Heart Failure Reviews (2020) 25:231–243

https://doi.org/10.1007/s10741-019-09801-5

Effectiveness of telemedicine systems for adults with heart failure:

a meta-analysis of randomized controlled trials
Ye Zhu 1,2 & Xiang Gu 1,2 & Chao Xu 3

Published online: 24 May 2019

# The Author(s) 2019

Abstract
Despite favorable effects from telemedicine (TM) on cardiovascular diseases, outcome and comparative impact of TM on heart
failure (HF) adults remain controversial. A meta-analysis was conducted to summarize the evidence from existing randomized
controlled trials (RCTs) which compared potential impact of TM on HF with conventional healthcare. TM mainly included
structure telephone support (STS), involving interactive vocal response monitoring and telemonitoring. PubMed, MEDLINE,
EMBASE, and the Cochrane Library were searched to identify RCTs to fit our analysis (1999 to 2018). Odds ratio (OR) with its
95% confidence interval (CI) was used. Sensitivity analysis, subgroup analysis, and tests for publication bias were conducted.
Heterogeneities were also evaluated by I2 tests. A total of 29 RCTs consisting of 10,981 HF adults were selected for meta-level
synthesis, with follow-up range of 1–36 months. Telemonitoring is associated with the reduction in total number of all-cause
hospitalization (OR 0.82, 95% CI 0.73–0.91, P = 0.0004) and cardiac hospitalization (OR 0.83, 95% CI 0.72–0.95, P = 0.007).
Telemonitoring resulted in statistically significant risk reduction of all-cause mortality (OR 0.75, 95% CI 0.62–0.90, P = 0.003).
However, the OR of HF-related mortality (OR 0.84, 95% CI 0.61–1.16, P = 0.28) is not significantly distinguishable from that of
conventional healthcare. Receiving STS interventions is likely to reduce the hospitalization for all causes (OR 0.86, 95% CI
0.78–0.96, P = 0.006, I2 = 6%) and the hospitalization due to HF (OR 0.74, 95% CI 0.65–0.85, P < 0.0001, I2 = 0%), compared
with interventions from conventional healthcare. OR of all-cause STS mortality (OR 0.96, 95% CI 0.83–1.11, P = 0.55) was
identified in meta-analyses of eight cases. OR of STS cardiac mortality (OR 0.54, 95% CI 0.34–0.86, P = 0.009) was identified in
meta-analyses of three cases. This work represents the comprehensive application of network meta-analysis to examine the
comparative effectiveness of telemedicine interventions in improving HF patient outcomes. Compared with conventional
healthcare, telemedicine systems with medical support prove to be more effective for HF adults, particularly in reducing all-
cause hospitalization, cardiac hospitalization, all-cause mortality, cardiac mortality, and length of stay. While further research is
required to confirm these observational findings and identify optimal telemedicine strategies and the duration of follow-up for
which it confers benefits.

Keywords Meta-analysis . Heart failure . Telemedicine . Cardiovascular disease . Randomized controlled trials

Introduction quality of life (QoL) of HF patients while increasing the bur-

den of morbidity, mortality, and healthcare. As the result of the
Heart failure (HF) is one of the most prevalent manifestations rising demand for acute hospital beds for HF, strategies to
of cardiovascular disorders [1], decreasing the health-related facilitate early discharge and reduce unplanned readmissions
are superior to improving patient outcomes and resource us-
age. Poor adherence to recommendations provided by
* Xiang Gu
[email protected]
healthcare professional is responsible for ~ 50% of HF hospi-
talization [2]. The overall purpose of education and other in-
1
tervention modalities, for instance, structured telephone sup-
Clinical Medical College of Yangzhou University,
port (STS), home visits, and nurse-led clinics, is to improve
Yangzhou 225001, Jiangsu, China
2
self-care and patients’ adherence. In recent years, promising
Department of Cardiology, Northern Jiangsu People’s Hospital,
results have been reported for multidisciplinary care strategies
Yangzhou, China
3
for HF patients with and without telemedicine systems. A
Department of Biostatistics and Epidemiology, University of
meta-analysis has shown positive results for home-based
Oklahoma Health Science Center, Oklahoma City, OK 73104, USA
232 Heart Fail Rev (2020) 25:231–243

interventions, including a reduction of 34% in all-cause mor- assessed. Most of the disagreements were resolved through
tality and a reduction of 30–56% in HF hospitalizations [3]. discussions. When there was any disagreement, the third au-
However, following prospective randomized multicenter clin- thor mediated the discussion to gain consensus.
ical trials of non-invasive approaches cannot confirm these
findings for morbidity-related and mortality-related endpoints Inclusion criteria
[4, 5]. Although earlier studies suggest a reduction in mortal-
ity, results of the study reported by Chaudhry and colleagues & Studies were included in the meta-analysis if they were
failed to show any beneficial effects of telemonitoring [6]. RCTs.
To determine whether telemedicine systems improve out- & Patients were objectively confirmed to have symptomatic
comes for adults following an unplanned admission due to HF (New York Heart Association [NYHA] Class I–IV)
HF, we conducted a meta-analysis of existing randomized characterized by impaired left ventricular function (left
controlled trials (RCTs) comparing different forms of telemed- ventricular ejection fraction < 45%).
icine systems with conventional healthcare after discharge. & Includes both an experimental group and a control group.
Interventions included conventional healthcare and the fol- & Telemedicine treatments include telephone support,
lowing forms of telemedicine: STS involving regular follow- telemonitoring involving interactive vocal response mon-
up calls between the health professional and the patient, itoring, and monitoring by ECG.
telemonitoring systems involving the transmission of informa- & Control group only receives conventional healthcare de-
tion on symptoms and signs (TM), and telemedicine systems fined as a guideline-based standard care with scheduled
involving interactive vocal response monitoring and electro- clinic visits but without any additional interventions.
cardiographic transmissions (ECG). The primary endpoints of & The primary outcome measurements include all-cause mor-
this study were all-cause hospitalization and all-cause mortal- tality and all-cause hospitalization (defined as an admission
ity during follow-up. As secondary endpoints, we explored to a healthcare facility for less than 24 h for all causes).
cardiac hospitalization, mortality, and length of stay. & The secondary outcome measurements include cardiac
hospitalization (defined as an admission to a healthcare
facility for less than 24 h due to heart failure), cardiac
Materials and methods mortality, length of hospital stay, health-related QoL, and
hospitalization costs.
Search strategy

A search of literature was performed for RCTs of non-invasive Exclusion criteria

telemedicine systems for HF patients compared with conven-
tional healthcare. The PubMed, MEDLINE, EMBASE, and the Studies were excluded if they met the following criteria:
Cochrane Library databases were searched to extract articles
(1999 to 2018) on telemedicine systems in adults. Literature & Not being RCTs and non-English language papers
search was informed using keywords heart failure, cardiac fail- & Not involving patients with acute HF
ure, remote, telemedicine, telecommunication, telehealth, & Not reporting numerical data on the outcomes of interest
telecardiology, health information systems, internet, home & Published in the form of letters, congress abstracts, review
monitor, and interactive voice response. The meta-analysis of articles, comments, editorials, case reports, technical re-
RCTs was performed in accordance with the latest meta- ports, or animal studies
analysis guidelines (PRISMA) [7]. Referenced studies and nar-
rative reviews were searched as well.
Data extraction
Study selection
A predesigned data abstraction form was used to obtain data
Studies involved in this meta-analysis were sorted by three on relevant results of the study. Following terms were record-
independent authors. In the preliminary stage, all publication ed for each study: authors, years of publication, patient demo-
titles and abstracts were examined by the first two authors to graphic data (gender, age, disease severity, etc.), intervention
exclude non-pertinent articles clearly not meeting the follow- features, outcomes parameters, as well as the quality of includ-
ing inclusion criteria. The two authors reviewed the results of ed RCT studies. Detailed information would be requested
each study with a standardized data extraction tool and also from the author if some necessary original data could not be
applied standard scales to judge study quality and risk of bias acquired. Data were then tabulated and a network meta-
independently. If any doubt of suitability remained after the analysis (NMA) of the following outcomes was deemed ap-
abstract was examined, the full manuscript was retrieved and propriate: all-cause mortality, all-cause hospitalization,
Heart Fail Rev (2020) 25:231–243 233

cardiac hospitalization, and cardiac mortality. Studies satisfy- Results

ing the inclusion criteria were assigned a quality score based
on the revised 7-point Jadad scale [8]. The scale consists of Trial flow
four aspects: generation of allocation sequence (2 points), al-
location concealment (2 points), investigator blindness (2 As shown in Fig. 1, 388 citations were identified from our
points), and withdrawals and dropouts (1 point). A total score search (up to August 2018). Fifty-two duplication cross-
of less than 4 indicates low quality, while the one of more than databases were excluded. Three hundred eight were excluded
5 indicates high quality [9]. after examining titles and abstracts of full-text articles.
Reasons for exclusion were not related to HF, not RCT, unre-
lated to home-based telemonitoring/telephone support, no out-
Statistical analysis come of interest, or non-English language papers and so forth.
From the remaining articles, we identified 29 non-duplicated
Separate analyses were performed for each outcome’s odds RCTs and 10,981 patients eligible for the meta-analysis.
ratio (OR) or weighted mean difference (WMD) using the Details about the searching strategy and the flow chart for
Mantel–Haenszel method. Pairwise meta-analyses were con- the identification of studies used in the network meta-
ducted by combining studies comparing the same interven- analysis of telemedicine interventions for HF patients were
tions using a random-effects model. Meta-analysis inconsis- provided in Fig. 1.
tency was assessed by comparing the deviance and deviance
information criteria in fitted consistency and inconsistency Characteristics of included trials
models across studies [10]. Specifically, we investigated the
heterogeneity through examining both forest plots and General characteristics of the population, interventions, and
Cochran’s Q quantified by I2 tests [11]. An I2 of 0–25% indi- comparison groups included in the 29 RCTs along with the
cates no heterogeneity, an I2 of 25–50% indicates moderate main outcomes of each study were summarized in Table 1.
heterogeneity, an I2 of 50–75% indicates large heterogeneity, All the RCT studies were classified into two groups based on
and an I2 of 75–100% indicates extreme heterogeneity. the type of telemedicine intervention(s): telemonitoring (n = 19)
Results with a P value less than 0.05 and 95% confidence and telephone-supported systems (n = 9). Please note that only
intervals (CIs) exceeding 1 were considered as statistical sig- one study reported outcomes for both telemonitoring and
nificance. The analyses were carried out using telephone-supported care. The average duration of the interven-
Comprehensive Meta-Analysis techniques in Review tions was 10.5 months (range 1 to 36 months). For most of the
Manager (RevMan, version 5.2, The Cochrane studies (25 out of 29), the number of males was greater than that
Collaboration, London, England, 2012) [12]. The results from of females. Endpoints and adopted telemedicine strategies were
our network meta-analysis were qualitatively compared with similar among the selected studies. In 22 of 29 trials, partici-
direct, frequent, pairwise estimates. Publication bias was test- pants were followed for six or more months. Despite differ-
ed by funnel plots and the Egger and Begger tests using Stata ences in the scope and range of included studies, most RCTs
version 12.0 software (Stata Corporation, College Station, reported on a number of similar outcomes. Most frequently
TX, USA) and P < 0.05 was considered significant [13]. reported outcomes included all-cause hospitalization, cardiac

Fig. 1 Selection process of the

studies
Table 1 Description of included studies
234

Author/year Study population Population Type of interventions Follow- Outcome parameters Jadad
up score
N Age (year) Female (%) lengths

Ewa Hägglund/2006 72 75 ± 8 5 Sweden Home intervention versus usual care. 4 months Health-related quality of life (HRQoL), hospital 5
days due to HF
Silvia Soreca/2012 118 ≥ 70 49 Italy Clinical and electrocardiographic 18 months 1. Rehospitalization for worsening of heart failure 5
evaluations and periodic home symptoms and/or for the appearance of major
echocardiographic examinations versus usual care vascular events
2. Home-treated vascular events, cardiovascular
death, and the composite endpoint of
death plus rehospitalization
Abul Kashem/2006 36 56.1 ± 12.6 30.5 America Telemedicine arm versus usual care 8 months 1. Total hospital days 4
2. Effect of outpatient monitoring on duration
of carvedilol titration
Abul Kashem/2008 48 53.6 ± 2.6 25 America Telemedicine group versus usual care 1 year Office visits, emergency department visits, 4
hospitalization, telephone calls
S Scalvini/2005 230 59 ± 9 Italy Home-based telecardiology versus usual care 1 year Readmission due to heart failure; 4
cardiovascular events
Jeffrey A. Spaeder/2006 49 54.5 33 America Telemedicine system versus usual care 3 months Adverse events 5
William T Abraham/2011 560 61 27 America A wireless implantable hemodynamic 6 months Heart failure-related hospitalizations 5
monitoring system versus usual care
Sarwat I. Chaudhry/2010 1653 61 42 America Telemonitoring of interactive voice 6 months 1. Readmission for any reason hospitalization 6
response system versus usual care for heart failure, number of days in the hospital,
and number of hospitalizations
Friedrich Koehler/2011 710 66.9 ± 10.7 19 Germany Remote telemedical management 26 months 1. Death from any cause 5
versus usual care 2. A composite of cardiovascular death and
hospitalization for HF
Christine S. Ritchie/2016 346 63.2 ± 13 48.5 America A care transition nurse (CTN), interactive 1 month 1. 30-day rehospitalization 5
voice response versus usual care 2. (1) Rehospitalization and death, (2) days in
the hospital and out of the community
Josiane J.J. Boyne/2012 382 71 ± 11 41 Netherland Telemonitoring versus usual care 1 year 1. Mean time to first heart failure-related hospitalization 5
2. Heart failure admission and all-cause mortality
A. Giordano/2009 455 57 ± 10 15 Italy Home-based telemanagement 24 months 1. All-cause hospital readmissions 5
versus usual care 2. Mean cost for hospital readmission
P. Dendale/2012 160 76 ± 10 35 Belgium Telemonitoring versus usual care 6 months 1. All-cause mortality 5
2. Hospitalization costs
Andrew Weintraub/2010 188 69 34 America Telemonitoring versus usual care 3 months 1. HF hospitalization 5
2. Heart failure inpatient days
3. Quality of life
Goldberg/2003 280 59 ± 15 32 America Alere net system versus usual care 6 months 1. Hospitalization rates 4
2. Mortality
Marcia J. Wade/2011 316 78.1 47.7 America Telemonitoring versus usual care 6 months Emergency department visits, hospital 5
admissions, and death
Patrik Lynga/2012 344 73 ± 10.2 25 Sweden Telemonitoring versus usual care 12 months Hospitalization and death 6
Roberto Antonicelli/2008 57 70 38.5 Italy Telemonitoring versus usual care 16 months Mortality and rate of hospitalization, 5
quality of life, and costs
Gallagher BD/2017 40 64 (50–77) 25 America Telemonitoring versus usual care 1 month Readmission and adherence 5
Claudio Pedone/2015 90 80 ± 7 61.2 Italy 6 months All-cause death and hospital admissions 5
Heart Fail Rev (2020) 25:231–243
Heart Fail Rev (2020) 25:231–243 235

Jadad
score
hospitalization, all-cause mortality, and cardiac mortality. Other

5
5

6
4

5
5
commonly reported outcomes comprised the impact of tele-
medicine interventions on quality of life, length of hospital stay,
as well as hospitalization costs. Acceptability, patient satisfac-
tion, and emergency room visits were rarely reported in the
studies and therefore were excluded from our final analysis.

6 months All-cause mortality; heart failure mortality

Readmission rate and readmission cost
In most of the trials, interventions were typically delivered by

6 months Hospital readmissions and mortality;

12 months The Packer clinical composite score;
nurses. Using the revised 7-point Jadad scale, all the selected
HF hospitalization and mortality

Rate of death or hospitalization

hospitalization for any cause

RCTs had Jadad scores greater than 3, which suggested a good

3 months Rehospitalizations and deaths

16 months Mortality and quality of life

study design and high study quality. A more detailed descrip-

Rate of rehospitalization
tion of included trials is provided in Table 1.

quality-of-life score
Outcome parameters

Health-related outcomes and meta-analysis

All-cause hospitalization was reported in 23 studies and car-

diovascular diseases-related hospitalization was reported in 16
studies. With respect to clinical outcomes, 22 trials contribut-
Follow-

90 days
lengths

ed to the analysis of the all-cause death and 9 trials analyzed

3 years
1 year

1 year

the death due to heart failure. We did the meta-analyses for the
up

outcomes of all-cause of hospitalization, all-cause of mortali-

ty, cardiac hospitalization, and cardiac mortality (Fig. 2).
Evidence network for interventions included in the analysis
of the outcomes of telemedicine versus conventional
healthcare was shown in Fig. 3.
Usual care and telephone support

All-cause hospitalization and cardiac hospitalization

Telemonitoring and telephone

Telephone versus usual care

Telephone versus usual care

Telephone versus usual care

Telephone versus usual care

Telephone versus usual care
Telephone versus usual care
Telephone versus usual care

Telephone versus usual care

Population Type of interventions

With respect to all-cause hospitalizations, most studies report-

versus usual care.

versus usual care

ed beneficial effects of different telemedicine system interven-

tions. Telemonitoring was associated with a reduction in a
total number of all-cause hospitalization (OR 0.82, 95% CI
0.73–0.91, P = 0.0004). The effect was statistically significant
for 17 out of 19 studies. Heterogeneity was moderate (I2 =
70%; Fig. 2a), which is supposed to result from individual
Argentina
Argentina
Australia

America
America
America

America

discrepancies (age, gender distribution, etc.) and interventions

Britain

Brazil

(methods, duration, etc.). Two studies reported HF readmis-

sion within 30 days [14, 15]. Thirty-day rehospitalization rates
Age (year) Female (%)

were similar in the telemonitoring intervention group and the

UC control group. Cardiac hospitalizations were reported in
32.0
29.2

39.5
37

69

52
29

32
28

13 studies. The OR of telemonitoring versus conventional

healthcare on cardiac hospitalization was 0.83 (95% CI
Study population

70.7 ± 11.8

65 ± 13.3
73 ± 10

63 ± 13

72 ± 11

0.72–0.95, P = 0.007), suggesting a significant difference be-

75 ± 8

67.2
63.5

tween the two groups (Fig. 2b). Compared with conventional

65

healthcare, structured telephone support interventions reduced

1518
1518
405

165

462
181
200
287
111

the hospitalization for all causes (OR 0.86, 95% CI 0.78–0.96,

N

P = 0.006, I2 = 6%) and due to HF (OR 0.74, 95% CI 0.65–

Fernanda B. Domingues/2010

Robert Frank DeBusk/2004

0.85, P < 0.0001, I2 = 0%) (Fig. 2c, d).

Table 1 (continued)

Ann S. Laramee/2003

Daniel Ferrante/2010
Henry Krum /2013

All-cause of mortality and cardiac mortality

Lynda Blue/2001

Edward K./2002
GESICA /2005

Wendy A/1999
Author/year

A significant reduction in the risk of all-cause mortality (OR

0.75, 95% CI 0.62–0.90, P = 0.003) was identified in meta-
analyses of 15 studies of telemonitoring. However, the OR of
236 Heart Fail Rev (2020) 25:231–243

telemonitoring versus conventional healthcare on cardiac of these two analyses was moderate (I2 = 49%; Fig. 2e) and
mortality was not significant (OR 0.84, 95% CI 0.61–1.16, none (I2 = 0%; Fig. 2f), respectively. No significant effect of
P = 0.28) in a meta-analysis of five studies. The heterogeneity telephone support intervention on all-cause mortality (OR
 Heart Fail Rev (2020) 25:231–243 237

R Fig. 2 a Effect of telemonitoring versus usual care on all-cause hospital Publication bias
admission in patients with chronic heart failure. CI, confidence interval;
M-H, Mantel–Haenszel. b Effect of telemonitoring versus usual care on
Funnel plots and Egger’s testing were performed to assess the
cardiac hospital admission in patients with chronic heart failure. CI, con-
fidence interval; M-H, Mantel–Haenszel. c Effect of telephone support publication bias of all of the studies. There was no significant
interventions versus usual care on all-cause hospital admission in patients evidence of publication bias for all-cause mortality of tele-
with chronic heart failure. CI, confidence interval; M-H, Mantel– medicine interventions, which was revealed by the Egger
Haenszel. .d Effect of telephone support interventions versus usual care
on cardiac hospitalization in patients with chronic heart failure. CI, con-
and Begger tests (Egger: P = 0.888; Begger: P = 0.582). The
fidence interval; M-H, Mantel–Haenszel. e Effect of telemonitoring ver- funnel plot did not display asymmetry, while both Egger’s and
sus usual care on all-cause mortality in patients with chronic heart failure. Begg’s test indicate no publication bias (Fig. 4).
CI, confidence interval; M-H, Mantel–Haenszel. f Effect of
telemonitoring versus usual care on cardiac mortality in patients with
chronic heart failure. CI, confidence interval; M-H, Mantel–Haenszel. g
Quality of life
Effect of telephone versus usual care on all cause of mortality in patients
with chronic heart failure. CI, confidence interval; M-H, Mantel– Ten telemedicine studies reported slight improvements in
Haenszel. h Effect of telephone versus usual care on cardiac mortality measures of QoL for HF adults receiving telemedicine as
in patients with chronic heart failure. CI, confidence interval; M-H,
compared to those receiving conventional healthcare [18, 22,
Mantel–Haenszel. i Effect of interventions versus usual care on length
of hospital stay in patients with chronic heart failure. M-H = Mantel– 25, 26, 28–33]. Quality of life was measured by various ques-
Haenszel risk ratio. Data are from full peer-reviewed publications only tionnaires, including the Minnesota Living with Heart Failure
and reflect the most recent meta-analysis of telemedicine in heart failure Questionnaire (MLHFQ), score for SF-36 and SF-12, and the
Kansas City Cardiomyopathy Questionnaire (KCCQ), mak-
ing it difficult to compare the outcomes. However, patients in
the telemedicine group were more likely to report stable or
0.96, 95% CI 0.83–1.11, P = 0.55) was identified in meta- deteriorated symptoms, compared with those in the control
analyses of eight studies (Fig. 2g). A significant effect of tele- group. SF-36 was applied in two studies. One study found
phone support intervention on cardiac mortality (OR 0.54, [18] that patients randomly allocated to the remote telemedical
95% CI 0.34–0.86, P = 0.009) was identified in meta- management (RTM) group showed an improved score for SF-
analyses of studies (Fig. 2h). 36 physical functioning over the entire study period (P < 0.05)
compared with the conventional healthcare group. Ewa et al.
[22] observed the improvements in health-related QoL using
Length of hospital stay disease-specific measures of the KCCQ: the telemedicine
group had significantly higher score than control group over
HF-related length of stay was reported in 13 studies compar- 3 months follow-up (P < 0.05). Three studies contributed to
ing interventions with conventional healthcare [16–28], in- the analysis of the MLHFQ, while the SF-12 was also applied
cluding 3 telephone-supported studies and 10 telemonitoring in three studies. Goldberg reported that among patients com-
studies. There was a significant heterogeneity when data from pleting their 6-month follow-up visit, patients in both groups
the telemedicine interventions studies were pooled (df = 5, experienced improvement between baseline and 6 months in
P = 0.002, I2 = 74%, as shown in Fig. 2i). The analysis re- their Minnesota Living with Heart Failure, SF-12, and Health
vealed that there was a significant difference in HF-related Distress scores. Although no difference was statistically sig-
length of hospital stay between the overall interventions and nificant, the intervention group tended to improve all the qual-
control groups (pooled standardized difference in means = − ity of life measures [30]. Additionally, GESICA [31] found
2.21, 95% CI − 4.35 to − 0.06, Z = 2.02, P = 0.04), the that the intervention group had a significantly better quality of
telemonitoring and control groups (pooled standardized dif- life than the conventional healthcare group (mean total score
ference in means = − 1.71, 95% CI − 4.83 to − 1.42, Z = 1.07, on MLHFQ (30.6 vs. 35, P = 0.001).
P = 0.28), or the telephone-supported care and control groups
(pooled standardized difference in means = − 3.41, 95% CI − Hospitalization costs
5.01 to − 1.82, Z = 4.2, P = 0.0001). William et al. [16] report-
ed that the length of hospital stay for HF-related hospitaliza- Five studies [16, 19, 26, 27, 34] examined the effects of tele-
tions was significantly shorter in the treatment group than in medicine interventions on hospitalization costs. Although
the control group (2.2 days [SD 6.8] vs. 3.8 days [SD 11.1], hospitalization costs were reported in the five original studies,
P = 0.02) during 6 months follow-up. However, no significant different applied methodologies did not allow the pooled re-
difference in HF-related length of hospital stay was observed sults into the meta-analysis. The general outcomes were sta-
comparing the telemonitoring group with the control group tistically inconclusive and varied depending on the context
(pooled standardized difference in means = − 1.71, 95% CI and specific telemedicine systems of the studies.
− 4.83 to − 1.42, Z = − 1.07, P = 0.28). Furthermore, many studies showed that there was no
238 Heart Fail Rev (2020) 25:231–243

Fig. 2 continued.

significant difference in hospitalization costs between inter- Discussion

vention groups and non-intervention groups. Dendale et al.
[19] reported that even though the total hospitalization cost In this meta-analysis, we performed an up-to-date assessment
for HF was almost doubled in the control group (1458 + of the effectiveness of telemedicine systems for the manage-
3420 Euro/patient) compared with the telemedicine group ment of HF patients. By summarizing the current best evi-
(902 + 2277 Euro/patient), the difference was not significant dence, this network meta-analysis showed that compared to
(P = 0.23). Laramee et al. reported that the intervention did not conventional healthcare, telemedicine intervention appears to
increase costs and no significant differences were found in be beneficial for patients with HF, particularly in reducing all-
both outpatient and inpatient resource utilization between the cause hospitalization, cardiac hospitalization, all-cause mor-
groups [27]. tality, cardiac mortality, and length of stay in HF patients. This
Heart Fail Rev (2020) 25:231–243 239

Fig. 2 continued.

Fig. 3 Evidence network for

interventions included in the
analysis of the outcomes of
telemedicine versus usual care.
Each node represents different
outcomes and the size of each
node indicates the total number of
studies included in the network
240 Heart Fail Rev (2020) 25:231–243

Fig. 4 Funnel plot comparing

interventions versus controls
reporting all-cause mortality.
Funnel plot assessing publication
bias

work represents the comprehensive application of network with or without telemedicine, the only possible advantage of
meta-analysis to examine the comparative effectiveness of telemedicine is shortening the time to a treatable, or
telemedicine interventions in improving HF patient outcomes. Bactionable,^ event. In this respect, we should consider that
Available studies examining the patient experience and ef- none of the telemedicine systems should be expected to pre-
fect of patient education on acceptance and adherence to the dict some events, e.g., acute cardiac death or no cardiac death.
intervention of telemonitoring were limited [34, 35]. The As was reported by Sarwat et al. [17], non-implantable
adoption of telemedicine is driven by the expectation that it telemonitoring systems for HF do not seem to improve out-
should improve patient outcomes and reduce healthcare costs. comes compared with conventional healthcare. Being the larg-
Few studies, however, have compared telemedicine with con- est RCT done so far, a comprehensive non-invasive
ventional health care with respect to healthcare utilization telemonitoring system did not reduce morbidity or mortality
[36]. In a recent network meta-analysis of RCTs by Kotb in 1653 patients who were randomly assigned to the
et al. [37], teletransmission was found to reduce the odds of telemonitoring system versus conventional healthcare. It is
mortality as well as the HF-related hospitalizations compared still unclear which factor could explain the significant reduc-
with the conventional healthcare. In the present meta-analysis tion in mortality achieved by telemedicine [6]. The inconsis-
of 29 studies, we have specifically considered planned or un- tent outcomes of the telemedicine program may be due to the
planned hospital visits that provide additional data to the field lack of consensus protocol or guideline for conducting tele-
with an updated literature search up to 2018. Most of the medicine care. The purpose of a remote interview may range
evidence that is currently available on the impact of telemed- from improving diet and treatment compliance to regular
icine interventions involves the comparison of an active form monitoring of the HF-related symptoms and self-manage-
of telemedicine to conventional healthcare. Findings from this ment. However, many home monitoring systems are designed
network meta-analysis are unique in that various comparisons for transmission of body weight, blood pressure, and heart rate
were examined across different forms of telemedicine inter- via a standard telephone line or network system to a central
ventions. Most available literature focused on the primary server. It may be helpful to monitor the real-time clinical con-
outcomes of mortality and hospitalizations. We have found dition of the patients for early treatment.
that telemedicine is associated with a significant reduction in The previous network meta-analysis included 21 studies
the total number of hospital visits and mortality. In other which included a control group and examined the impact of
words, telemedicine safely reduces healthcare utilization by telemonitoring. Finally, only nine of them were followed par-
reducing elective face-to-face hospital visits. ticipants for more than 6 months [19]. In this meta-analysis,
HF is a complex illness, and optimal management requires most of the studies had longer than 6 months of follow-up.
regular patient monitoring. However, the financial and orga- This may suggest that the potential benefits of telemedicine
nizational strain on healthcare systems prevents timely moni- require longer periods of follow-ups before they are observed.
toring frequently. This could lead to reliance on patient help- Telemedicine is part of a comprehensive package of care that
seeking which often occurs when it is too late to prevent hos- includes education and empowerment of the patients, early
pitalization [38]. Essentially, telemedicine is a diagnostic mo- warning of deterioration, access to health professionals’ ad-
dality which, without an appropriate therapeutic intervention, vice and moral support, and pharmacological intervention. For
could not be expected to alter clinical outcomes. If one as- instance, one of the studies reported [39] that remote monitor-
sumes that the appropriateness of such interventions is similar ing with an automated telemedicine system can successfully
Heart Fail Rev (2020) 25:231–243 241

facilitate titration of carvedilol in outpatients with class II and studies, publication and reporting biases were inevitable to
III HF defined by NYHA. Telemedicine titrations were as some extent. It is difficult to testify differences between inter-
successful as titrations in the clinic. Further, the time to reach vention duration and intervention designs. These differences
the final dose of carvedilol was significantly shorter in the in study designs resulted in low to large scores of heterogene-
intervention group compared to that of conventional ity (0% to 75%). Furthermore, some trials were underpowered
healthcare group (33.6 vs. 63.7 days, P < 0.001). to detect the primary outcome and did not report outcome
Telemedicine interventions, therefore, reflect complex assessor blinding. Telemedicine usually builds on self-moni-
healthcare strategies and are not limited to simple data- toring, with evidence that it can help educate patients about
gathering. which symptoms and signs are most important and what mea-
Positive results were found on health-related QoL in pa- sures can be taken to destabilize the syndrome. The content of
tients with chronic HF, although data were limited. Indeed, telemedicine interventions was often poorly described, mak-
QoL is an important measure of health, particularly for older ing it difficult to understand exactly what was provided.
people and those suffering from HF. Maintaining moderate or Moreover, due to the differences in the selection criteria of
a high and improving level of physical activities is associated the included studies (e.g., LVEF and New York Heart
with a better health-related QoL in patients with chronic HF Association), the generalizability of the treatment effect is
[40]. Regarding the quality of the studies included in this unclear.
meta-analysis, different questionnaires were used (SF-36,
SF-12, MLHF, KCCQ), making it difficult to compare out-
comes. However, the majority demonstrated improvements in
the patients who underwent telemedicine intervention, espe- Conclusions
cially in terms of the MLHFQ and SF-36 physical aspects, and
these improvements led to both better life quality and favor- Telemedicine interventions appear to lead to benefits for pa-
able prognoses. Further information is required to assess the tients with CHF, decrease all-cause hospital admissions, and
effect of the telemonitoring use on the patient’s QoL, and improve QoL, although there are still several important issues
perceptions of health status, as this was cited as a barrier to consider. Only limited studies are available on the cost-
against uptake and adherence. This network analysis was lim- benefits and appropriate business models for the interventions;
ited to RCTs only. This was deemed appropriate, however, impacts of these interventions on patient QoL have only been
given the availability of a substantial amount of evidence reported in a few studies, and the optimal duration of these
and the reduced likelihood of bias and confounding associated interventions is still not clear. There is also a significant lack
with this study design. For the most part, the risk of bias of researchers concentrating on HF patients in remote rural
associated with included studies was found to be either low areas who might benefit from a telemedicine service.
or moderate and further sensitivity analyses did not signifi- Further research is therefore required to fill in this gap in
cantly differ from the main analysis of the study. knowledge. We suggest that other aspects should also be
With constrained resources for healthcare expenditures, it addressed, as telemedicine is not the only one component
is reasonable to expect an evaluation of new technologies not of managing HF and could not replace face-to-face consul-
just on safety and efficacy but also on cost-effectiveness. With tations between healthcare providers and patients. Such
the substantial societal and economic impacts of HF, telemed- studies need to investigate, from the patients’ perspective,
icine interventions that can help minimize the likelihood of the effect of educational methods and technological sup-
cost of care associated with HF may provide significant ben- ports, benefits of tailored monitoring, and cost-
efits to the healthcare system. Researchers that used an ana- effectiveness analysis. The efficacy of telemedicine using
lytical approach for assessing healthcare and patient-related an advanced telemonitoring device and newly developed
costs showed that most of the telemedicine was cost saving guidelines in the remote follow-up and management of HF
as the telemedicine groups have shorter Bstay of length.^ patients should also be investigated.
There is no consensus on hospitalization costs for HF patients.
Some differences across models may be expected. However, Acknowledgements We thank Northern Jiangsu People’s Hospital for
helpful statistical advice.
many reports showed that there was no significant difference
in hospitalization cost use of the intervention and conventional
Funding This work was supported by funds from the Science and
healthcare groups [27]. Technology Department of Jiangsu Province, China (No. BL2013022).
Admittedly, our meta-analysis has its limitations.
Telemedicine interventions were heterogeneous in terms of Compliance with ethical standards
monitored parameters and HF selection criteria. Our study
depended only on the data reported in studies, some endpoint Conflict of interest The authors declare that they have no conflict of
data were unavailable, and considering the limited number of interest.
242 Heart Fail Rev (2020) 25:231–243

Open Access This article is distributed under the terms of the Creative 13. Cao J, Li Z, Yang L, Liu C, Luan X (2016) Association between
Commons Attribution 4.0 International License (http:// tissue inhibitor of metalloproteinase-3 gene methylation and gastric
creativecommons.org/licenses/by/4.0/), which permits unrestricted use, cancer risk: a meta-analysis. Genet Test Mol Biomarkers 20(8):
distribution, and reproduction in any medium, provided you give appro- 427–431. https://doi.org/10.1089/gtmb.2015.0332
priate credit to the original author(s) and the source, provide a link to the 14. Ritchie CS, Houston TK, Richman JS, Sobko HJ, Berner ES,
Creative Commons license, and indicate if changes were made. Taylor BB, Salanitro AH, Locher JL (2016) The E-Coach technol-
ogy-assisted care transition system: a pragmatic randomized trial.
Transl Behav Med 6(3):428–437. https://doi.org/10.1007/s13142-
016-0422-8
15. Gallagher BD, Moise N, Haerizadeh M, Ye S, Medina V, Kronish
References IM (2017) Telemonitoring Adherence to Medications in Heart
Failure Patients (TEAM-HF): a pilot randomized clinical trial. J
1. Rathi S, Deedwania PC (2012) The epidemiology and pathophys- Card Fail 23(4):345–349. https://doi.org/10.1016/j.cardfail.2016.
iology of heart failure. Med Clin North Am 96(5):881–890. https:// 11.001
doi.org/10.1016/j.mcna.2012.07.011 16. Abraham WT, Adamson PB, Bourge RC, Aaron MF, Costanzo
2. Desai AS (2012) Home monitoring heart failure care does not im- MR, Stevenson LW, Strickland W, Neelagaru S, Raval N,
prove patient outcomes: looking beyond telephone-based disease Krueger S, Weiner S, Shavelle D, Jeffries B, Yadav JS (2011)
management. Circulation 125(6):828–836. https://doi.org/10.1161/ Wireless pulmonary artery haemodynamic monitoring in chronic
circulationaha.111.031179 heart failure: a randomised controlled trial. Lancet (London,
3. Savard LA, Thompson DR, Clark AM (2011) A meta-review of England) 377(9766):658–666. https://doi.org/10.1016/s0140-
evidence on heart failure disease management programs: the chal- 6736(11)60101-3
lenges of describing and synthesizing evidence on complex inter- 17. Chaudhry SI, Mattera JA, Curtis JP, Spertus JA, Herrin J, Lin Z,
ventions. Trials 12(194). https://doi.org/10.1186/1745-6215-12- Phillips CO, Hodshon BV, Cooper LS, Krumholz HM (2010)
194 Telemonitoring in patients with heart failure. N Engl J Med
4. Pandor A, Gomersall T, Stevens JW, Wang J, Al-Mohammad A, 363(24):2301–2309. https://doi.org/10.1056/NEJMoa1010029
Bakhai A, Cleland JG, Cowie MR, Wong R (2013) Remote mon- 18. Koehler F, Winkler S, Schieber M, Sechtem U, Stangl K, Bohm M,
itoring after recent hospital discharge in patients with heart failure: a Boll H, Baumann G, Honold M, Koehler K, Gelbrich G, Kirwan
systematic review and network meta-analysis. Heart 99(23):1717– BA, Anker SD (2011) Impact of remote telemedical management
1726. https://doi.org/10.1136/heartjnl-2013-303811 on mortality and hospitalizations in ambulatory patients with chron-
ic heart failure: the telemedical interventional monitoring in heart
5. Koehler F, Winkler S, Schieber M, Sechtem U, Stangl K, Bohm M,
failure study. Circulation 123(17):1873–1880. https://doi.org/10.
de Brouwer S, Perrin E, Baumann G, Gelbrich G, Boll H, Honold
1161/circulationaha.111.018473
M, Koehler K, Kirwan BA, Anker SD (2012) Telemedicine in heart
19. Dendale P, De Keulenaer G, Troisfontaines P, Weytjens C, Mullens
failure: pre-specified and exploratory subgroup analyses from the
W, Elegeert I, Ector B, Houbrechts M, Willekens K, Hansen D
TIM-HF trial. Int J Cardiol 161(3):143–150. https://doi.org/10.
(2012) Effect of a telemonitoring-facilitated collaboration between
1016/j.ijcard.2011.09.007
general practitioner and heart failure clinic on mortality and rehos-
6. Clarke M, Shah A, Sharma U (2011) Systematic review of studies
pitalization rates in severe heart failure: the TEMA-HF 1
on telemonitoring of patients with congestive heart failure: a meta-
(TElemonitoring in the MAnagement of Heart Failure) study. Eur
analysis. J Telemed Telecare 17(1):7–14. https://doi.org/10.1258/jtt.
J Heart Fail 14(3):333–340. https://doi.org/10.1093/eurjhf/hfr144
2010.100113
20. Kashem A, Droogan MT, Santamore WP, Wald JW, Bove AA
7. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, (2008) Managing heart failure care using an internet-based telemed-
Ioannidis JP, Clarke M, Devereaux PJ, Kleijnen J, Moher D icine system. J Card Fail 14(2):121–126. https://doi.org/10.1016/j.
(2009) The PRISMA statement for reporting systematic reviews cardfail.2007.10.014
and meta-analyses of studies that evaluate healthcare interventions: 21. Kashem A, Cross RC, Santamore WP, Bove AA (2006)
explanation and elaboration. BMJ (Clinical research ed) 339: Management of heart failure patients using telemedicine commu-
b2700. https://doi.org/10.1136/bmj.b2700 nication systems. Curr Cardiol Rep 8(3):171–179
8. Clark HD, Wells GA, Huet C, McAlister FA, Salmi LR, Fergusson 22. Hagglund E, Lynga P, Frie F, Ullman B, Persson H, Melin M,
D, Laupacis A (1999) Assessing the quality of randomized trials: Hagerman I (2015) Patient-centred home-based management of
reliability of the Jadad scale. Control Clin Trials 20(5):448–452 heart failure. Findings from a randomised clinical trial evaluating
9. Buzzetti E, Kalafateli M, Thorburn D, Davidson BR, Tsochatzis E, a tablet computer for self-care, quality of life and effects on knowl-
Gurusamy KS (2017) Interventions for hereditary edge. Scand Cardiovasc J 49(4):193–199. https://doi.org/10.3109/
haemochromatosis: an attempted network meta-analysis. 14017431.2015.1035319
Cochrane Database Syst Rev 3:Cd011647. https://doi.org/10. 23. Boyne JJ, Vrijhoef HJ, Crijns HJ, De Weerd G, Kragten J, Gorgels
1002/14651858.CD011647.pub2 AP (2012) Tailored telemonitoring in patients with heart failure:
10. Stonehouse W, Wycherley T, Luscombe-Marsh N, Taylor P, results of a multicentre randomized controlled trial. Eur J Heart
Brinkworth G, Riley M (2016) Dairy intake enhances body weight Fail 14(7):791–801. https://doi.org/10.1093/eurjhf/hfs058
and composition changes during energy restriction in 18-50-year- 24. Blue L, Lang E, McMurray JJ, Davie AP, McDonagh TA, Murdoch
old adults—a meta-analysis of randomized controlled trials. DR, Petrie MC, Connolly E, Norrie J, Round CE, Ford I, Morrison
Nutrients 8(7). https://doi.org/10.3390/nu8070394 CE (2001) Randomised controlled trial of specialist nurse interven-
11. Higgins JP, Thompson SG, Deeks JJ, Altman DG (2003) tion in heart failure. BMJ (Clinical research ed) 323(7315):715–718
Measuring inconsistency in meta-analyses. BMJ (Clinical research 25. Wade MJ, Desai AS, Spettell CM, Snyder AD, McGowan-
ed) 327(7414):557–560. https://doi.org/10.1136/bmj.327.7414.557 Stackewicz V, Kummer PJ, Maccoy MC, Krakauer RS (2011)
12. Zhang Y, Ma XJ, Shi DZ (2015) Effect of trimetazidine in patients Telemonitoring with case management for seniors with heart fail-
undergoing percutaneous coronary intervention: a meta-analysis. ure. Am J Manag Care 17(3):e71–e79
PLoS One 10(9):e0137775. https://doi.org/10.1371/journal.pone. 26. Kasper EK, Gerstenblith G, Hefter G, Van Anden E, Brinker JA,
0137775 Thiemann DR, Terrin M, Forman S, Gottlieb SH (2002) A
Heart Fail Rev (2020) 25:231–243 243

randomized trial of the efficacy of multidisciplinary care in heart 35. Anker SD, Koehler F, Abraham WT (2011) Telemedicine and re-
failure outpatients at high risk of hospital readmission. J Am Coll mote management of patients with heart failure. Lancet (London,
Cardiol 39(3):471–480 England) 378(9792):731–739. https://doi.org/10.1016/s0140-
27. Laramee AS, Levinsky SK, Sargent J, Ross R, Callas P (2003) Case 6736(11)61229-4
management in a heterogeneous congestive heart failure popula- 36. Omboni S, Gazzola T, Carabelli G, Parati G (2013) Clinical useful-
tion: a randomized controlled trial. Arch Intern Med 163(7):809– ness and cost effectiveness of home blood pressure telemonitoring:
817. https://doi.org/10.1001/archinte.163.7.809 meta-analysis of randomized controlled studies. J Hypertens 31(3):
28. Lynga P, Persson H, Hagg-Martinell A, Hagglund E, Hagerman I, 455–467; discussion 467-458. https://doi.org/10.1097/HJH.
Langius-Eklof A, Rosenqvist M (2012) Weight monitoring in pa- 0b013e32835ca8dd
tients with severe heart failure (WISH). A randomized controlled 37. Kotb A, Cameron C, Hsieh S, Wells G (2015) Comparative effec-
trial. Eur J Heart Fail 14(4):438–444. https://doi.org/10.1093/ tiveness of different forms of telemedicine for individuals with heart
eurjhf/hfs023 failure (HF): a systematic review and network meta-analysis. PLoS
29. Soreca S, Aprile S, Cardone A, Carella G, Fimiani B, Guarnaccia F, One 10(2):e0118681. https://doi.org/10.1371/journal.pone.
Santoro G, Apuzzi V, Bosso G, Valvano A, Zito G, Oliviero U 0118681
(2012) Management of chronic heart failure: role of home echocar- 38. O'Connor M, Asdornwised U, Dempsey ML, Huffenberger A, Jost
diography in monitoring care programs. World J Cardiol 4(3):72– S, Flynn D, Norris A (2016) Using telehealth to reduce all-cause 30-
76. https://doi.org/10.4330/wjc.v4.i3.72 day hospital readmissions among heart failure patients receiving
30. Goldberg LR, Piette JD, Walsh MN, Frank TA, Jaski BE, Smith skilled home health services. Appl Clin Inform 7(2):238–247.
AL, Rodriguez R, Mancini DM, Hopton LA, Orav EJ, Loh E https://doi.org/10.4338/aci-2015-11-soa-0157
(2003) Randomized trial of a daily electronic home monitoring 39. Spaeder J, Najjar SS, Gerstenblith G, Hefter G, Kern L, Palmer JG,
system in patients with advanced heart failure: the Weight Gottlieb SH, Kasper EK (2006) Rapid titration of carvedilol in
Monitoring in Heart Failure (WHARF) trial. Am Heart J 146(4): patients with congestive heart failure: a randomized trial of auto-
705–712. https://doi.org/10.1016/s0002-8703(03)00393-4 mated telemedicine versus frequent outpatient clinic visits. Am
31. (2005) Randomised trial of telephone intervention in chronic heart Heart J 151(4):844.e841–844.e810. https://doi.org/10.1016/j.ahj.
failure: DIAL trial. BMJ (Clinical research ed) 331(7514):425. 2005.06.044
https://doi.org/10.1136/bmj.38516.398067.E0
40. van Tol BA, Huijsmans RJ, Kroon DW, Schothorst M, Kwakkel G
32. DeBusk RF, Miller NH, Parker KM, Bandura A, Kraemer HC, Cher
(2006) Effects of exercise training on cardiac performance, exercise
DJ, West JA, Fowler MB, Greenwald G (2004) Care management
capacity and quality of life in patients with heart failure: a meta-
for low-risk patients with heart failure: a randomized, controlled
analysis. Eur J Heart Fail 8(8):841–850. https://doi.org/10.1016/j.
trial. Ann Intern Med 141(8):606–613
ejheart.2006.02.013
33. Ferrante D, Varini S, Macchia A, Soifer S, Badra R, Nul D,
Grancelli H, Doval H (2010) Long-term results after a telephone
intervention in chronic heart failure: DIAL (Randomized Trial of
Phone Intervention in Chronic Heart Failure) follow-up. J Am Coll
Cardiol 56(5):372–378. https://doi.org/10.1016/j.jacc.2010.03.049
34. Antonicelli R, Testarmata P, Spazzafumo L, Gagliardi C, Bilo G, Publisher’s note Springer Nature remains neutral with regard to
Valentini M, Olivieri F, Parati G (2008) Impact of telemonitoring at jurisdictional claims in published maps and institutional affiliations.
home on the management of elderly patients with congestive heart
failure. J Telemed Telecare 14(6):300–305. https://doi.org/10.1258/
jtt.2008.071213