a home health visit. Yet the patients’ personal characteristics added considerable
specificity to the solution sets and the coverage on the 75% consistency (see Figure 9).
Figure 9: Results of Models 49 – 53
75% CONSISTENCY 80% CONSISTENCY 85-90% CONSISTENCY
#49: Home Health and 5
personal
none No home health for some
Discharged home + English speaking
Young and English speaking
none
#50: RTMR and 5 personal none Men not discharged home or having a RTMR
or older non-English speaking
(12% coverage)
none
#51: MD visit and 5 personal Older age
Men
English language
(98% coverage)
Not discharged home
MD visit and non-white women or younger
age
MD visit, older age, and women or white
race
MD visit and white or young men
(36% coverage)
none
#52: Call and 5 personal Women
Younger age
Follow-up call
White race
(96% coverage)
Men non-English speaking or not discharged
home
Younger men with a follow-up call
(29% coverage)
Older men not English
language
(3% coverage)
#53: All 10 conditions except
HF program participation
Women
Younger age
MD visit
Non-white and call
White and no home
health
(93% coverage)
Not discharged home
Women with no home health and MD visit
White men and MD visit
White younger women with no call
Younger white women or non-white men
with home health
Older non-white or younger white men with
no home health and with a call
(43% consistency)
MD visit and non-white
women (7%)
MODEL # AND CONDITIONS
INCLUDED
PARSIMONIOUS RESULTS: SOLUTION PATHWAYS LEADING TO A NON-READMISSION
85
I ran one last model with all conditions except participation in the heart failure program
to understand how exclusion of that variable may affect the results. Again, the similar
themes of a physician visit emerged at 75% consistency, and the model yielded the exact
same results at 85% consistency. Otherwise, the details varied considerably. At 75%
younger patients and women each appeared in the solution. At 80% patients not
discharged home as well as multiple combinations of specific patient groups with varying
interventions appeared.
These results support by initial theory and themes I read in the literature: not every
patient may benefit from the exact same treatment. Care needs to be tailored to the
specific individual.
From this analysis, I observed several themes. First, not being discharged home is more
related to a non-readmission than any other variable. This is consistent with the literature,
which indicates patients who are discharged to a non-community setting typically have
lower readmission rates. Second, the physician visit within seven days of discharge is the
intervention most associated with a pathway to a non-readmission; even so, for certain
patient groups in some models the lack of a physician visit led to a non-readmission. This
too is consistent with the literature in that some studies have found this intervention aids
in reducing readmissions while others have found the converse (Bakhtiari, 2010). Third,
receipt of a home health visit within 48 hours did not appear in many solution pathways.
In fact, the lack of a home health visit was more commonly reflected as a pathway to a
86
non-readmission. I was surprised by this result as the intervention was intended to
provide a bridge in the care transition before the patient sees their physician visit. I do not
know if the visit itself is not of value, if the visit was not being done reliably enough to
yield better results, if perhaps the quality of the intervention was not sufficient, or if this
is a mere coincidence. Fourth, language appears to be an important factor in
understanding which interventions benefit patients most. English speaking patients
appear to benefit more from certain interventions. Also in many models’ intermediate
solutions, English was a consistent condition present in most or all solutions. This does
not imply that patients who are not English speaking have a higher likelihood of
readmission, which is what some literature suggests. On the contrary, when I examined
readmission rates on the basis of language alone, both cohorts had roughly a 21%
readmission rate.
Based on the findings of the literature review and the previous models, I wanted to
further explore two distinct groups: 1) patients whose primary language is not English
and 2) patients who are discharged home. I stratified the data set and ran two additional
models. These results are in Figure 10.
87
Figure 10: Results of Models 54 – 55
Model 54 – Stratification by Language
I stratified my dataset to include only non-English speaking patients (n=833). I included
all conditions except language. I observed the continuing themes related to not being
discharged home and a physician visit at 75% consistency and not being discharged home
continued to be associated with a non-readmission at 80% and 85%. This was the
strongest association of not being discharged home I observed in any of the models. I also
found it interesting at all consistency levels specific patient groups, especially older
patients, who had a home health visit were associated with a non-readmission; older men
with a home health visit at 85% consistency accounted for 11% of the 40% total
coverage. Overall, this model had one of the strongest solution sets at 85% consistency.
75% CONSISTENCY 80% CONSISTENCY 85-90% CONSISTENCY
#54: Stratified to only
include non-English
speaking patients. All other
conditions
Not discharged home
MD visit
Older white patients
Non-white men in the heart
failure program
Older patients with a home
health visit and a follow-up
call or heart failure program
Older women with no home
health and either no call or no
heart failure program
(77% coverage)
Not discharged home
MD visit for younger age or those
without a home health visit
White women
Men with a home health visit
Older non-white men with a call or
enrolled in the HF program
(60% coverage)
Not discharged home
Women of white race or
younger with an MD visit
Younger women with no
home health but with an MD
visit
Older men and a home
health visit
(40% coverage)
#55: Stratified to only
include patients
discharged home. All other
conditions.
Women
Younger age
MD visit
Non-white and follow-up call
White race and no home
health visit
(93% coverage)
Women with an MD visit and no
home health
White men with an MD visit
Younger non-white men with home
health
Younger white women with a home
health visit or no call
Older non-white or younger non-
white men with a call and no home
health
(39% coverage)
Non-white women with an
MD visit
(7% coverage)
MODEL # AND CONDITIONS
INCLUDED
RESULTS: SOLUTION PATHWAYS LEADING TO A NON-READMISSION
88
These results continued to support my theory that care needs to be individualized to the
patient and that for patients whose primary language is not English; they may have
unique cultural needs. This subject might be an opportunity for further research to
understand which interventions may be most effective by primary language or ethnicity.
Model 55 – Discharge Setting
I stratified the data set to only those patients discharged home (n=8269). At both 75%
consistency, the parsimonious solution contained five solution pathways that led to a non-
readmission with 93% coverage: women, younger age, having a physician visit, being
non-white and having follow-up call, and being white but not having a home health visit.
At 80% consistency, the results were similar with MD visit and home health being
associated with non-readmissions for some populations. At 85% consistency, the sole
parsimonious solution path indicated non-white women who see a physician are less
consistently readmitted, but this only covered 7% of the population. This was the
identical result as model 1, which included the entire dataset. The primary difference I
observed between this model and other models was the frequency with which the
younger age condition and female gender condition were present. This too may be an
opportunity for further analysis.
Model 56 – Stratification by Hospital
In talking with people in different hospitals, they each believed their patient population
varied from the other hospitals either by socio-economic status, education, and/or co-
89
morbidities. The literature is now also discovering a communities infrastructure to help
its citizens, the relative wealth or poverty, and prevalence of mental illness may also be
drivers of readmission (Joynt & Jha, 2012). I was unable to obtain that data by patient, so
as a proxy I re-ran Model 53 containing all nine variables except inclusion in the heart
failure program stratified by hospital. I found far greater coverage levels and consistency
perhaps due to the smaller dataset. I obtained solutions at 85% and 90% for all models, so
I focused my analysis on those as reflected in Figure 11. I found it interesting that
different hospitals had very different solutions. For example, not discharged home only
appeared as an independent solution at both 85% and 90% for hospital A, while women
not discharged home was a solution for hospital I; medication reconciliation appeared in
several solutions for hospital J and non-English speaking women were included in
hospital H. I also observed that English was a consistent condition in the Intermediate
solution for each hospital.
90
Figure 11: Results of Model 56: Stratified by Hospital
85% CONSISTENCY 90% CONSISTENCY
Hospital A Not discharged home
Non-white
Younger women with a follow-up call
Women with home health, RTMR, and an MD visit
Men with no MD visit, no home health, and RTMR
Men with an MD visit, a call, and no RTMR
Older men with no MD visit and no call
(47% coverage)
Not discharged home
Younger women with no follow-up call
Men with RTMR and no home health or MD visit
Older men with a home health visit and an MD
visit and no RTMR
Non-white and women or older or MD visit or
follow-up call
(32% coverage)
Hospital B Younger and white race
Call and (women with home health) or (older patients
with no home health) or (men with no call and an MD
visit)
White women discharged home with no home health
No home health and no MD visit and (men with a call)
or (younger English speaking women with no call)
White or younger patients with a home health and MD
visit
(53% coverage)
Younger white patients with either no call or
women with an MD visit
Older or white women with a call
Younger or white men with an MD visit and no
call
Older white patients with no MD visit and a call
White women discharged home with no home
health visit and no MD visit
(29% coverage)
Hospital C White men with an MD visit
Non-white women with no home health and an MD
visit
Younger white women with home health and no MD
visit
Non-white or younger men not discharged home
Older white patients with home health and no call
Older white or male patients with no home health
and an MD visit
Older white patients with no home health and a call
and either no MD visit or male gender
(31% coverage)
White men with an MD visit
Non-white women with no home health and an
MD visit
Older non-white or male patients with no home
health and an MD visit
(17% coverage)
Hospital D Non-white women with MD visit and home health or
no call
White women with a call and no home health and an
MD visit or younger age
Younger white men with an MD visit and a home
health visit or no call
Younger non-white men with no call and no MD visit
or no home health visit
Non-white men with home health or an MD visit
No home health and an MD visit for older women or
(non-white patients or older men with no call)
(29% coverage)
White women with no home health and MD visit
and call
Home health and no MD visit and no call and
non-white or younger men
(8% coverage)
Hospital E Men not discharged home
Older patients with an MD visit and either a call or
white race
White men with RTMR or home health
Older white patients with no MD visit and a call
Men with home health and no RTMR and a call
Non-white men with a call and MD visit and either no
home health or no RTMR
(35% coverage)
Older white patients with an MD visit
Younger white patients with no MD visit and a
call
Younger men with home health and a call and
no RTMR
(13% coverage)
MODEL # AND
CONDITIONS
INCLUDED
RESULTS: SOLUTION PATHWAYS LEADING TO A NON-READMISSION
91
85% CONSISTENCY 90% CONSISTENCY
Hospital F English language or women and home health
Women and MD visit or younger with no call or white
with a call
Older white men with no call
Older non-white mean with a call and no home health
Younger white patients with home health and a call
and no MD visit
(45% coverage)
White women either younger or with an MD visit
Older women with an MD visit
Younger white patients with home health and a
call and no MD visit
(14% coverage)
Hospital G Men
(46% coverage)
Men
(46% coverage)
Hospital H Not discharged home
Non-English speaking women
White patients with an MD visit and no call
Older men with an MD visit and either a home health
visit or (no home health visit and a call)
Older or white women with no call
(36% coverage)
Non-English speaking women
White patients with an MD visit and no call
Non-white men with no home health and an MD
visit and a call
Younger or male and discharged home
Older or white women discharged home with no
call
(27% coverage)
Hospital I Women not discharged home
Older patients with either no call or (a home health
visit and MD visit)
A home health visit and no call
Men with a home health visit and RTMR and an MD
No home health and RTMR and young age or no MD
Women with MD visit and RTMR or no home health
Younger non-white women with RTMR or no home
health
(47% coverage)
Women not discharged home
Younger and no call
Younger with RTMR and either an MD visit or
(home health and no MD visit)
Non-white with a home health visit and either
an MD visit or (no RTMR and younger age)
(32% coverage)
Hospital J RTMR and no home health
Younger age and an MD visit
Men with RTMR and an MD visit
(31% coverage)
No home health and RTMR
Younger age and RTMR and MD visit
(16% coverage)
Hospital K Non-white women and home health visit
Men with no home health and an MD visit
Young women with no MD visit and a call
White women with a call and no home health and no
MD visit
(30% coverage)
Older, non-white women with a home health
visit
(6% coverage)
Hospital L Non-white women with a home health visit
Women with a call and no MD visit and either (white
and no home health) or (younger English speaking)
Men with an MD visit and call and no home health
(28% coverage)
Home health and non-English speaking or
older, non-white women
(9% coverage)
MODEL # AND
CONDITIONS
RESULTS: SOLUTION PATHWAYS LEADING TO A NON-READMISSION
92
Conclusion
My study suggests delivering the same bundle of interventions to all heart failure may not
be either the most beneficial for patients or the most effective use of limited resources to
prevent heart failure readmissions. Personal issues appear to be important drivers of heart
failure readmissions. Therefore understanding each patient’s background, lifestyle,
culture, and individual needs including other health conditions may be important to
determining the optimal plan of care for avoiding a rehospitalization. These results are
consistent with my literature review, which found care providers who developed personal
care plans with their patients achieved the lowest readmission rates.
93
Chapter Five: Discussion and Extensions
Introduction
There is mounting industry pressure on reducing readmissions to align with the Centers
for Medicare and Medicaid five-star ratings and to avoid financial penalties. Hospitals
across the US are looking to the literature to understand what has worked in other
institutions. The literature tells us that adopting specific, isolated interventions may not
always result in reduced readmissions due to differences in usual care and individual
patient needs. The hospital system I studied adopted specific practices cited in the
literature; yet after an initial period of readmission reduction, results plateaued. From my
data analysis and case studies, I found all patients within the health system I studied were
supposed to receive the same interventions despite potential differences in individual
needs, but processes, roles and metrics were not designed to be reliable or achieve the
desired outcome.
I discuss my research findings and potential extensions in three parts. First, I explore the
concept of patient-centered care including a brief review of what the literature indicates is
missing from health care today. Second, I propose a patient-centered framework to
reducing heart failure readmissions. I conclude that health care organizations seeking to
reduce heart failure readmissions should adopt a flexible, patient-centered model that
94
encourages providers to get to know their patients and collaboratively determine the best
care plan for each individual.
Patient-Centered Care
Patient centeredness is an increasingly common term in health care literature, although
there is not one standard definition for what constitutes patient centeredness or the
elements of it (Lauver, et al., 2002). At its core, this concept generally refers to inclusion
of a patient’s values, objectives, and support structure when planning and delivering care.
The literature defines multiple types of centeredness. In addition to patient-centeredness,
there is client-centeredness, family-centeredness, person-centeredness and relationship-
centered care (Hughes, Bamford, & May, 2008). For the purposes of this paper, I will use
the term “patient-centeredness” to broadly refer to the centeredness concept. Despite the
different labels, each approach is rooted in the notion of shared decision-making which is
a paradigm shift from historical, paternalistic approaches to delivering care.
Hughes reviewed the literature and found ten common themes among these centeredness
models (Hughes, Bamford, & May, 2008):
1. Respect for the individual person and their own needs and rights
2. Understanding the person’s view of their illness
95
3. Empathetic, honest, accommodating relationships to facilitate a partnership
between care giver and patient
4. Understanding the person’s social network and their phase of life
5. Integrated holistic understanding of the individual’s unique world with a
recognition of his or her idiosyncratic and broader life-setting
6. Recognize the individual’s or family’s expert knowledge about the person
7. Shared responsibility and collaborative decision making
8. Bi-directional communication including listening, and verbal and non-verbal
queues
9. Independence of individuals to make their own choices
10. Acknowledging the physicians and other professional staff as individuals who
may also need assistance fulfilling their roles
The literature reflects concerns that the relationship between patients and providers has
been eroding. In 2006, Bergeson and Dean found physicians and other clinicians do not
always attend to patients’ questions and concerns, attempt to understand patients’ goals
and beliefs, or engage them in discourse about care delivery options (Bergeson & Dean,
2006). This may be a reflection of the burgeoning growth of people with chronic
conditions or even multiple chronic conditions and medical community’s trend over the
past two decades to focus more on caring for diseases rather than caring for people
(Holman 2006).
96
Oeseburg and Abma suggest health care should be viewed as a shared undertaking with
both patient and provider responsible for planning and adhering to agreements (Oeseburg
and Abma 2006). For individuals who have chronic conditions, patient-centeredness is of
even greater importance to improve quality of life (Downs & Mackenzie, 2006) (Sabat,
2006). This point was reinforced in a 2009 article in which Annema et al studied
readmission causes for 173 patients with heart failure. They interviewed the patients,
families and providers involved in the cases to gain their opinions of what led to the
readmission. Over half of all patients and providers believed comorbidities, patient non-
adherence, and medications that were not optimized played a part (Annema, Luttik,, &
Jaarsma, 2009). Patients and their care providers only agreed on the readmissions cause
in one-third of the cases.
Jencks argues patients and their families need to trained and empowered to insist upon
the appropriate health care (Jencks S. , 2010). Bentley et al and Bergeson and Dean also
suggest improved patient and family engagement can avoid medical errors and health
care costs since treatment guidelines are generally developed for the “average person”
(Bentley, Effros, Palar, & Keeler, 2008) (Bergeson & Dean, 2006). More recently, Cowie
observed our national health policy is changing to what he called a “broader bio-psycho-
social and spiritual view” with greater collaboration between patient and provider,
although he noted the evidence suggests cardiologists have been slower to adopt patient-
centeredness concepts than primary care providers (Cowie, 2011).
97
A Patient-Centered Approach to Reducing Heart Failure Readmissions
I recently attended a meeting at The Joint Commission headquarters. During the
introduction, the President, Mark Chassin, MD, presented a paper he published regarding
a hand hygiene improvement initiative with eight hospitals. The teams found in aggregate
there were fifteen barriers to instituting effective hand hygiene practices, but each
participating hospital experienced a differing subset (Chassin & Loeb, 2011). Therefore,
if any one hospital had taken the prescribed solution set that had worked at another and
implemented it without understand their own specific needs; they would not have
achieved the same success. Chassin and Loeb concluded the key to improvement is to
first understand the root cause of the problem each hospital is experiencing and then
apply the appropriate subset of solutions.
It struck me that this concept could be extended to the delivery of health care. Perhaps
there is a broader universe of potential interventions every person with heart failure may
need, but only by knowing their unique goals, beliefs, and challenges could we know
which care would best benefit them at any given time. This theory is consistent with my
research and data analysis. I found there seems to be a causal relationship between the
physician visit, the real-time medication reconciliation, and the follow-up call and a
lower readmission rate when each intervention is examined individually, but as I added
interventions to the analysis they were not always incrementally beneficial to every
patient.
98
I build upon Hansen’s Readmissions Reduction Intervention Framework in Figure 12 by
modifying the interventions included in the original model, adding potential care
interventions cited in the literature cited as benefiting some patients, and adding further
interventions derived my observations and study (Hansen, Young, Hinami, Leung, &
Williams, 2011).
Figure 12: Hansen's Readmission Reduction Intervention Framework
The resulting framework encompasses a collection of potential interventions from which
providers may draw based on their patient’s specific needs as indicated in Figure 13. I
add to Hansen’s model a fourth time period, preadmission interventions, to reflect the
need to provide the optimal usual care from the time of diagnosis.
The model I propose is based on themes derived from the literature that care should be:
led by a cardiologist, intensive and ongoing, and personalized. I suggest it begins with the
care team getting to know the specific individual, including their living situation,
99
personal values and preferences, and social support structure to identify their unique care
needs. This may also include using a risk stratification methodology such as LACE to
assess how at risk the person is for a readmission (van Walraven, et al., 2012). This
model also requires primary care, cardiology, and hospital-based providers to collaborate
with each other to provide a team-based approach to the patient’s care.
Figure 13: Garofalo’s Patient-Centered Framework for Reducing Heart Failure Readmissions
Predischarge Interventions
o Identify patient’s unique care needs – At the heart of patient-centeredness
is understanding the patient and their unique needs; (Moser, Watkins, & J,
Preadmission Interventions
• Identify patient’s unique
care needs
• Patient education and
coaching to self-care
• Medication reconciliation
• Physician appointments as
needed
• Patient and family
discussion and care
planning
• Personalized dietary and
exercise plans
• Patient hotline
• Measure and monitor
depression
• Measure and monitor
quality of life
• End of life planning
• Any other interventions as
needed by the person
Predischarge interventions
• Identify patient’s unique
care needs
• Patient education and
coaching to self-care
• Discharge criteria
• Comprehensive discharge
planning
• Medication reconciliation
• Appointment scheduled
after discharge
• Patient and family
discussion and care
planning
• Personalized dietary and
exercise plans
• Any other interventions as
needed by the person
Postdischarge interventions
• Continue the care started in
the hospital
• Patient hotline
• Measure and monitor
depression
• Measure and monitor
quality of life
• End of life planning
• Any other interventions as
needed by the person
Interventions bridging the transition
• Transition coach
• Patient-centered discharge instructions
• Provider continuity
100
2008) (Rich, Beckham, Wittenburg, Leven, Freedland, & Carney, 1995)
this would include, but not be limited to, their literacy and whether they
live alone which the literature indicated are important influencers on
readmission (Powell & Kripalani, 2005) (Baker, Parker, Williams, &
Clark, 1998) (Bathaei, Ashktorab, Zohari Anbuhi, Alavi Majd, & Ezzati,
2009) (Evangelista, Doering, & Dracup, 2000) (DiIorio, et al., 1998). Each
care provider that comes in contact with a patient can learn about them
and contribute to this learning process. The information gleaned can serve
as input to the care plan and determining which other interventions may be
necessary.
o Patient education and coaching to self-care – Rather than focus just on
education, I recommend modifying this process to reflect the need for both
intensive education and coaching to self care, that should include the
family and/or care giver. (Yu, Thompson, & Lee, 2006) (Krumholz, et al.,
2002) (Ni, Burgess, Wise, Crispell, & Hershberger, 1999) (Ditewig, Blok,
Havers, & van Veenendaal, 2010) (Krumholz, et al., 2002) (Moser,
Watkins, & J, 2008).
o Discharge criteria – Before we send patients home from the hospital, the
literature suggest we should ensure the person is physically and
emotionally prepared to leave the hospital. Kossovsky et al found one
driver of readmission is patients not ready to leave the hospital
(Kossovsky, Sarasin, Perneger, Chopard, Sigaud, & Gaspoz, 2000). I
101
identified two examples of clinical discharge criteria in the literature.
Ledwidge et al waited until heart failure patients were at their dry weight
for two days before they discharged a person to ensure they were more
stable (Ledwidge, et al., 2003). Mejhert and others have also evaluated B-
type natriuretic peptide (BNP) as a predictive indicator of mortality and
readmission (Mejhert, Kahan, Persson, & Edner, 2006).
o Comprehensive discharge planning – Discharge planning should
encapsulate ensuring all of the aforementioned interventions have
appropriately taken place and the patient and family feel appropriately
prepared to transition to the next care venue, be it a skilled nursing facility
or the home. This process is especially important for older patients who
need greater support (Phillips, Wright, Kern, Singa, Shepperd, & Rubin,
2004). Discharge planning should be a collaborative process that evaluates
if the necessary support structure exists and working with family, friends
and community resources to provide a safe and healing environment. This
process should ideally include the family and other members of the
patient’s support network in this process as literature as indicates informal
caregivers often feel ill-prepared for their responsibilities (Foust,
Vuckovic, & Henriquez, 2012)
o Medication reconciliation – Building upon Hansen’s model, I suggest
clarifying medication reconciliation to include medication reconciliation
both at admission and discharge to ensure patients both receive the correct
102
medication while hospitalized and after discharge. Anecdotally I heard
examples of patients receiving medications during hospitalization that had
previously been discontinued, but it was not appropriately documented in
their medication list.
o Appointment scheduled after discharge – The literature suggests patients
who receive follow-up care from a cardiologist have lower readmission
rates and better outcomes than those cared for by generalists (Ansari,
Alexander, Tutar, Bello, & Massie, 2003) (Reis, Edmundowicz,
McNamara, Zell, Detre, & Feldman, 1997) (Laramee, Levinsky, Sargent,
Ross, & Callas, 2003) (Capomolla, et al., 2002) (Atienza, et al., 2004)
(Yu, Thompson, & Lee, 2006). I suggest modifying this intervention to
reflect the need of a cardiologist in heart failure patients’ care in
collaboration with other disciplines caring for other aspects of the person’s
health to ensure the resulting care plans are holistic and not conflicting.
Furthermore, I suggest scheduling the appointment day and time with the
patient’s and family’s input to ensure the patient agrees to the scheduled
visit, is able to pay, and has transportation on the agreed upon day and
time. Some patients may also need an alternative approach to conducting
this visit, such as having the physician come to their home if they are
unable to travel (Ekman, Andersson, Ehnfors, Persson, & Fagerberg,
1998).
103
o Patient and family discussion and care planning – A key tenet of patient-
centeredness, in this first step key clinicians involved in the patient’s care
while in the hospital and after discharge should be involved in a process of
understanding patient and family goals, values, and beliefs, and in
developing an agreed upon plan of care that extends into activities of daily
life (Hughes, Bamford, & May, 2008). If necessary a warm handoff to
ambulatory care providers should occur to ensure continuity of this
process.
o Personalized dietary and exercise plans – The literature indicates heart
failure patients benefit from nutrition and exercise plans that reflect their
preferences (Capomolla, et al., 2002) (Courtney, Edwards, Chang, Parker,
Finlayson, & Hamilton, 2009). I suggest beginning the appropriate
nutrition and exercise plans with the patient and family while the patient is
in the hospital so the home can be prepared before the patient returns.
o Any other interventions as needed by the person – Although the
aforementioned list may account for most of what our patients may need,
in my case studies I observed a random assortment of interventions. For
example, one lady who had heart failure was also on dialysis. She was
challenged getting to her dialysis appointments regularly because she did
not drive and had a problem of persistent diarrhea. On more than one
occasion, she arranged for someone else to take her to her dialysis
appointment, but was then struck with the diarrhea en route and chose to
104
return home. Thus, she had two problems requiring resolution: the
diarrhea and lack of a driver’s license. After her physician worked with
her to resolve her diarrhea, a social worker arranged for a ride to take her
to the department of motor vehicles so she could get her license.
Postdischarge Interventions
o Continue the care started in the hospital – Continue the care started pre-
discharge, including coaching to self-care, support and adjustment of
individualized nutrition and exercise plans, social support, education,
medication management, and physician or cardiology clinic visits based
on the patient’s individual needs. As the literature indicates, continued
care seems to be a necessary component of keeping people healthy and out
of the hospital (Capomolla, et al., 2002) (Ledwidge, et al., 2003)
(McDonald, et al., 2002) (Kasper, et al., 2002) (Ni, Burgess, Wise,
Crispell, & Hershberger, 1999) (Krumholz, et al., 2002).
o Patient hotline – Hansen’s framework included the concept of a patient
hotline, yet the literature varied in terms of what constituted a hotline.
Some studies offered a phone number that was available during business
hours and others offered it twenty-four hours per day. It may be optimal to
have a single number available to patients twenty-four hours a day, seven
days per week to avoid emergency department visits that may lead to a
readmission.
105
o Measure and monitor depression – Depression is common in heart failure
patients and can affect their adherence to care plans and outcomes (Moser,
Watkins, & J, 2008). I suggest regularly evaluating a person’s depression
using a standard tool such as the PHQ-2 (patient health questionnaire) or
PHQ-9 to identify mood changes and determine if other support needs
may be warranted.
o Measure and monitor quality of life – In concert with depression, people
with heart failure may also benefit from monitoring their quality of life.
The literature indicated elderly adults with lower life satisfaction or
quality of life, including poor satisfaction with their social conditions led
to more admissions (Fethke, Smith, & Johnson, 1986) (DiIorio, et al.,
1998) (Mejhert, Kahan, Persson, & Edner, 2006). As the disease
progresses and quality of life changes, it may serve as a trigger for
monitoring plans of care. Mejhert et al studied use of a heart failure-
specific quality of life tool and suggested it is more valuable than a
generalized instrument (Mejhert, Kahan, Persson, & Edner, 2006).
o End of life planning – Research suggests people would rather plan their
end of life and advanced directives before they become ill and with a
physician who they know well (Wissow, Belote, Kramer, Compton-
Phillips, Kritzler, & Weiner, 2004). Rather than begin these discussions in
a hospital when a person is most ill and vulnerable, I suggest we begin
these conversations early in the disease progression rather than waiting
106
until they become very ill and are hospitalized. Quill suggests planning the
end of one’s death in a collaborative, caring fashion is also a vital
component of quality health care (Quill, 2000).
o Any other interventions as needed by the person – Continue to understand
the patient’s and family’s needs to identify any other unique needs.
Hansen’s framework included three elements to help bridge the transition from the
hospital setting to after discharge: transition coach, patient-centered discharge
instructions, and provider continuity. I do not offer any suggested modifications to these
components except to continue to emphasize the same patient-centeredness approach.
I also suggest adding a fourth time period to the framework: preadmission interventions.
If patients are diagnosed with heart failure before they are admitted to the hospital, I
believe starting the appropriate interventions sooner may improve health outcomes and
avoid the index admission altogether. Ideally this would begin with understanding the
patient’s unique needs, building a collaborative care plan, intensive family and patient
education and coaching, personalized dietary and exercise plans, and a schedule of
follow-up care.
107
Conclusion
My data analysis indicates it is a fallacy that providing more care consistently yields
better patient outcomes. I posit a flexible model of care that is highly personalized and
encourages providers to get to know their patients may yield better outcomes and may be
more cost effective. As with Chassin’s hand hygiene findings, I suggest there is a toolbox
of potential interventions providers can employ based on their patient’s and patient’s
family’s need and their best judgment. My recommended framework is based in part on
observations from the literature and impressions from my case studies, thus there is an
opportunity for other researches to further explore and extend it.
108
Chapter Six: Conclusion
The objective of this study was to understand how health care providers with limited
financial resources and increasing pressure to reduce avoidable readmissions can most
effectively and efficiently care for people with heart failure. I began by spending several
months interviewing and observing frontline workers, during which time I developed a
theory that a more flexible approach to caring for heart failure patients may yield better
outcomes and be a more prudent use of resources.
Based on the findings of Chassin, Capomolla, Barth, Jaarsma, Ledwidge, and McDonald,
I further refined my theory to extend Hansen’s heart failure readmission reduction
framework into a more flexible, patient-centered model (Chassin & Loeb, 2011)
(Capomolla, et al., 2002) (Barth, 2001) (Jaarsma, et al., 1999) (Ledwidge, et al., 2003)
(McDonald, et al., 2002). It enables providers to use their best judgment in collaboration
with patients and families to determine the best treatment plan to manage their heart
failure.
Yet even with this proposed model, Pronovost et al write that delivering care consistently
is far more difficult than determining the right interventions to provide (Pronovost, et al.,
2006). Thus, even a well developed patient-centered readmission reduction framework
will not be effective if the intended care is not delivered reliably.
109
My case studies indicated hospitals with lower readmission rates had engaged program
leadership and oversight, an aligned organizational structure, and effective process and
role design. This is consistent with Donabedian’s model for analyzing health systems
quality, which posits processes, and their resulting outcomes, are constrained by the
organizational structures in which they operate (Donabedian, 1966). Therefore, health
care providers attempting to establish a health care program may wish to also focus on
their structure to optimally deliver processes and achieve the desired outcomes.
I provide health care providers who wish to reduce their heart failure readmissions with
guidance for effectively designing program leadership and oversight, organizational
structure, and processes below.
Program Leadership and Oversight
In hospitals with lower readmission rates, empowered administrative and physician
leaders co-sponsored the program, set the tempo of improvement, and removed process
or resource barriers to running an effective program. An oversight group met at least once
per month to maintain a constant focus on measuring results and improving frontline
processes, and keep executive leadership apprised of progress. Often, smaller work teams
met weekly or bi-weekly to make process improvements identified in the oversight
meeting. The program sponsors organized and facilitated oversight meetings, to enforce
attendance by other necessary leaders, managers, and front line staff, drive improvement
110
and hold people accountable for results. In organizations that had higher readmission
rates, sponsors were not highly engaged, the oversight team met less than once per
month, oversight meetings were often facilitated by a person not empowered to enforce
participation, oversight group attendance was irregular, and no one was held accountable
for process consistency or outcomes.
I suggest health care organizations appoint jointly accountable administrative and
physician leaders to sponsor their heart failure readmissions reduction program and lead
an oversight process. The sponsors must play an active role in the oversight process to
demonstrate the importance of the work to the organization and set a tempo of rapid
improvement.
Organizational Structure
I found that hospitals with the lowest heart failure readmission rates had organizational
reporting relationships that aligned frontline staff across the care continuum with a single
or few leader(s) who sponsored the program, which enabled leadership to set clear
priorities and hold people accountable. Conversely, in hospitals with higher heart failure
readmission rates, the frontline staff involved in the program reported into multiple
organizational leaders who had varied or even competing priorities. The program
sponsors were not empowered to set individual goals, direct changes to individual staff
111
members’ work processes, or hold people accountable to previously made commitments.
Thus, there were no consequences for not delivering care reliability or improving results.
I recommend health care leaders should evaluate their organizational structure and
culture to determine if incentives are appropriately aligned and the program sponsors are
empowered to influence and hold accountable the necessary individuals across the care
continuum to develop and refine processes as needed.
Process Design
The hospitals with the lowest readmission rates and highest process reliability designed
processes to reflect customer needs to ensure the patient was getting the care they needed
while being highly reliable: simplified and standardized, using controls to help prevent
errors (e.g. checklists), and using mitigation strategies (e.g., redundancies) to identify and
interrupt errors. In addition, processes were designed to learn from errors and
continuously improve. Hospitals with lower consistency did not have formally
documented processes and thus no basis from which to improve. They tended to rely
upon providers’ memory and provided care in a way that was simplest and most
convenient for them rather than for the patient and their family.
Once a health care organization defines its heart failure readmissions reduction program,
I propose designing the processes for delivering patient care to clarify roles, identify the
112
necessary handoffs, and determine when to employ process controls to ensure they are as
efficient and reliable as possible; for example, once a care plan is developed between
patients and their providers, use reminders and checklists to implement the plan. I suggest
documenting processes and so that as errors or improvement opportunities are identified
the oversight team can agree to the specific improvements needed and quickly implement
them across the system.
Conclusion
There is a paradigm shift in the health care industry from a paternalistic, physician-driven
model to one in which patients and providers collaborate and share in decision-making.
As health care leaders seek to understand what has worked in other organizations, I
suggest they should evaluate not only which interventions to provide patients, but also the
organizational construct in which those processes are delivered. I outline a patient-
centered framework for reducing heart failure readmissions that begins with
understanding each individual to determine which interventions are optimal for that
person at that particular time and guidance for effectively designing a heart failure
readmission reduction program with engaged leaders, program oversight that includes
leaders and frontline staff, and well designed processes that focus on the patient and
encourage continuous improvement. I believe as we get to know our patients and plan
care with them in the room, we too will learn more about how we can provide the most
effective care.
113
Bibliography
Alexander, M., Grumbach, K., Remy, L., Rowell, R., & Massie, B. (1999). Congestive
heart failure hospitalizations and survival in California: patterns according to
race/ethnicity. American Heart Journal , 919-927.
Allaudeen, N., Vidyarthi, A., Maselli, J., & Auerbach, A. (2011). Redefining readmission
risk factors for general medicine patients. Journal of Hospital Medicine , 54-60.
Anderson, M. A., Helms, L. B., Hanson, K. S., & DeVilder, N. W. (1999). Unplanned
hospital readmissions: a home care perspective. Nursing Research , 299-307.
Annema, C., L. M.-L., & Jaarsma, T. (2009). Reasons for readmission in heart failure:
perspectives of patients, caregivers, cardiologists, and heart failure nurses. Heart & Lung:
The Journal of Acute and Critical Care , 38 (5), 427-434.
Ansari, M., Alexander, M., Tutar, A., Bello, D., & Massie, B. (2003). Cardiology
participation improves outcomes in patients with new-onset heart failure in the outpatient
setting . Journal of the American College of Cardiology , 62-68.
Ansari, M., Collopy, B., & Booth, J. (1995). Hospital characteristics associated with
unplanned readmissions. Australian Health Review , 63-75.
Ashton, C. M., Del Junco, D. J., Souchek, J., Wray, N. P., & Mansyur, C. L. (1997). The
association between the quality of inpatient care and early readmission: a meta analysis
of the evidence. Medical Care , 1044-1059.
Ashton, C. M., Kuykendall, D. H., Johnson, M. L., Wray, N. P., & Wu, L. (1995). The
association between the quality of inpatient care and early readmission. Annals of
Internal Medicine , 415-421.
Atienza, F., Anguitac, M., Martinez-Alzamorad, N., Oscab, J., Ojedac, S., Almenarb, L.,
et al. (2004). Multicenter randomized trial of a comprehensive hospital discharge
andoutpatient heart failure management program. The European Journal of Heart Failure
, 643– 652.
114
Baker, D. W., Parker, R. M., Williams, M. V., & Clark, W. S. (1998). Health literacy and
the risk of hospital admission. Journal of General Internal Medicine , 791-798.
Bakhtiari, E. (2010, June 24). Follow-up appointments may not reduce readmissions.
Retrieved February 7, 2011, from Health Leaders Media:
hppt://www.healthleadersmedia.com/print/QUA-252982/Followup-Appointments-May-
Not-Reduce...
Barth, V. (2001). A nurse-managed discharge program for congestive heart failure
patients: outcomes and costs. Home Health Care Management & Practice , 436-443.
Bathaei, S., Ashktorab, T., Zohari Anbuhi, S., Alavi Majd, H., & Ezzati, J. (2009).
Personal factors contributing to readmission of patients with congestive heart failure.
Iranian Journal of Critical Care Nursing , 109-112.
Bathaei, S., Ashktorab, T., Zohari, A. S., Alavi, M. H., & Ezati, J. (2009). Personal
factors contributing to readmission of patients with congestive heart failure. Iranian
Journal of Critical Care Nursing , 109-112.
Bentley, T., Effros, R., Palar, K., & Keeler, E. (2008). Waste in the U.S. health care
system: a conceptual framework. The Milbank Quarterly , 86 (4), 629-659.
Bergeson, S., & Dean, J. (2006). A systems approach to patient-centered care. JAMA ,
296 (23), 2848 - 2851.
Bero, L. A., Lipton, H. L., & Bird, J. A. (1991). Characterization of geriatric drug-related
hospital readmissions. Medical Care , 989-1003.
Billings, J., Dixon, J., Mijanovich, T., & Wennberg, D. (2006, June 30). Case finding for
patients at risk of readmission to hospital: development of algorithm to identify high risk
patients. British Medical Journal , 1-6.
Billups, S. J., Malone, D. C., & Carter, B. L. (2000). The relationship between drug
therapy noncompliance and patient characteristics, health-related quality of life, and
health care costs. Pharmacotherapy , 941-949.
115
Blue, L., Lang, E., McMurray, J., Davie, A., McDonagh, T., Murdoch, D., et al. (2001).
Randomised controlled trial of specialist nurse intervention in heart failure. BMJ , 323,
715-718.
Bowles, K. H., Naylor, M. D., & Foust, J. B. (2002). Patient characteristics at hospital
discharge and a comparison of home care referral decisions. Journal of the American
Geriatric Society , 336-342.
Camberg, L. C., Smith, N. E., Beaudet, M., Daley, J., Cagan, M., & Thibault, G. (1997).
Discharge destination and repeat hospitalizations. Medical Care , 756-767.
Capomolla, S., Febo, O., Ceresa, M., Caporotondi, A., Guazzotti, G., La Rovere, M., et
al. (2002). Cost/utility ratio in chronic heart failure: comparison between heart failure
management Program Delivered by Day-Hospital and Usual Care. Journal of the
American College of Cardiology , 1259-1266.
Chassin, M. R., & Loeb, J. M. (2011). The ongoing quality improvement journey: next
stop, high reliability. Health Affairs , 559-568.
Chin, M., & Goldman, L. (1997). Correlates of early hospital readmission or death in
patients With congestive heart failure. The American Journal of Cardiology , 1640-1644.
Chu, L., & Pei, C. (1999). Risk factors for early emergency hospital readmission in
elderly medical patients. Gerontology , 220–226.
Clark, R. A., Inglis, S. C., McAlister, F. A., Cleland, J. G., & Stewart, S. (2007).
Telemonitoring or structured telephone support programmes for patients with chronic
heart failure: systematic review and meta-analysis. BMJ , 1-9.
Cleland, J., Louis, A., Rigby, A., Janssens, U., Balk, A., & Investigators, T.-H. (2005).
Noninvasive Home Telemonitoring for Patients With Heart Failure at High Risk of
Recurrent Admission and Death: The Trans-European Network-Home-Care Management
System (TEN-HMS) study. Journal of American College of Cardiology , 45, 1654-1664.
Cline, C., Israelsson, B., Willenheimer, R., & Broms, K. E. (1998). Cost effective
management programme for heart failure reduces hospitalisation. Heart , 80, 442–446.
116
Corrigan, J. M., & Martin, J. B. (1992). Identification of factors associated with hospital
readmission and development of a predictive model. Health Services Research , 81-101.
Courtney, M., Edwards, H., Chang, A., Parker, A., Finlayson, K., & Hamilton, K. (2009).
Fewer emergency readmissions and better quality of life for older adults at risk of
hospital readmission: a randomized controlled trial to determine the effectiveness of a 24-
week exercise and telephone follow-up program. Journal of American Geriatrics Society
, 395-402.
Cowie, M. (2011, September 15). Person-centred care: more than just improving patient
satisfaction? (Editorial). European Heart Journal , 1 - 3.
DeBusk, R., Miller, N., Parker, K., Bandura, A., Kraemer, H., Cher, D., et al. (2004).
Care management for low-risk patients with heart failure. Annals of Internal Medicine ,
141, 606 - 613.
DiIorio, A., Longo, A., Mitidieri Costanza, A., Bandinelli, S., Capasso, S., Gigante, M.,
et al. (1998). Characteristics of geriatric patients related to early and late readmissions to
hospital. Aging (Milano) , 339-346.
Ditewig, J., Blok, H., Havers, J., & van Veenendaal, H. (2010). Effectiveness of self-
management interventions on mortality, hospital readmissions, chronic heart failure
hospitalization rate and quality of life in patients with chronic heart failure: A systematic
review. Patient Education and Counseling , 297–315.
Donabedian, A. (1966). Evaluating the quality of medical care. Milbank Memorial Fund
Quarterly , 166–206.
Downs, M. C., & Mackenzie, J. (2006). Understandings of dementia: explanatory models
and their implications for the person with dementia and therapeutic effort. In J. Hughes,
& S. S. Louw, Dementia: mind, meaning, and the person (pp. 235–258). Oxford: Oxford
University Press.
Ekman, I., Andersson, B., Ehnfors, M. M., Persson, B., & Fagerberg, B. (1998).
Feasibility of a nurse-monitored, outpatient-care programme for elderly patients with
moderate-to-severe, chronic heart failure. European Heart Journal , 1254-1260.
117
Evangelista, L. S., Doering, L. V., & Dracup, K. (2000). Usefulness of a history of
tobacco and alcohol use in predicting multiple heart failure readmissions among veterans.
The American Journal of Cardiology , 86, 1339-1342.
Farisa, R., Purcella, H., Heneina, M., & Coatsa, A. (2002). Clinical depression is
common and significantly associated with reduced survival in patients with non-
ischaemic heart failure. The European Journal of Heart Failure , 541–551.
Fethke, C. C., Smith, I. M., & Johnson, N. (1986). "Risk" factors affecting readmission of
the elderly into the health care system. Medical Care , 429-437.
Foust, J., Vuckovic, N., & Henriquez, E. (2012). Hospital to home health care transition:
patient, caregiver, and clinician perspectives. Western Journal of Nursing Research , 34
(2), 194 - 212.
Friedmann, J., Jensen, G., Smiciklas-Wright, H., & McCamish, M. (1997). Predicting
early nonelective hospital readmission in nutritionally compromised older adults.
American Journal of Clinical Nutrition , 1714-1720.
Gattis, W., Hasselblad, V., Whellan, D., & O’Connor, C. (1999). Reduction in Heart
Failure Events by the Addition of a Clinical Pharmacist to the Heart Failure Management
Team. Archives of Internal Medicine , 1939 - 1945.
GESICA Investigators. (2005, August 1). Randomised trial of telephone intervention in
chronic heart failure:DIAL trial. BMJ , 1 - 5.
Goldberg, L., Piette, J., Walsh, M., & Frank, T. (2003, October). Randomized trial of a
daily electronic home monitoring system in patients with advanced heart failure: The
Weight Monitoring in Heart Failure (WHARF) trial. American Heart Journal , 705 - 712.
Gonseth, J., Guallar-Castill, P., Banegas, J., & Rodrıguez-Artalejo, F. (2004). The
effectiveness of disease management programmes in reducing hospital re-admission in
older patients with heart failure: a systematic review and meta-analysis of published
reports. European Heart Journal , 1570–1595.
118
Goodyer, L., Miskelly, F., & Milligan, P. (1995). Does encouraging good compliance
improve patients' clinical condition in heart failure? The British Journal of Clinical
Practice , 49 (4), 173 - 176.
Gudmundsson, G., Gislason, T., Janson, C., Lindberg, E., Hallin, R., Ulrik, C., et al.
(2005). Risk factors for rehospitalisation in COPD: role of health status, anxiety and
depression. European Respiratory Journal , 414-419.
Gwadry-Sridhar, F. H., Flintoft, V., Lee, D. S., Lee, H., & Guyatt, G. H. (2004). A
systematic review and meta-analysis of studies comparing readmission rates and
mortality rates in patients With heart failure. Archives of Internal Medicine , 2315-2320.
Hansen, L., Young, R., Hinami, K., Leung, A., & Williams, M. (2011). Interventions to
reduce 30-day rehospitalization: a systematic review. Annals of Internal Medicine , 520-
528.
Harrison, M., Browne, G., Roberts, J., Tugwell, P., Gafni, A., & Graham, I. (2002,
April). Quality of Life of Individuals with Heart Failure: A Randomized Trial of the
Effectiveness ofTwo Models of Hospital-to-Home Transition. Medical Care , 271 - 282.
Hasan, O., Meltzer, D. O., Shaykevich, S. A., Bell, C. M., Kaboli, P. J., Auerbach, A. D.,
et al. (2009). Hospital readmission in general medicine patients: a prediction model.
Journal of General Internal Medicine , 211-219.
Heggestad, T. (2002). Do hospital length of stay and staffing ratio affect elderly patients’
risk of readmission? A nation-wide study of norwegian hospitals. Health Services
Research , 647-665.
Holland, R., Battersby, J., Harvey, I., Lenaghan, E., Smith, J., & Hay, L. (2005).
Systematic review of multidisciplinary interventions in heart failure. Heart , 899–906.
Holland, R., Lenaghan, E., Harvey, I., Smith, R., Shepstone, L., Lipp, A., et al. (2005).
Does home based medication review keep older people out of hospital? The HOMER
randomised controlled trial. BMJ , 1-5.
Holloway, J. J., & Thomas, J. W. (1989). Factors influencing readmission risk:
implications for quality monitoring. Health Care Financing Review , 19-32.
119
Hughes, J., Bamford, C., & May, C. (2008, April 9). Types of centredness in health care:
themes and concepts. Medical Health Care and Philosophy , 455–463.
Jaarsma, T., Halfens, R., Huijer Abu-Saad, H., Dracup, K., Gorgels, T., van Ree, J., et al.
(1999). Effects of education and support on self-care and resource utilization in patients
with heart failure. European Heart Journal , 673–682.
Jencks, S. (2010). Defragmenting care. Annals of Internal Medicine , 757-758.
Jencks, S. W. (2009). Rehospitalizations among patients in the Medicare fee-for-service
program. New England Journal of Medicine , 360, 1418-28.
Jiang, H. J., Andrews, R., Stryer, D., & Friedman, B. (2005). Racial/ethnic disparities in
potentially preventable readmissions: the case of diabetes. American Journal of Public
Health , 1561-1567.
Joynt, K. E., Orav, E. J., & Jha, A. K. (2011). Thirty-day readmission rates for Medicare
beneficiaries by race and site of care. JAMA , 675-681.
Joynt, K., & Jha, A. (2012). Thirty-day readmissions - truth and consequences. NEJM ,
366, 1366-1369.
Karliner, L. S., Kim, S. E., Meltzer, D. O., & Auerbach, A. D. (2010). Influence of
language barriers on outcomes of hospital care for general medicine inpatients. Journal of
Hospital Medicine , 276-282.
Kasper, E., Gerstenblith, G., Hefter, G., Van Anden, E., Brinker, J., Thiemann, D., et al.
(2002). A randomized trial of the eficacy of mltidisciplinary cre in heart failure
outpatients at high risk of hospital readmission. Journal of the American College of
Cardiology , 39 (3), 471–480.
Kimmelstiel, C., Levine, D., Perry, K., Patel, A., Sadaniantz, A., Gorham, N., et al.
(2004). Randomized, controlled evaluation of short- and long-term benefits of heart
failure disease management within a diverse provider network. 110, 1450-1455 .
120
Kossovsky, M. P., Perneger, T. V., Sarasin, F. P., Bolla, F., Borst, F., & Gaspoz, J.-M.
(n.d.). Comparison between planned and unplanned readmissions to a department of
internal medicine. Journal of Clinical Epidemiology , 151-156.
Kossovsky, M., Sarasin, F., Perneger, T., Chopard, P., Sigaud, P., & Gaspoz, J. (2000).
Unplanned readmissions of patients with congestive heart failure: do they reflect in-
hospital quality of care or patient characteristics? The American Journal of Medicine ,
386-390.
Krumholz, H., Amatruda, J., Smith, G., Mattera, J., Roumanis, S., Radford, M., et al.
(2002). Randomized trial of an education and support intervention to prevent readmission
of patients with heart failure. Journal of the American College of Cardiology , 39 (1), 83–
89.
Laramee, A., Levinsky, S., Sargent, J., Ross, R., & Callas, P. (2003). Case management
in a heterogeneous congestive heart failure population. Archives of Internal Medicine ,
809-817.
Lauver, D., Ward, S., Heidrich, S., Keller, M., Bowers, B., Brennan, P., et al. (2002).
Patient-centered interventions. Research in Nursing & Health , 246–255.
Ledwidge, M., Barry, M., Cahill, J., Ryan, E., Maurer, B., Ryder, M., et al. (2003). Is
multidisciplinary care of heart failure cost-beneficial when combined with optimal
medical care? The European Journal of Heart Failure , 381–389.
McDonald, K., Ledwidge, M., Cahill, J., Quigley, P., Maurer, B., Travers, B., et al.
(2002). Heart failure management: multidisciplinary care has intrinsic benefit above the
optimization of medical care. Journal of Cardiac Failure , 142-148.
Mejhert, M., Kahan, T., Persson, H., & Edner, M. (2004). Limited long term effects of a
management programme for heart failure. Heart , 90 (9), 1010-1015.
Mejhert, M., Kahan, T., Persson, H., & Edner, M. (2006). Predicting readmissions and
cardiovascular events in heart failure patients. International Journal of Cardiology , 108-
113.
121
Mistiaen, P., Francke, A. L., & Poot, E. (2007). Interventions aimed at reducing problems
in adult patients. BMC Health Services Research , ????
Moser, D., Watkins, & J. (2008). Conceptualizing self-care in heart failure: a life course
model of patient characteristics. Journal of Cardiovascular Nursing , 205Y218.
Naylor, M., Brooten, D., Campbell, R., Maislin, G., McCauley, K., & Schwartz, J.
(2004). Transitional Care of Older Adults Hospitalized with Heart Failure: A
Randomized, Controlled Trial. Journal of the American Geriatrics Society , 52, 675 –
684.
Ni, H. N., Burgess, D., Wise, K., Crispell, K., & Hershberger, R. (1999). Factors
influencing knowledge of and adherence to self-care among patients with heart failure.
Archives Internal Medicine , 159, 1613-1619.
Oddone, E., Weinberger, M., Giobbie-Hurder, A., Landsman, P., & Henderson, W.
(1999). Enhanced access to primary care for patients with congestive heart failure.
Veterans Affairs Cooperative Study Group on Primary Care and Hospital Readmission.
Effective Clinical Practices , 2 (5), 201 - 209.
Ottenbacher, K. J., Smith, P. M., Illig, S. B., Fiedler, R. C., & Granger, C. V. (2000).
Length of stay and hospital readmission for persons with disabilities. American Journal
of Public Health , 1920-1923.
Philbin, E., Dec, G. W., Jenkins, P. L., & DiSalvoi, T. G. (2001). Risk factor for hospital
readmission for heart failure. American Journal of Cardiology , 1367–1371.
Philbin, E., Rocco, T., Lindenmuth, N., Ulrich, K., McCall, M., & Jenkins, P. (2000). The
results of a randomized trial of a quality improvement intervention in the care of patients
with heart failure. The MISCHF Study Investigators. American Journal of Medicine , 109
(6), 443-449.
Phillips, C., Wright, S., Kern, D., Singa, R., Shepperd, S., & Rubin, H. (2004).
Comprehensive discharge planning with postdischarge support for older patients with
congestive heart failure: a meta-analysis. JAMA , 291 (11), 1358 - 1367.
122
Powell, C. K., & Kripalani, S. (2005). Brief report: resident recognition of low literacy as
a risk factor in hospital readmission. Journal of General Internal Medicine , 1042-1044.
Pronovost, P., Berenholtz, S., Goeschel, C., Needham, D., Sexton, J., Thompson, D., et
al. (2006). Creating high reliability in health care organizations. Health Services
Research , 41, 1599-1617.
Quill, T. (2000). Initiating end-of-life discussions With seriously ill patients: addressing
the “elephant in the room” . JAMA , 284 (19), 2502 – 2507.
R.Doughty, J., Wright, S., Pearl, A., Walsh, H., Muncaster, S., Whalley, G., et al. (2002).
Randomized, controlled trial of integrated heart failure management. European Heart
Journal , 23, 139–146.
Ragin, C. C. (1987). The comparative method: moving beyond qualitative and
quantitative strategies. Berkeley/Los Angeles/London: University of California Press.
Ragin, C. (2008). Redesigning social inquiry: fuzzy sets and beyond. Chicago: University
of Chicago Press.
Ragin, C. (2006). User's guide to fuzzy-set/Qualitative Comparative Analysis 2.0.
Tucson, Arizona: Department of Sociology, University of Arizona.
Ragin, C., Drass, K., & Davey, S. (2009). fsQCA. Fuzzy-Set/Qualitative Comparative
Analysis 2.5 . Tucson, Arizona: Department of Sociology, University of Arizona.
Ragin, C., Drass, K., & Davey, S. (2006). Fuzzy-Set/Qualitative Comparative Analysis
2.0. (Department of Sociology, University of Arizona.) Retrieved April 10, 2012, from
http://www.u.arizona.edu/~cragin/fsQCA/citing.shtml
Reis, S. H., Edmundowicz, D., McNamara, D., Zell, K., Detre, K., & Feldman, A. (1997).
Treatment of patients admitted to the hospital With congestive heart failure: specialty-
related disparities in practice patterns and outcomes. Jounal of the American College of
Cardiology , 733-738.
123
Rich, M., Beckham, V., Wittenburg, C., Leven, C., Freedland, K., & Carney, R. (1995).
Multidisciplinary intervention to prevent the readmission of elderly patients with
congestive heart failure. New England Journal of Medicine , 333 (18), 1190-1195.
Riegel, B., Carlson, B., & Kopp, Z. (2002). Effect of a standardized nurse case-
management telephone intervention on resource use in patients with chronic heart failure.
Archives of Internal Medicine , 705-712.
Riegel, B., Carlson, B., Glaser, D., & Romero, T. (2006). Randomized Controlled Trial
of Telephone Case Management in Hispanics of Mexican Origin With Heart Failure.
Journal of Cardiac Failure , 12 (3), 211-219.
Roccaforte, R., Demers, C., Baldassarre, F., Teo, K., & Yusuf, S. (2005). Effectiveness of
comprehensive disease management programmes in improving clinical outcomes in heart
failure patients. A meta-analysis. The European Journal of Heart Failure , 1133 – 1144.
Sabat, S. (2006). Mind, meaning, and personhood in dementia: the effects of positioning .
In J. Hughes, S. Louw, & S. Sabat, Dementia: Mind, meaning, and the person (pp. 287–
302). Oxford: Oxford University Press.
Selected Medicare Hospital Quality Provisions Under the ACA. (n.d.). Retrieved March
11, 2012, from Association of American Medical Colleges:
https://www.aamc.org/advocacy/medicare/153882/selected_medicare_hospital_quality_p
rovisions_under_the_aca.html
Silverstein, M., Qin, H., Mercer, S., Fong, J., & Z, H. (2008). Risk factors for 30-day
hospital readmission in patients ≥65 years of age. Baylor University Medical Center
Proceedings , 363-372.
Smith, D. A., Norton, J. A., & McDonald, C. A. (1985). Nonelective readmissions of
medical patients. Journal of Chronic Diseases , 213-224.
Smith, D. M., Giobbie-Hurderb, A., Weinbergera, M., Oddonec, E. Z., Hendersonb, W.
G., Aschd, D. A., et al. (2000). Predicting non-elective hospital readmissions: a multi-site
study. Journal of Clinical Epidemiology , 113-118.
124
Sreeramoju, P., Montie, B., & Ramirez, A. M. (2010). Healthcare ‐associated infection: a
significant cause of hospital readmission. Infection Control and Hospital Epidemiology ,
1195-1197.
Stewart, S., Marley, J., & Horowitz, J. (1999). Effects of a multidisciplinary, home-based
intervention on planned readmissions and survival among patients with chronic
congestive heart failure: a randomised controlled study. Lancet , 354, 1077-1083.
Stewart, S., Pearson, S., & Horowitz, J. (1998). Effects of a Home-Based Intervention
Among Patients With Congestive Heart Failure Discharged From Acute Hospital Care.
Archives of Internal Medicine , 158, 1067 - 1082.
Stromberg, A., Martensson, J., Fridlund, B., Levin, L., Karlsson, J., & Dahlström, U.
(2003). Nurse-led heart failure clinics improve survival and self-care behaviour in
patients with heart failure: results from a prospective, randomised trial. European Heart
Journal , 1014-1023.
van Walraven, C., Bennett, C., Jennings, A., Austin, P. C., & Forster, A. J. (2011).
Proportion of hospital readmissions deemed avoidable: a systematic review. Canadian
Medical Association Journal , 183 (7), E391 - E402.
van Walraven, C., Dhalla, I., Bell, C., Etchells, E., Stiell, I., Zarnke, K., et al. (2012).
Derivation and validation of an index to predict early death or unplanned readmission
after discharge from hospital to the community. Canadian Medical Association Journal ,
551-557.
Varma, S., McElnay, J., Hughes, C., Passmore, A., & Varma, M. (1999). Pharmaceutical
care of patients with congestive heart failure: interventions and outcomes.
Pharmacotherapy: The Journal of Human Pharmacology and Drug Therapy , 19 (7),
860-869.
Victor, C. R., & Vetter, N. J. (1985). The early readmission of the elderly to the hospital.
Age and Ageing , 37-42.
Weinburger, M., Oddone, E., & Henderson, W. (1996). Does increased access to primary
care reduce hospital readmissions? The New England Journal of Medicine , 1441-1447.
125
Wells, B., Gentry, M., Ruiz-Arango, A., Dias, J., & Landolfo, C. K. (2006). Relation
between body mass index and clinical outcome in acute myocardial infarction. American
Journal of Cardiology , 474–477.
Wissow, L., Belote, A., Kramer, W., Compton-Phillips, A., Kritzler, R., & Weiner, J.
(2004). Promoting advance directives among elderly primary care patients. Journal of
General Internal Medicine , 19 (9), 944 - 951.
Yu, D. S., Thompson, D. R., & Lee, D. T. (2006). Disease management programmes for
older people with heart failure: crucial characteristics which improve post-discharge
outcomes. European Heart Journal , 596–612.
126
Appendix
Model 1: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, in_hfprogram, older_age, white_race, discharged_home,
male_gender, english_languag)
Truth table after delete and code (75% consistency):
127
Results (75% consistency):
128
Results (80% consistency):
Results (85% consistency):
129
Model 2: ~readmit_in_30 = f(homehealth_in_48)
Truth table after delete and code (79% consistency):
Results (79% consistency):
Model 3: ~readmit_in_30 = f(rtmr_in_48hour)
Truth table after delete and code (80% consistency):
130
Results (80% consistency):
Model 4: ~readmit_in_30 = f(md_in_7days)
Truth table after delete and code (80% consistency)
131
Results after delete and code (80% consistency)
132
Model 5: ~readmit_in_30 = f(call_in_7days)
Truth table after delete and code (80% consistency)
Results at 80% consistency:
133
Model 6: ~readmit_in_30 = f(in_hfprogram)
Truth table after delete and code (79% consistency):
Results (79% consistency):
134
Model 7: ~readmit_in_30 = f(older_age)
Truth table after delete and code (79% consistency):
Results (79% consistency):
135
Model 8: ~readmit_in_30 = f(white_race)
Truth table after delete and code (79% consistency)
Results (79% consistency): No negative cases.
Model 9: ~readmit_in_30 = f(discharged_home)
Truth table after delete and code (80% consistency)
136
Results (80% consistency):
Model 10: ~readmit_in_30 = f(male_gender)
Truth table after delete and code (79% consistency):
137
Results (79% consistency):
Model 11: ~readmit_in_30 = f(english_languag)
Truth table after delete and code (79% consistency):
138
Model 12: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour)
Truth table after delete and code (80% consistency):
139
Results (80% consistency):
Model 13: ~readmit_in_30 = f(homehealth_in_4, md_in_7days)
Truth table after delete and code (80% consistency):
140
Results (80% consistency):
141
Model 14: ~readmit_in_30 = f(homehealth_in_4, call_in_7days)
Truth Table after Delete and Code (75% consistency)
Results (75% consistency)
142
Results (80% consistency)
Model 15: ~readmit_in_30 = f(md_in_7days, call_in_7days)
Truth table after delete and code (75% consistency)
143
Results (75% consistency):
144
Results (80% consistency):
Model 16: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days)
Truth table after delete and code (80% consistency):
145
Results (80% consistency):
Model 17: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, call_in_7days)
Truth table after delete and code (75% consistency):
146
Results (75% consistency):
147
Results (80% consistency):
148
Model 18: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days)
Truth table after delete and code (75% consistency):
149
Results (75% consistency):
150
Results (80% consistency):
151
Results (85% consistency):
152
Model 19: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age)
Truth table after delete and code (75% consistency):
153
Results (75% consistency):
154
Results (80% consistency):
155
Results (85% consistency):
156
Model 20: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, white_race)
Truth table after delete and code (75% consistency):
157
Results (75% consistency):
158
Results (80% consistency):
159
Results (85% consistency):
160
Model 21: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, discharged_home)
Truth table after delete and code (75% consistency):
161
Results (75% consistency):
162
Results (80% consistency):
163
Results (85% consistency):
164
Model 22: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, male_gender)
Truth table after delete and code (75% consistency):
165
Results (75% consistency):
166
Results (80% consistency):
167
Results (85% consistency):
Model 23: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, english_languag)
168
Truth table after delete and code (75% consistency):
Results (75% consistency):
169
Results (80% consistency):
170
Results (85% consistency):
Model 24: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, white_race)
Truth table after delete and code (75% consistency):
171
172
Results (75% consistency):
173
Results (80% consistency):
Results (85% consistency): There are no positive cases in the truth table.
Model 25: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, discharged_home)
174
Truth table after delete and code (75% consistency):
175
Results (75% consistency):
176
Results (80% consistency):
177
Results (85% consistency):
Model 26:
178
Truth table after delete and code (75% consistency):
179
Results (75% consistency):
180
Results (80% consistency):
181
Results (85% consistency):
Model 27: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, english_languag)
182
Truth table after delete and code (75% consistency):
183
Results (75% consistency):
184
Results (80% consistency):
185
Results (85% consistency):
Model 28: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, white_race, discharged_home)
186
Truth table after delete and code (75% consistency):
187
Results (75% consistency):
188
Results (80% consistency):
189
Results (85% consistency):
Model 29: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, white_race, male_gender)
190
Truth table after delete and code (75% consistency):
191
Results (75% consistency):
192
Results (80% consistency):
193
Results (85% consistency):
Model 30: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, white_race, english_languag)
Truth table after delete and code (75% consistency):
194
Results (75% consistency):
195
Results (80% consistency):
196
Results (85% consistency):
Model 31: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, discharged_home)
197
Truth table after delete and code (75% consistency):
198
Results (75% consistency):
199
Results (80% consistency):
200
Results (85% consistency):
Model 32~: f(homehealth_in_4, rtmr_in_48hour, md_in_7days, call_in_7days,
discharged_home, english_languag)
201
Truth table after delete and code (75% consistency):
Results (75% consistency):
202
Results (80% consistency):
Results (85% consistency):
203
Model 33: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, male_gender, english_languag)
Truth table after delete and code (75% consistency):
204
Results (75% consistency):
205
Results (80% consistency):
206
Results (85% consistency):
Model 34: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, white_race, discharged_home)
207
Truth table after delete and code (75% consistency):
Results (75% consistency):
208
209
Results (80% consistency):
Results (85% consistency): No positive cases.
Model 35~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race, male_gender)
210
Truth table after delete and code (75% consistency):
211
Results (75% consistency):
212
Results (80% consistency):
213
Results (85% consistency):
Model 36~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, white_race, english_languag)
214
Truth table after delete and code (75% consistency):
215
Results (75% consistency):
216
Results (80% consistency):
Results (85% consistency): No positive cases.
Model 37~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, discharged_home, male_gender)
217
Truth table after delete and code (75% consistency):
218
Results (75% consistency):
219
Results (80% consistency):
220
Results (85% consistency):
Model 38~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, discharged_home, english_languag)
221
Truth table after delete and code (75% consistency):
Results (75% consistency):
222
Results (80% consistency):
Results (85% consistency):
223
Model 39~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, older_age, male_gender, english_languag)
Truth table after delete and code (75% consistency):
224
Results (75% consistency):
225
Results (80% consistency):
Results (85% consistency): No positive cases.
226
Model 40~: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour,
md_in_7days, call_in_7days, white_race, discharged_home, male_gender)
Truth table after delete and code (75% consistency):
227
Results (75% consistency):
228
Results (80% consistency):
229
Results (85% consistency):
Model 41~ ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, white_race, discharged_home, english_languag)
230
Truth table after delete and code (75% consistency):
231
Results (75% consistency):
232
Results (80% consistency):
233
Results (85% consistency):
Model 42: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, white_race, male_gender, english_languag)
234
Truth table after delete and code (75% consistency):
235
Results (75% consistency):
236
Results (80% consistency):
237
Results (85% consistency):
Model 43: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, discharged_home, male_gender, english_languag)
238
Truth table after delete and code (75% consistency):
239
Results (75% consistency):
240
Results (80% consistency):
241
Results (85% consistency):
Model 44: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, white_race, discharged_home, male_gender)
242
Truth table after delete and code (75% consistency):
243
Results (75% consistency):
244
Results (80% consistency):
245
Results (85% consistency):
Model 45: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, white_race, discharged_home, english_languag)
246
Truth table after delete and code (75% consistency):
247
Results (75% consistency):
248
Results (80% consistency):
Results (85% consistency): No positive cases.
Model 46: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, white_race, male_gender, english_languag)
249
Truth table after delete and code (75% consistency):
250
Results (75% consistency):
251
Results (80% consistency):
Results (85% consistency):
252
Model 47: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
253
Results (75% consistency):
254
Results (80% consistency):
Results (85% consistency): No positive cases.
Model 48: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, white_race, discharged_home, male_gender, english_languag)
255
Truth table after delete and code (75% consistency):
256
Results (75% consistency):
257
Results (80% consistency):
258
Results (85% consistency):
Model 49: ~readmit_in_30 = f(homehealth_in_4, older_age, white_race,
discharged_home, male_gender, english_languag)
259
Truth table after delete and code (75% consistency):
260
Results (at 75% consistency): No negative cases.
Results (at 80% consistency):
Results (at 85% consistency): No positive cases.
261
Model 50: ~readmit_in_30 = f(rtmr_in_48, older_age, white_race,
discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
Results (75% consistency): No negative cases.
262
Results (80% consistency):
Results (85% consistency): No positive cases.
263
Model 51: ~readmit_in_30 = f(md_in_7days, older_age, white_race,
discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
264
Results (75% consistency):
265
Results (80% consistency):
Results (85% consistency): No positive cases.
266
Model 52~: ~readmit_in_30 = f(call_in_7days, older_age, white_race,
discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
267
Results (75% consistency):
268
Results (80% consistency:
269
Results (85% consistency):
Model 53: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, white_race, discharged_home, male_gender,
english_languag)
270
Truth table after delete and code (75% consistency):
Results (75% consistency):
271
Results (80% consistency):
272
Results (85% consistency):
Model 54: ~readmit_in_30 = f(homehealth_in_48, rtmr_in_48, md_in_7days,
call_in_7days, older_age, white_race, discharged_home, male_gender) – non-
English speaking patients only (n=833)
Truth table after delete (8) and code (80%):
273
Results (75% consistency):
274
Results (80% consistency):
275
Results (85% consistency):
r
Results (90% consistency): No positive cases.
276
Model 55: ~readmit_in_30 = f(homehealth_in_4, rtmr_in_48hour, md_in_7days,
call_in_7days, older_age, white_race, male_gender, english_languag) – patients
discharged home or to another community setting (n=8269)
Truth table after delete (80) and code (75%):
277
Results (75% consistency):
278
Results (80% consistency):
Results (85% consistency):
279
Model 56: Hospital A
Stratified only to include discharges from hospital A (n=799) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
280
Results (75% consistency):
281
Results (80% consistency):
282
Results (85% consistency):
283
Results (90% consistency):
Model 56: Hospital B
Stratified only to include discharges from hospital B (n=769) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
284
285
Results (75% consistency):
286
Results (80% consistency):
287
Results (85% consistency):
288
Results (90% consistency):
Model 56: Hospital C
Stratified only to include discharges from hospital C (n=631) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
289
290
Results (75% consistency):
291
Results (80% consistency):
292
Results (85% consistency):
Results (90% consistency):
293
Model 56: Hospital D
Stratified only to include discharges from hospital D (n=1145 ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
Results (75% consistency):
294
295
Results (80% consistency):
296
Results (85% consistency):
297
Results (90% consistency):
Model 56: Hospital E
Stratified only to include discharges from hospital E (n=706) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
298
Truth table after delete and code (75% consistency):
299
Results (75% consistency):
300
Results (80% consistency):
301
Results (85% consistency):
Results (90% consistency):
302
Model 56: Hospital F
Stratified only to include discharges from hospital F (n=1155) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
303
Results (75% consistency):
304
Results (80% consistency):
305
Results (85% consistency):
Results (90% consistency):
306
Model 56: Hospital G
Stratified only to include discharges from hospital G (n=33) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
Results (75% - 90% consistency):
307
Model 56: Hospital H
Stratified only to include discharges from hospital H (n=595) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
308
Results (75% consistency):
309
Results (80% consistency):
310
Results (85% consistency):
311
Results (90% consistency):
Model 56: Hospital I
Stratified only to include discharges from hospital I (n=676) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
312
Truth table after delete and code (75% consistency):
313
Results (75% consistency):
Results (80% consistency):
314
315
Results (85% consistency):
316
Results (90% consistency):
Model 56: Hospital J
Stratified only to include discharges from hospital J (n=1211) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
317
Truth table after delete and code (75% consistency):
Results (75% consistency):
318
Results (80% consistency):
Results (85% consistency):
319
Results (90% consistency):
Model 56: Hospital K
Stratified only to include discharges from hospital K (n=613) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
320
Truth table after delete and code (75% consistency):
321
Results (75% consistency):
322
Results (80% consistency):
Results (85% consistency):
323
Results (90% consistency):
Model 54: Hospital L
Stratified only to include discharges from hospital L (n=580) ~readmit_in_30 =
f(homehealth_in_4, rtmr_in_48hours, md_in_7days, call_in_7days, older_age,
white_race, discharged_home, male_gender, english_languag)
Truth table after delete and code (75% consistency):
324
Results (75% consistency):
325
Results (80% consistency):
326
Results (85% consistency):
Results (90% consistency):
Abstract (if available)
Abstract
If US health care organizations did not already deem readmissions reduction an important undertaking, the Health Care Accountability and Affordability Act has made it so. Over two thousand hospitals are already being penalized for having higher than average readmission rates for acute myocardial infarction, pneumonia, and heart failure. I conducted a two-part study across a southern California health system to understand how health care providers can most effectively reduce heart failure readmissions with constrained finances and found a static approach that provides every patient with the same intervention at every discharge is neither optimal for patients nor an effective use of resources. I propose a patient-centered framework for reducing heart failure readmissions that begins with understanding each patient and their family, and collaborating with them to determine the optimal care plan based on multiple potential interventions. Furthermore I discuss the importance delivering these interventions with fidelity to achieve the best patient outcomes, and provide guidance for designing an effective heart failure readmission reduction program with engaged leaders, program oversight that includes leaders and frontline staff, and well designed processes that focus on the patient and encourage continuous improvement. I believe as we get to know our patients and plan care with them in the room, we too will learn more about how we can provide the most effective care.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Improving hospital readmission rates: the benefit of a transitional case management program
PDF
Investigating the effectiveness of a social work intervention on reducing hospital readmissions among older adults
PDF
Improvement of health care delivery in America: medical office compliance certification system implementation
PDF
Delivering better care for children with special health care needs: analyses of patient-centered medical home and types of insurance
PDF
Impact of pharmacy-based transitional care on healthcare utilization and costs
PDF
Developing an agent-based simulation model to evaluate competition in private health care markets with an assessment of accountable care organizations
PDF
Designing health care provider payment systems to reduce potentially preventable medical needs and patient harm: a simulation study
PDF
A series of longitudinal analyses of patient reported outcomes to further the understanding of care-management of comorbid diabetes and depression in a safety-net healthcare system
PDF
A public sector organizational change model: prioritizing a community-focused, inclusive, and collaborative approach to strategic planning
PDF
Green healthcare, an environmentally sustainable methodology: an investigation of the ecological impacts of the healthcare industry and the role of green initiatives in sustainable medical services
PDF
Clinical prediction models to forecast depression in patients with diabetes and applications in depression screening policymaking
PDF
Effect of α₁-adrenergic antagonist use on clinical outcomes in patients with heart failure
PDF
Implicit provider bias in cardiovascular disease care of Black women: the lives of Black women with heart disease matter
PDF
Intersect policing: bringing CompStat to the field level to reduce the fear and incidence of crime
PDF
The impact of public expenditures on health care on total health expenditures: an exploratory analysis of selected OECD countries
PDF
A system framework for evidence based implementations in a health care organization
PDF
Topics in U.S. healthcare
PDF
Simulation modeling to evaluate cost-benefit of multi-level screening strategies involving behavioral components to improve compliance: the example of diabetic retinopathy
PDF
Using a human factors engineering perspective to design and evaluate communication and information technology tools to support depression care and physical activity behavior change among low-inco...
PDF
Research to develop a manual for parents, caregivers, and volunteers to teach children from birth to five years of age with intellectual disability living in rural communities in the Dominican Re...
Asset Metadata
Creator
Garofalo, Lynn Marie
(author)
Core Title
Planning care with the patient in the room: a patient-focused approach to reducing heart failure readmissions
School
School of Policy, Planning and Development
Degree
Doctor of Policy, Planning & Development
Degree Program
Policy, Planning, and Development
Publication Date
01/22/2013
Defense Date
12/17/2012
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
health care,heart failure,OAI-PMH Harvest,patient-centered care,readmissions
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Myrtle, Robert C. (
committee chair
), Adler, Paul S. (
committee member
), Belson, David (
committee member
), Eytan, Ted (
committee member
), Schilling, Lisa (
committee member
), Wu, Shinyi (
committee member
)
Creator Email
lynn.g.wright@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-129272
Unique identifier
UC11292799
Identifier
usctheses-c3-129272 (legacy record id)
Legacy Identifier
etd-GarofaloLy-1401.pdf
Dmrecord
129272
Document Type
Dissertation
Rights
Garofalo, Lynn Marie
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
heart failure
patient-centered care
readmissions