Reducing Heart Failure Hospital Readmissions: A Systematic Review of Disease Management Programs

Articles © The authors | Journal compilation © Cardiol Res and Elmer Press Inc™ | www.cardiologyres.org
This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction
in any medium, provided the original work is properly cited
126
Review Cardiol Res. 2014;5(5):126-138 ressElmer
Reducing Heart Failure Hospital Readmissions: A
Systematic Review of Disease Management Programs
Janardhana Gorthia, Claire B. Huntera, Ayran N. Mooss a, Venkata M. Alla a, Daniel E. Hillemana, b
Abstract
The recent enactment of the Patient Protection and Affordable Care
Act which established the federal Hospital Readmissions Reduction
Program (HRRP) has accelerated efforts to develop heart failure (HF)
disease management programs (DMPs) that reduce readmissions in
patients hospitalized for HF. This systematic review identified ran-
domized controlled trials of HF DMPs which included home care,
outpatient clinic interventions, structured telephone support, and
non-invasive and invasive telemonitoring. These different types of
DMPs have been associated with conflicting results. No specific type
of DMP has produced consistent benefit in reducing HF hospitaliza-
tions. Although probably effective at reducing readmissions, home
visits and outpatient clinic interventions have substantial limitations
including cost and accessibility. Telemanagement has the potential
to reach a large number of patients at a reasonable cost. Structured
telephone support follow-up has been shown to significantly reduce
HF readmissions, but does not significantly reduce all-cause mortality
or all-cause hospitalization. A meta-analysis of 11 non-invasive te-
lemonitoring studies demonstrated significant reductions in all-cause
mortality and HF hospitalizations. Invasive telemonitoring is a poten-
tially effective means of reducing HF hospitalizations, but only one
study using pulmonary artery pressure monitoring was able to dem-
onstrate a reduction in HF hospitalizations. Other studies using inva-
sive hemodynamic monitoring have failed to demonstrate changes in
rates of readmission or mortality. The efficacy of HF DMPs is associ-
ated with inconsistent results. Our review should not be interpreted to
indicate that HF DMPs are universally ineffective. Rather, our data
suggest that one approach applied to a broad spectrum of different pa-
tient types may produce an erratic impact on readmissions and clini-
cal outcomes. HF DMPs should include the flexibility to meet the
individualized needs of specific patients.
Keywords: Heart failure; Hospitalizations; Heart failure clinics; Tele-
Manuscript accepted for publication October 24, 2014
aThe Creighton University Cardiac Center, Creighton University School of
Medicine, Omaha, NE, USA
bCorresponding Author: Daniel E. Hilleman, Creighton University Cardiac
Center, 3006 Webster Street, Omaha, NE 68131, USA.
Email: [e-mail]
doi: http://dx.doi.org/10.14740/cr362w
management; Telemonitoring
Introduction
The medical and financial burden of heart failure (HF) hos-
pitalizations has led to a substantive body of research charac-
terizing the timing and etiology of readmissions, identifying
methods that predict readmission, and evaluating strategies
that reduce readmissions. Findings from epidemiologic sur-
veys of HF patients indicate that 30% of readmissions occur
during the first 2 months after hospital discharge, 50% of re-
admissions occur within the last 2 months prior to death, and
the remaining 20% of readmissions occur between these time
periods [1, 2]. This pattern of readmissions has been referred
to as the “three-phase terrain” of HF readmissions [3].
The Patient Protection and Affordable Care Act estab-
lished the federal Hospital Readmissions Reduction Program
(HRRP) through which Medicare payments to hospitals that
have excess readmissions following an admission for HF, my-
ocardial infarction, or pneumonia would be reduced [4]. The
HRRP took effect on October 1, 2012 using claims data from
July 2008 through June 2011. The CMS defines a readmission
as any hospital admission that occurs within 30 days of a dis-
charge from the same or other hospital [5]. Excess readmis-
sions are calculated by comparing a hospital’s rate of readmis-
sion for an applicable condition against the national average
for similar hospitals. For fiscal year 2013, excessive readmis-
sions can result in a maximal loss of up to 1% of Medicare re-
imbursement for the coming year [6]. The HRRP is expanding
in 2015 to include readmissions for chronic obstructive pulmo-
nary disease, coronary artery bypass graft surgery, percutane-
ous coronary interventions, and other vascular interventions
with penalties increasing to a maximum payment withholding
of 3% [7].
The ability of HF disease management programs (DMPs)
to routinely reduce all-cause hospital readmissions at 30 days
has not been documented. Many HF DMPs have reported
morbidity and/or mortality outcomes or have used different
follow-up time points [3]. Many programs have not focused
on clinical outcomes or reductions in unplanned healthcare
contacts, but rather have evaluated the rate at which a DMP

Articles © The authors | Journal compilation © Cardiol Res and Elmer Press Inc™ | www.cardiologyres.org 127
Gorthi et al Cardiol Res. 2014;5(5):126-138
has been successful in changing the process of care in HF pa-
tients [8]. Since the inception of the HRRP, greater emphasis
has been place on HF DMPs [9].
The purpose of the present systematic review was to criti-
cally evaluate all available studies meeting minimal inclusion
criteria to define the efficacy of DMPs in reducing hospitali-
zations and/or mortality in patients with chronic HF. The re-
cent HRRP initiative has provided substantial motivation to
minimize hospital readmissions in patients discharged with a
diagnosis of HF.
Methods
Studies were identified using the guidelines defined by the
Cochrane Handbook or Systemic Reviews and the Meta-anal-
ysis of Observational Studies in Epidemiology (MOOSE) [10,
11]. The on-line databases of PubMed (Medline), EBSCOHost,
and the Cochrane Library were searched from January 1975
through August 2014 for studies reporting the outcomes of HF
DMPs. The medical subject heading terms used in the search
included HF DMPs, HF, hemodynamics, structured telephone
support, telemonitoring, telemanagement, and implantable
hemodynamic devices. A manual search of the bibliographies
of the identified reports and reviews was also performed.
Only studies published in English were included in the
analysis. Studies published only as abstracts were excluded.
Only prospective, randomized studies including a minimum
of 50 patients were included. Eligible studies had to report
either hospitalizations (all-cause or heart failure specific) or
mortality (all-cause or cardiovascular). Efficacy was based on
study reported outcomes concerning hospital readmissions or
mortality comparing the intervention and control or usual care
treatment arms. Studies using pre- and post-disease manage-
ment intervention analyses were excluded. Studies that were
published as preliminary reports that were subsequently re-
ported in a later publication with a larger sample size were
not included in this analysis. In addition, studies reporting on
patients with disease states other than HF which did not report
outcomes for HF patients separately from other patient types
were also excluded.
Results
In-home care interventions
A total of eight randomized controlled studies meeting eligi-
bility criteria evaluating the efficacy of in-home visits as part
of an HF DMP were identified (Table 1) [12-19]. One study
included only one home visit and two others included only two
home visits [13, 14, 17]. Six studies compared home visit in-
terventions to usual care, one study compared home visits in
addition to outpatient visits to usual care, and one study com-
pared home visits with outpatient visits. Of the seven studies
comparing home visits to usual care, three were associated
with a significant improvement in the primary outcome [14-
Table 1. Randomized Trials of Heart Failure Disease Management Programs Using Home Visits
Reference Duration of
intervention
Duration of
follow-up
No. of control
patients
No. of interven-
tion patients Primary outcome
Results
Primary
outcome ACM ACH HFH
Rich et al 1995 [12] 3 months 3 months 140 142 90 days ACH-free survival 0 0 + +
Jaarsma et al 1999 [13] 1 visit 9 months 95 84 ACH 0 0 0 0
Blue et al 2001 [14] 12 months 12 months 81 84 ACH plus HFH + 0 0 +
Harrison et al 2002 [15] 2 weeks 5 months 100 92 QOL + 0 0 0
Stewart and Horow-
itz 2002 [16]
6 months 6 months 148 149 ACH plus ACM + 0 0 0
Holland et al 2007 [17] 6 months 6 months 144 149 ACH 0 0 0 NR
Jaarsma et al 2008 [18] 18 months 18 months 348 701 ACM plus HFH 0 0 0 0
ACM: all-cause mortality; ACH: all-cause hospitalization; HFH: heart failure hospitalization; 0: not significant; +: significant; NR: not reported.

Articles © The authors | Journal compilation © Cardiol Res and Elmer Press Inc™ | www.cardiologyres.org128
Reducing Heart Failure Hospital Readmissions Cardiol Res. 2014;5(5):126-138
16]. None of the studies were able to demonstrate a significant
reduction in all-cause mortality. One study was able to dem-
onstrate a reduction in all-cause hospitalization which was
driven by a reduction in HF hospitalizations [12]. Two studies
significantly reduced HF hospitalizations [12, 14]. Three stud-
ies evaluating home visits failed to demonstrate a significant
improvement in hospitalization or mortality [13, 17, 18].
In the largest published study to incorporate home visits
into the disease management intervention, home visits had no
favorable impact on outcomes [18]. The Coordinating Study
Evaluating Outcomes of Advising and Counseling in Heart
Failure (COACH) randomized 1,023 patients with NYHA
class II/III HF to one of three interventions including a control
group (n = 339), a basic support group (n = 340), and an inten-
sive support group (n = 344). All three interventions included
four visits to a cardiologist over an 18-month follow-up period
after an HF hospital discharge. The basic support intervention
included nine additional visits to an HF specialist nurse at an
outpatient clinic. The intensive support intervention included
18 additional visits to an HF specialist nurse at an outpatient
clinic, two home visits by the nurse specialist with one occur-
ring in the first month after discharge, and two multidiscipli-
nary advice sessions. The usual care group included only the
four outpatient visits to a cardiologist. The primary endpoint
of the composite of HF readmission or all-cause mortality oc-
curred in 141 (42%) control patients, 138 (38%) patients in the
basic support group, and 132 (38%) patients in the intensive
support group. Analysis of the time to the first event deter-
mined hazard ratios of 0.96 (95% CI 0.76 - 1.21; P = 0.73)
and 0.93 (95% CI 0.73 - 1.17; P = 0.53) for the composite
outcome comparing basic and intensive support against the
control group. All-cause mortality and hospitalizations were
not different among the patients randomized to the three inter-
ventions. The frequency of healthcare contacts initiated by the
patient was greater than prescribed in the protocol in all three
interventions. This was the greatest in the basic support group
where the increase in healthcare contacts was 40% while the
increase in the control group was 33%. The increase was only
10% greater than prescribed in the intensive support group.
The most recently published trial including home visits
was a randomized comparison against patients who were seen
in a walk-in specialty HF clinic. The WHICH (Which Heart
Failure Intervention Is Most Cost-Effective & Consumer
Friendly in Reducing Hospital Care) study randomized 143 pa-
tients to a home-based intervention (HBI) and 137 patients to
a specialized HF clinic-based intervention (CBI) with a 12- to
18-month follow-up [19]. The primary outcome was the com-
posite of all-cause unplanned hospitalizations or death. Since
there was no control group in this study, conclusions about the
relative effectiveness of the either DMP cannot be reached.
There was no significant difference in the primary composite
outcome between the HBI (71%) and the CBI (76%) (adjusted
hazard ratio 0.97; 95% CI 0.73 - 1.30; P = 0.86). There were
also no significant differences in unplanned hospitalizations
between the HBI (67%) and the CBI (69%) (P = 0.88) or in
all-cause mortality between the HBI (22%) and the CBI (28%)
(P = 0.25). Patients in the HBI group did have a significantly
shorter median duration of days during hospitalizations. The
Table 2. Randomized Trials of Heart Failure Disease Management Programs Using Outpatient Visits
Reference Duration of
intervention
Duration of
follow-up
No. of control
patients
No. of interven-
tion patients
Primary
outcome
Results
Primary
outcome ACM ACH HFH
Cline et al 1998 [20] 12 months 12 months 110 80 Time to
readmission
+ 0 0 NR
Ekman et al 1998 [21] 6 months 6 months 79 79 ACH plus ACM 0 0 0 0
Kasper et al 2002 [22] 6 months 6 months 98 102 ACM plus HFH 0 0 0 0
Doughty et al 2002 [23] 12 months 12 months 97 100 ACH plus ACM 0 0 + 0
Ledwidge et al 2002 [24] 3 months 3 months 47 51 Cost benefit + 0 + +
Capomolla et al 2002 [25] 12 months 12 months 122 112 Cost utility + + + NR
Stromberg et al 2003 [26] 12 months 12 months 54 52 ACM plus ACH + + + 0
de la Porte et al 2007 [27] 12 months 12 months 122 118 ACM plus HFH + 0 + +
Powell et al 2010 [28] 12 months 31 months 451 451 ACM plus HFH 0 0 0 0
ACM: all-cause mortality; ACH: all-cause hospitalization; HFH: heart failure hospitalization; 0: not significant; +: significant; NR: not reported.

Articles © The authors | Journal compilation © Cardiol Res and Elmer Press Inc™ | www.cardiologyres.org 129
Gorthi et al Cardiol Res. 2014;5(5):126-138
median duration of hospital length of stay with HBI was 4.0
days (interquartile range of 2.0 - 7.0 days) compared to 6.0
days (interquartile range 3.5 - 13 days) with CBI (P = 0.004).
Although the HBI was not associated with a significant im-
provement in the primary outcome compared to the CBI, the
shorter hospital stay with HBI was associated with a lower
overall healthcare cost (P = 0.03). The costs of providing the
patient interventions were not significantly different between
HBI ($1,813 per patient) and CBI ($1,829 per patient).
Outpatient visit interventions
A total of 11 randomized controlled studies meeting eligibil-
ity criteria evaluating the efficacy of outpatient clinic visits
as part of an HF DMP were identified (Table 2) [18-28]. Two
of these trials were previously discussed: the COACH study
which found no benefit of frequent visits to a nurse specialist in
an outpatient setting compared to usual care and the WHICH
study comparing HBI and CBI [18, 19]. Of the remaining
nine studies, the primary outcome was significantly improved
in five studies [20, 24-27]. However, only two of these stud-
ies used hospitalizations or mortality in the primary outcome
[26, 27]. Three other studies achieved a statistically significant
improvement in their primary outcome [20, 24, 25]. The pri-
mary outcomes in these studies were time to readmission, cost-
benefit, and cost-utility. All-cause mortality was significantly
reduced in two studies, but one of these studies only enrolled
a total of 106 patients [25, 26]. Of the seven studies reporting
HF-related readmissions, only two significantly reduced those
events. The most consistent effect found in the studies utilizing
outpatient clinic visits was a significant reduction in all-cause
hospitalization which was achieved in five of the nine studies.
In the largest study using outpatient clinic visits, the Heart
Failure Adherence and Retention Trial (HART), 902 patients
with NYHA class II/III HF were randomized to one of two
interventions [28]. The self-management plus education inter-
vention included 18 two-hour group meetings offered over the
first year after randomization. The HF education alone group
received 18 “Heart Failure Tip Sheets” mailed on the same
schedule as the group meetings. Telephone calls were made
within 2 - 3 days after each mailing to ensure receipt and com-
prehension. Patients were followed for a minimum of 2 years
(1 year of treatment and 1 year of post-treatment follow-up).
The rate of the primary composite outcome of HF hospitaliza-
tion plus all-cause mortality was not different in the self-man-
agement plus education group (163 events, 40%) compared
to the education alone group (171 events, 41%) after a mean
follow-up of 2.56 years (odd ratio 0.95; 95% CI 0.72 - 1.26).
There were also no significant differences in the secondary
endpoints of death, HF hospitalization, all-cause hospitaliza-
tion, or quality of life.
Structured telephone support interventions
Disease management interventions relying on outpatient or
home visits are resource intensive, costly, and are limited in
the numbers of patients that can be impacted. This is especially
true for patients in geographically remote areas or those with
transportation limitations. Telemanagement using phone calls
or the more complex transmission of patient-related clinical
data (telemonitoring) over telephone or internet connections
have the potential to reach unlimited numbers of HF patients.
A total of 13 randomized controlled studies meeting eli-
gibility criteria evaluating the efficacy of structured telephone
support as part of an HF DMP were identified (Table 3) [29-
41]. All but two studies used hospitalization or mortality in the
primary efficacy outcome [33, 34]. In these two studies, time
to hospitalization for HF and medication adherence were the
primary outcomes, and neither achieved their primary efficacy
endpoint. In the 11 studies using hospitalization, mortality, or
both as the primary efficacy endpoint, four studies achieved
their primary efficacy endpoint [29, 30, 35, 37]. Two studies
were associated with a significant reduction in all-cause mor-
tality, one study was associated with a significant reduction in
all-cause hospitalization, and four studies were associated with
a significant reduction in HF hospitalizations [29-31, 35-37,
39]. A 2007 meta-analysis which pooled the results of 10 stud-
ies of structured telephone support concluded that telephone
follow-up significantly reduced HF readmissions, but did not
significantly reduce all-cause mortality or all-cause hospitali-
zation [42]. Two of the structured telephone support studies
were randomized comparisons against non-invasive telemoni-
toring DMPs [36, 40]. These studies are discussed further un-
der the non-invasive telemonitoring intervention section.
Non-invasive telemonitoring interventions
A total of 14 randomized controlled studies meeting eligibility
criteria evaluating the efficacy of non-invasive telemonitoring
support as part of an HF DMP were identified (Table 4) [36,
40, 43-54]. Thirteen of the 14 studies used a primary efficacy
endpoint that included hospitalizations, mortality, or both. The
lone study that did not include hospitalizations or mortality in
the primary outcome used changes in b-type naturetic peptide
levels and quality of life [54]. This study did demonstrate a
significant improvement in both of the primary endpoints us-
ing a mobile-phone-based telemonitoring system.
Of the remaining 13 studies, three achieved their primary
efficacy endpoint [44, 46, 51]. Twelve of the 13 studies report-
ed the effect of the DMP on cardiac or all-cause mortality with
only two studies demonstrating a significantly positive effect
on this outcome [36, 43]. Three studies significantly reduced
all-cause hospitalizations [44, 46, 50]. Ten of the 14 studies re-
ported rates of HF hospitalizations with three of the 10 studies
demonstrating significant reductions in these hospitalizations
[44, 50, 51].
A Cochrane database review conducted a meta-analysis
published in 2010 which included a total of 27 controlled
studies including 11 studies using non-invasive telemonitor-
ing (2,710 patients) and 16 studies using structured telephone
support (5,613 patients) [55]. All-cause mortality was signifi-
cantly reduced by non-invasive telemonitoring (RR 0.66; 95%
CI 0.54 - 0.81; P < 0.001). While structured telephone support