Global mortality variations in patients with heart failure: Results from the International Congestive Heart Failure (INTER-CHF) prospective cohort study

Abstract

Background: Most data on mortality and prognostic factors in patients with heart failure come from North America and Europe, with little information from other regions. Here, in the International Congestive Heart Failure (INTER-CHF) study, we aimed to measure mortality at 1 year in patients with heart failure in Africa, China, India, the Middle East, southeast Asia and South America; we also explored demographic, clinical, and socioeconomic variables associated with mortality.
Methods: We enrolled consecutive patients with heart failure (3695 [66%] clinic outpatients, 2105 [34%] hospital in patients) from 108 centres in six geographical regions. We recorded baseline demographic and clinical characteristics and followed up patients at 6 months and 1 year from enrolment to record symptoms, medications, and outcomes. Time to death was studied with Cox proportional hazards models adjusted for demographic and clinical variables, medications, socioeconomic variables, and region. We used the explained risk statistic to calculate the relative contribution of each level of adjustment to the risk of death.
Findings: We enrolled 5823 patients within 1 year (with 98% follow-up). Overall mortality was 16·5%: highest in Africa (34%) and India (23%), intermediate in southeast Asia (15%), and lowest in China (7%), South America (9%), and the Middle East (9%). Regional differences persisted after multivariable adjustment. Independent predictors of mortality included cardiac variables (New York Heart Association Functional Class III or IV, previous admission for heart failure, and valve disease) and non-cardiac variables (body-mass index, chronic kidney disease, and chronic obstructive pulmonary disease). 46% of mortality risk was explained by multivariable modelling with these variables; however, the remainder was unexplained.
Interpretation: Marked regional differences in mortality in patients with heart failure persisted after multivariable adjustment for cardiac and non-cardiac factors. Therefore, variations in mortality between regions could be the result of health-care infrastructure, quality and access, or environmental and genetic factors. Further studies in large, global cohorts are needed.
Funding: The study was supported by Novartis.

Full text

www.thelancet.com/lancetgh Vol 5 July 2017 e
665
Articles
Global mortality variations in patients with heart failure:
results from the International Congestive Heart Failure
(INTER-CHF) prospective cohort study
Hisham Dokainish, Koon Teo, Jun Zhu, Ambuj Roy, Khalid F AlHabib, Ahmed ElSayed, Lia Palileo-Villaneuva, Patricio Lopez-Jaramillo,
Kamilu Karaye, Khalid Yusoff, Andres Orlandini, Karen Sliwa, Charles Mondo, Fernando Lanas, Dorairaj Prabhakaran, Amr Badr,
Mohamed Elmaghawry, Albertino Damasceno, Kemi Tibazarwa, Emilie Belley-Cote, Kumar Balasubramanian, Shofiqul Islam, Magdi H Yacoub,
Mark D Huffman, Karen Harkness, Alex Grinvalds, Robert McKelvie, Shrikant I Bangdiwala, Salim Yusuf, on behalf of the INTER-CHF
Investigators*
Summary
Background Most data on mortality and prognostic factors in patients with heart failure come from North America
and Europe, with little information from other regions. Here, in the International Congestive Heart Failure (INTER-
CHF) study, we aimed to measure mortality at 1 year in patients with heart failure in Africa, China, India, the Middle
East, southeast Asia and South America; we also explored demographic, clinical, and socioeconomic variables
associated with mortality.
Methods We enrolled consecutive patients with heart failure (3695 [66%] clinic outpatients, 2105 [34%] hospital in
patients) from 108 centres in six geographical regions. We recorded baseline demographic and clinical characteristics
and followed up patients at 6 months and 1 year from enrolment to record symptoms, medications, and outcomes.
Time to death was studied with Cox proportional hazards models adjusted for demographic and clinical variables,
medications, socioeconomic variables, and region. We used the explained risk statistic to calculate the relative
contribution of each level of adjustment to the risk of death.
Findings We enrolled 5823 patients within 1 year (with 98% follow-up). Overall mortality was 16·5%: highest in Africa
(34%) and India (23%), intermediate in southeast Asia (15%), and lowest in China (7%), South America (9%), and the
Middle East (9%). Regional dierences persisted after multivariable adjustment. Independent predictors of mortality
included cardiac variables (New York Heart Association Functional Class III or IV, previous admission for heart
failure, and valve disease) and non-cardiac variables (body-mass index, chronic kidney disease, and chronic obstructive
pulmonary disease). 46% of mortality risk was explained by multivariable modelling with these variables; however,
the remainder was unexplained.
Interpretation Marked regional dierences in mortality in patients with heart failure persisted after multivariable
adjustment for cardiac and non-cardiac factors. Therefore, variations in mortality between regions could be the result
of health-care infrastructure, quality and access, or environmental and genetic factors. Further studies in large, global
cohorts are needed.
Funding The study was supported by Novartis.
Copyright © The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY-NC-ND 4.0
license.
Introduction
Heart failure is an important global health problem,
aecting about 26 million people worldwide;1 it is
associated with high mortality and is estimated to have
cost about US$100 billion in 2012.2 Most data on
outcomes in patients with heart failure come from North
America and Europe, with much less information from
Africa, Asia, the Middle East, and South America.3–5 The
few data from low-income and middle-income countries
(LMIC) suggest that mortality in patients with heart
failure in these countries is greater than that in high-
income countries.6–8 However, reasons for variation in
out comes between regions remain unclear. Therefore,
we designed a prospective registry of patients with heart
failure in 16countries in Africa, Asia, the Middle East,
and South America to document 1-year mortality in
patients from these regions, and to explore variables
associated with mortality.
Methods
Study design and participants
The International Congestive Heart Failure Study
(INTER-CHF) is a prospective cohort study, conducted in
108 centres in 16 countries with follow-up at 12months.
The rationale and design for this study have been
published elsewhere,9 but are briefly described here.
Lancet Glob Health 2017;
5: e665–72
Published Online
April 30, 2017
http://dx.doi.org/10.1016/
S2214-109X(17)30196-1
This online publication has
been corrected. The corrected
version first appeared at
thelancet.com on May 9, 2017
See Comment page e634
*Members listed in appendix
Population Health Research
Institute, McMaster University,
Hamilton, Canada
(H Dokainish MD, K Teo PhD,
E Belley-Cote MD,
K Balasubramanian MSc,
S Islam MSc, K Harkness PhD,
A Grinvalds BSc, R McKelvie PhD,
S I Bangdiwala PhD,
S Yusuf DPhil); Cardiovascular
Institute and Fuwai Hospital,
Chinese Academy of Medical
Sciences and Peking Union
Medical College, Beijing, China
(J Zhu MD); All India Institute of
Medical Science, New Delhi,
India (A Roy MD); King Fahad
Cardiac Center, King Saud
University, Riyadh, Saudi
Arabia (K F AlHabib MBBS);
AlShaab Teaching Hospital,
Khartoum, Sudan
(A ElSayed MD); University of
the Philippines, Manila,
Philippines
(L Palileo-Villaneuva, MD);
Fundación Oftalmológica de
Santander, (FOSCAL) and
Medical School, Universidad de
Santander (UDES),
Bucaramanga, Colombia
(P Lopez-Jaramillo PhD); Aminu
Kano Teaching Hospital and
Bayero University, Kano,
Nigeria (K Karaye MD); UCSI
University and UiTM Selayang,
Selangor University, Cheras,
Malaysia (K Yusoff MD); ECLA
Foundation, Instituto
Cardiovascular de Rosario,
Argentina (A Orlandini MD);
Downloaded for Anonymous User (n/a) at Saint Louis University from ClinicalKey.com by Elsevier on September 30,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.

Articles
666
www.thelancet.com/lancetgh Published online April 30, 2017 http://dx.doi.org/10.1016/
S2214-109X(17)30196-1
We enrolled patients with a clinical diagnosis of heart
failure in Africa (Mozambique, Nigeria, South Africa,
Sudan, and Uganda), China, India, the Middle East
(Egypt, Qatar, and Saudi Arabia), southeast Asia
(Malaysia, and the Philippines), and South America
(Argentina, Chile, Colombia, and Ecuador; appendix).
We included consecutive heart failure patients from
outpatient clinics and inpatient hospital wards at
participating centres. We aimed to recruit two-thirds of
the study population from out patient clinics since most
previous data on heart failure have come from hospital
inpatients.4–7,10 Patients were aged 18 years or older and
enrolled from academic health-care centres, community
health centres, and specialist and primary care clinics.
Where feasible, at least one centre in each country was in
a rural area. We excluded patients with severe non-
cardiac diseases that could aect survival within 1 year,
and patients who were dicult to follow-up (for example,
because of their migratory status).
The study was approved by institutional review boards
or independent ethics committees at participating sites
and the international coordinating centre (McMaster
University, Hamilton, ON, Canada). Participants or their
substitute decision maker provided written informed
consent.
Procedures
At enrolment, we recorded information on demographics,
clinical factors, medications, and socioeconomic factors.
Echocardiograms, if done for clinical care, were used for
information about left ventricular systolic function and
valve disease. Left ventricular ejection fraction (LVEF)
was defined as reduced when less than 40%.11 Valvular
heart disease was defined as the presence of at least
moderate stenosis or regurgitation in at least one cardiac
valve.
A local physician determined the cause of heart failure
using all available clinical data. Although we did not
specify that patients had to meet specific criteria to be
included, we prospectively collected information to assess
the proportion of participants who met the Boston criteria
for heart failure.12,13 Patients had follow-up visits at
6 months and 12 months, at which symptoms, medi-
cations and outcomes were recorded.
Outcomes
Primary outcome was time to all-cause mortality
within 1 year. Cause of death was also recorded, and
categorised by local investigators as cardiac, non-cardiac,
or unknown.
Statistical analysis
We used univariable and multivariable Cox proportional
hazards models for time-to-event analysis of mortality.
We verified the proportionality assumptions with
standard log[-log(survival)] plots. In model 1, we included
17 variables from the Meta-Analysis Global Group in
Chronic Heart Failure (MAGGIC) criteria,14 plus socio-
economic variables and region, in a multivariable model.
These variables (see appendix) were divided into
demographic variables, clinical variables, medications,
socio economic variables, and region to determine the
eect of each level of adjustment on the conditional
instan taneous risk (hazard) of death within 1 year. To
calculate regional hazard of death, we used South
America as the reference region because it had a low
death rate.
In model 2, we performed a second analysis using the
same variables as in model 1, but excluded data derived
from echocardiograms (LVEF, valve disease).
We conducted a sensitivity analysis including patients
who had clinical variables that fulfilled the Boston heart
Research in Context
Evidence before this study
We searched MEDLINE on Feb 10, 2017, for English-language
cohort studies published since 1990 of patients with heart
failure recruited from outpatient clinics or hospital settings.
Studies had to include data on mortality. We used the search
terms “heart failure”, “global”, “international”, “worldwide”,
“outcomes”, and “mortality”. There were no cohort studies
identified that included patients with heart failure from
different world regions that compared outcomes.
Added value of this study
The current study adds to the findings of previously published
local registries and randomised controlled trials by directly
demonstrating that mortality in patients with heart failure
was highest in Africa and India, intermediate in southeast
Asia, and lowest in China, South America, and the Middle East.
These differences persisted after adjustment for clinical
variables, medication use, and socioeconomic variables,
suggesting that a large part of the variations in mortality
could be the result of unrecorded or unknown factors.
Implications of all of the available evidence
The evidence suggests that there are important differences
between regions in death rates for patients with heart failure.
In general, patients with heart failure from richer regions have
lower mortality than do patients from poorer regions. These
regional differences persist after adjustment for risk factors
and treatments, which only explained about half of the risk of
death. Therefore, the data suggest that regional variations in
heart failure outcomes might be due to factors that are not
well described or measured in the current literature. These
include health-care quality, access, and infrastructure as well
as environmental factors and genetics. Further research into
these factors is, therefore, warranted.
Hatter Institute for
Cardiovascular Research in
Africa, SAMRC, Faculty of
Health Sciences, University of
Cape Town, South Africa
(K Sliwa PhD); Mulago National
Referral Hospital, Kampala,
Uganda (C Mondo MD);
Universidad de La Frontera,
Temuco, Chile (F Lanas, MD);
Centre for Chronic Disease
Control and Public Health
Foundation of India,New Delhi,
India (D Prabhakaran MD);
Hamad Medical Center, Doha,
Qatar (A Badr MD,
M H Yacoub, MD); Aswan Heart
Centre, Aswan, Egypt
(M H Yacoub,
M Elmaghawry PhD); Eduardo
Mondlane University, Maputo,
Mozambique
(A Damasceno MD);
Department of Cardiovascular
Medicine, Muhimbili National
Hospital, Dar es Salaam,
Tanzania (K Tibazarwa MD);
Northwestern University
Feinberg School of Medicine,
Chicago, USA
(M D Huffman MD)
Correspondence to:
Dr Hisham Dokainish,
McMaster University,
237 Barton Street East, CVSRI
#C3 111, Hamilton, ON,
Canada L8L 2X2
hisham.dokainish@phri.ca
See Online for appendix
Downloaded for Anonymous User (n/a) at Saint Louis University from ClinicalKey.com by Elsevier on September 30,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.

Articles
www.thelancet.com/lancetgh Vol 5 July 2017 e
667
failure criteria.11 We did further analyses by admission
status at enrolment (ie, hospital inpatients or clinic
outpatients). In Cox proportional hazards models, we
calculated the explained risk statistic to measure the
Overall
(N=5823)
Africa
(N=1294)
India
(N=858)
SoutheastAsia
(N=811)
Middle East
(N=1000)
China
(N=991)
South America
(N=869)
Demographic variables
Age (years) 59 (15) 53 (14) 56 (15) 57 (14) 56 (16) 66 (16) 67 (15)
Male sex 3495 (61%) 662 (52%) 531 (62%) 474 (59%) 721 (72%) 570 (57%) 537 (61%)
Clinical variables
Body-mass index (kg/m2) 26 (6·1) 26 (6·1) 23 (6·2) 26 (6) 30 (6) 24 (6.3) 29 (6.2)
Systolic blood pressure (mm Hg) 125 (23) 124 (21) 125 (21) 128 (23) 126 (22) 126 (22) 123 (24)
Diastolic blood pressure (mm Hg) 76 (13) 79 (16) 77 (12) 76 (12) 72 (13) 77 (12) 75 (13)
History of diabetes mellitus 1728 (29%) 201 (17%) 217 (26%) 328 (41%) 541 (57%) 216 (19%) 225 (21%)
History of chronic kidney disease 487 (8%) 45 (4%) 30 (3%) 107 (13%) 121 (12%) 67 (6%) 117 (11%)
Current tobacco use 554 (6%) 78 (4%) 102 (7%) 81 (6%) 132 (7%) 106 (8%) 55 (4%)
History of chronic obstructive pulmonary
disease
450 (6%) 26 (2%) 139 (16%) 35 (4%) 41 (4%) 99 (8%) 110 (10%)
Reduced left ventricular ejection fraction
(<40%)*
2486 (50%) 526 (54%) 392 (53%) 247 (39%) 743 (73%) 212 (27%) 366 (53%)
Valve disease*† 2286 (46%) 582 (57%) 309 (42%) 265 (40%) 479 (50%) 306 (41%) 345 (48%)
NYHA functional class III or IV 2470 (40%) 702 (56%) 415 (50%) 127 (16%) 360 (37%) 574 (56%) 292 (32%)
Admission for heart failure in previous year 1567 (27%) 420 (36%) 122 (14%) 285 (35%) 219 (22%) 308 (34%) 213 (28%)
Recruited as hospital inpatient 2105 (34%) 616 (48%) 389 (45%) 187 (23%) 310 (31%) 367 (35%) 236 (26%)
Main cause of heart failure
Ischaemic heart disease 2433 (39%) 242 (20%) 399 (46%) 449 (56%) 521 (50%) 519 (45%) 303 (25%)
Hypertensive heart disease 1096 (17%) 392 (35%) 116 (14%) 115 (15%) 93 (10%) 165 (14%) 215 (21%)
Idiopathic dilated cardiomyopathy 838 (12%) 212 (14%) 118 (11%) 31 (3%) 220 (18%) 141 (15%) 116 (15%)
Valvular heart disease 739 (11%) 185 (11%) 135 (12%) 122 (12%) 93 (8%) 97 (11%) 107 (13%)
Endocrine or metabolic disease 224 (4%) 67 (5%) 36 (4%) 46 (6%) 14 (1%) 31 (3%) 30 (4%)
Hypertrophic cardiomyopathy 71 (1%) 3 (0·2%) 15 (1·7%) 7 (0·8%) 6 (0·6%) 14 (1·6%) 26 (3·4%)
Congenital heart disease 57 (0·5%) 4 (0·1%) 10 (0·5%) 11 (1%) 7 (0·3%) 13 (1·2%) 12 (1·5%)
Alcohol or drugs 68 (0·4%) 22 (0·7%) 6 (0·2%) 6 (0·3%) 19 (0·6%) 7 (0·4%) 8 (0·5%)
Cardiac medications
Beta blocker 3768 (67%) 634 (48%) 495 (57%) 543 (66%) 866 (85%) 594 (60%) 636 (73%)
ACE inhibitor 2924 (49%) 774 (59%) 449 (51%) 383 (46%) 641 (62%) 336 (34%) 341 (40%)
Angiotensin receptor blocker 1443 (24%) 226 (19%) 143 (17%) 215 (27%) 191 (20%) 323 (29%) 345 (36%)
ACE inhibitor/angiotensin receptor blocker 4322 (74%) 990 (78%) 586 (68%) 593 (73%) 828 (82%) 654 (64%) 671 (76%)
Aldosterone inhibitor 2913 (48%) 787 (59%) 421 (47%) 229 (27%) 480 (46%) 533 (56%) 463 (55%)
Loop diuretic 4414 (78%) 1214 (94%) 691 (81%) 361 (45%) 878 (88%) 598 (61%) 672 (78%)
Digoxin 1550 (26%) 443 (32%) 224 (25%) 241 (29%) 186 (18%) 262 (29%) 194 (25%)
Long-acting nitrate 1075 (15%) 58 (5%) 114 (13%) 188 (23%) 256 (25%) 360 (31%) 99 (8%)
Aspirin 3293 (56%) 558 (46%) 440 (52%) 469 (58%) 753 (75%) 620 (58%) 453 (46%)
Warfarin 858 (14%) 222 (17%) 84 (10%) 81 (10%) 194 (20%) 97 (10%) 180 (22%)
Socioeconomic factors
Illiterate 1319 (15%) 495 (43%) 229 (29%) 20 (2%) 303 (36%) 210 (15%) 62 (4%)
Education ≥ grade 5 level 2496 (42%) 714 (61%) 428 (56%) 161 (20%) 464 (54%) 423 (37%) 306 (29%)
Education grade 6 –10 level 1382 (23%) 178 (13%) 135 (15%) 312 (37%) 207 (19%) 285 (30%) 265 (32%)
Education grade 11–12 level 1047 (17%) 208 (13%) 183 (18%) 175 (19%) 161 (13%) 170 (18%) 150 (19%)
Education post secondary 896 (13%) 193 (13%) 112 (11%) 163 (18%) 168 (13%) 113 (11%) 147 (17%)
Dwelling in rural area 2080 (36%) 414 (42%) 492 (48%) 298 (37%) 261 (26%) 434 (44%) 181 (21%)
Health insurance 3345 (61%) 416 (33%) 166 (19%) 357 (44%) 743 (74%) 906 (91%) 757 (76%)
Data adjusted for age and sex. Data are mean (SD) or n (%). NYHA=New York Heart Association. ACE=angiotensin converting enzyme. *Data available for 4716 patients
(81%) who had echocardiograms. †Valve disease defined as moderate or greater stenosis or regurgitation in a cardiac valve on echocardiogram; may co-exist with any of the
main causes of heart failure.
Table 1: Baseline demographic and clinical characteristics
Downloaded for Anonymous User (n/a) at Saint Louis University from ClinicalKey.com by Elsevier on September 30,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.

Articles
e668
www.thelancet.com/lancetgh Vol 5 July 2017
change in randomness (entropy) due to the introduction
or removal of a covariate.15 The explained risk statistics
were calculated using R software version 3.2.5,
(R Foundation for Statistical Computing, Vienna,
Austria). All other analyses were performed using SAS
version 9.4 for UNIX (SAS Institute Inc, Cary, NC, USA).
Given the observational nature of this study, reported
pvalues are considered as informative statistics. We set
the threshold of p<0·001 for a finding to be considered as
persuasive evidence of a meaningful finding, because
of the number of research questions addressed in
analysis
Results
Between September 10, 2012, and February 21, 2014, we
screened 7176 consecutive patients for eligibility and
Africa India Southeast Asia Middle East South America China Overall
0
Mortality at 1 year (%)
Region
5
10
15
20
25
30
35 Unknown cause of death
Non-cardiac death
Cardiac death
Figure 1: Unadjusted mortality at 1 year, by region and cause
Univariable analysis Multivariable analysis
Hazard ratio
(95% CI)
p χ²* Hazard ratio
(95% CI)
p χ²
Demographic variables
Age (per 10 year increase) 1 (1·0–1·1) 0·28 1·2 1·1 (1·05–1·17) 0·005 7·7
Male sex 1 (0·9–1·2) 0·84 0 1·0 (0·9–1·3) 0·7 0·1
Clinical variables
Enrolled as hospital inpatient 2·9 (2·5–3·6) <0·0001 158 1·9 (1·6–2·2) <0·0001 44
Valve disease on echocardiogram†2·0 (1·7–2·4) <0·0001 64 1·6 (1.3 – 1.9) <0·0001 25
Admission for hearth failure in previous year 1·9 (1·6–2·2) <0·0001 60 1·6 (1·3–1·9) <0·0001 25
History of chronic kidney disease 1·7 (1·3–2·2) <0·0001 19 1·9 (1·5–2·5) <0·0001 24
Systolic blood pressure (per 10 mmHg increase) 0·9 (0·8–0·9) <0·0001 30 0.92 (0.88 – 0.96) <0·0001 17
NYHA functional class III or IV (vs class I and II) 2·2 (1·8–2·6) <0·0001 83 1·4 (1·2–1·7) 0·0003 13
Body mass index (per 1 kg/m2 increase) 0·94 (0·93–0·96) <0·0001 52 0·97 (0·96–0·99) 0·001 11
History of chronic obstructive pulmonary disease 1·6 (1·3–2·1) 0·0002 14 1·6 (1·2–2·1) 0·001 11
History of diabetes mellitus 1·1 (0·9–1·3) 0·26 1·3 1·2 (1·0–1·5) 0·06 3·5
Reduced left ventricular function (EF <40%) 1·3 (1·1–1·5) 0·003 8·7 1·1 (0·9–1·4) 0·2 1·6
Current tobacco use 0·8 (0·6–1·1) 0·22 1·5 0·9 (0·7–1·3) 0·69 0·1
Ischaemic cause of heart failure 1·0 (0·8–1·1) 0·66 0·2 1·0 (0·8–1·3) 0·72 0·1
Medications
ACE inhibitor or ARB use 0·7(0·6–0·8) <0·0001 21 0·8 (0·7–0·9) 0·01 5·8
Digoxin use 1·0 (0·8–1·2) 0·83 0 0·80 (0·7–0·9) 0·03 4·7
Beta blocker use 0·6 (0·5–0·7) <0·0001 33 0·9 (0·7–1·1) 0·17 1·9
Aldosterone inhibitor use 1·0 (0·9–1·2) 0·60 0·3 1·0 (0·8–1·2) 0·95 0
Socioeconomic factors
Illiterate 1·8 (1·5–2·1) <0·0001 40 1·2 (0·9–1·5) 0·13 2·2
Dwelling in rural area 1·4 (1·2–1·7) <0·0001 17 1·2 (1·0–1·4) 0·12 2·4
No health insurance 1·6 (1·3–1·9) <0·0001 29 0·9 (0·7–1·0) 0·12 2·4
Region (vs South America)
Africa 3·7 (2·7–5·1) <0·0001 60 3·8 (2·6–5·5) <0·0001 48
India 2·9 (2·1–4·1) <0·0001 37 2·9 (1·9–4·3) <0·0001 28
Southeast Asia 1·9 (1·3–2·8) 0·0004 12 2·6 (1·7–3·9) <0·0001 21
Middle East 1·2 (0·8–1·7) 0·43 0·6 1·3 (0·9–1·9) 0·23 1·4
China 0·8 (0·6–1·3) 0·86 0·4 0·7 (0·4–1·1) 0·14 2·1
ARB=angiotensin receptor blocker; EF=ejection fraction; NYHA=New York Heart Association.*Degrees of freedom=1 for all tests. †Defined as moderate or greater stenosis or
regurgitation in a cardiac valve on echocardiography. N=4347; 565 deaths; global χ2=519; p <0·0001
Table 2: Variables associated with all-cause mortality at 1 year
Downloaded for Anonymous User (n/a) at Saint Louis University from ClinicalKey.com by Elsevier on September 30,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.

Articles
www.thelancet.com/lancetgh Vol 5 July 2017 e
669
enrolled 5823 participants (81%). Table 1 shows baseline
characteristics, adjusted for age and sex, by region;
further cohort data have been reported previously.16 Mean
age of the cohort was 59 years (SD 15); 2328 (39%) were
female, 3549 (64%) had a history of hypertension,
1728(29%) had a history of diabetes mellitus, 487 (8%)
had a history of chronic kidney disease, and 2486 (50%)
had a LVEF less than 40%. The cause of heart failure was
ischaemia in 2433 (39%), hypertension in 1096 (17%),
idiopathic dilated cardiomyopathy in 838 (12%), and
primary valve disease in 739 (11%). 1319 (15%) of
participants were illiterate and 2477 (39%) had no health
insurance. As per our study design, 3695 participants
(66%) were recruited from outpatient clinics and
2105 (34%) from inpatient hospital wards; 2080 (36%)
were recruited in rural settings.
Follow-up data at 12 months were available in
5689/5823 (98%) of patients: 811/811 (100%) in southeast
Asia, 989/991 (100%) in China, 856/858 (100%) in India,
995/1000 (99%) in the Middle East, 830/869 (95%) in
South America, and 1208/1294 (93%) in Africa.
Data from echocardiograms were available for
4716patients (81%). In model 2, with data derived from
echo cardiograms excluded, the number of patients for
which all data were available increased from 4347patients
with 565 deaths (in model 1) to 5341 patients with
766deaths. The events per variable (EPV) of 565 deaths/
21 variables=27 (model 1), and 766 deaths/19 variables=40
(model 2), are well above the recommended threshold of
10 EPV, therefore minimising bias in parameter esti-
mates of our Cox proportional hazards model.17,18
Findings from model 1 are reported here, and results
from model 2 are shown in the appendix.
Unadjusted all-cause mortality within 1 year was
16·5% (95% CI 15·4–17·6), and varied substantially
between regions. Mean age at time of death was 56 years
(SD 16) in Africa, 59 years (15) in India, 57 years (15) in
southeast Asia, 60 years (14) in the Middle East,
69 years (13) in China, and 72 years (14) in South
America. Despite being in the youngest cohorts at
baseline, patients in Africa and India had the highest
mortality (33·6% [95% CI 30·2–37·4]), 23·3%
[19·9–27·0]), respectively), and participants from south-
east Asian had an intermediate rate (15·0% [12·4–18·0]),
compared with patients in China, South America and
the Middle East patients who had the lowest rates of
death (7 ·3% [5·7–9·3], 9·1% [7·1–11·4] and 9·4%
[7·5–11·5], respectively).
Of the 858 deaths within 1 year, cardiac deaths were
more common than non-cardiac deaths (398 [46%] and
136 [16%], respectively), and deaths from an unknown
cause (324 [38%]). The proportion of deaths from an
unknown cause was much higher in Africa (189 deaths,
55%) than in other regions (figure 1). When we excluded
Africa from analysis, 282 (55%) deaths were from cardiac
causes, 99 (19%) from non-cardiac causes, and 135 (26%)
from unknown cause.
In unadjusted analyses, variables associated with death
within 1 year included clinical variables, medi cations,
socioeconomic variables and region (table 2). After
multivariable adjustment, clinical and demographic
variables independently associated with death within
1 year included: age (hazard ratio [HR] 1·1; 95% CI
1·05–1·17), systolic blood pressure (0·92 per 10 mm Hg
increase; 0·88–0·96), body mass index (0·97 per 1 kg/m2
increase; 0·96–0·99), history of chronic kidney disease
(1·9; 1·5–2·5), New York Heart Association (NYHA)
functional class III or IV heart failure (1·4; 1·2–1·7),
enrolment as a hospital inpatient (1·9; 1·6–2·2),
admission for heart failure in the previous year (1·6;
1·3–1·9), history of chronic obstructive pulmonary
disease (1·6; 1·2–2·1), and valve disease shown on echo-
cardiograms (1·6; 1·3–1·9). Medications associated with
death rates at 1 year were ACE inhibitors or angiotensin
receptor blockers (0·8; 0·7–0·9), and di goxin use at
enrolment (0·8; 0·7–0·9). With South America as the
reference, the variable of region was associated with
1-year mortality in Africa (3·8; 2·6–5·5), India (2·9;
1·9–4·3), and southeast Asia (2·6; 1·7–3·9). Figure 2
shows unadjusted and adjusted survival curves.
Figure 2: Unadjusted and adjusted survival curves by region
Number at risk
China
India
Southeast Asia
Middle East
Africa
South America
03
991
858
811
1000
1294
869
969
768
776
973
1070
811
952
722
747
952
984
791
921
579
723
914
914
772
731
667
642
695
742
627
6 9 12
Unadjusted log-rank test=<0·0001
Adjusted log-rank p value (compared with South America):
China=0·1402
India<0·0001
Southeast Asia<0·0001
Middle East=0·2349
Africa<0·0001
0
20
40
60
80
100
Survival (%)
Adjusted
03 6
Months
9 12
0
20
40
60
80
100
Survival (%)
Unadjusted
China
South America
Middle East
Southeast Asia
India
Africa
Downloaded for Anonymous User (n/a) at Saint Louis University from ClinicalKey.com by Elsevier on September 30,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.

Articles
e670
www.thelancet.com/lancetgh Vol 5 July 2017
In our analyses of cause-specific mortality, we noted
that for cardiac death alone, results were similar to those
for all-cause deaths: patients in Africa (HR 2·3; 95% CI
1·4–3·8) and India (2·5; 1·5–4·2) were at highest hazard
of cardiac death, followed by those in southeast Asian
(2·0; 1·2–3·4), with no increased hazard in patients in
the Middle East (1·0; 0·6–7·7) or China (0·7, 0·4–1·2),
compared with the South American cohort. Likewise, for
non-cardiac deaths, participants in Africa (2·7; 1·2–6·2)
and India (2·4; 1·0–5·9) were at highest hazard, followed
by participants in southeast Asia (2·0; 0·8–4·9), the
Middle East (1·4, 0·6–3.4) and China (0·5; 0·2–1·3),
compared with South American patients.
When data were stratified by patients’ admission status,
we noted that of the 2105 patients recruited as hospital
inpatients, unadjusted mortality within 1 year was
markedly higher (30·6%, 95% CI 28·0–33·3), than in the
3695 patients recruited from outpatient clinics (9·6%,
8·6–10·6) (figure 3). As in the overall cohort, patients
recruited as hospital inpatients in Africa had the highest
adjusted hazard of death within 1 year (HR 3·7; 95% CI
2·2–6·2), followed by inpatients in India (2·4; 1·4–4·0)
and southeast Asia (2·4; 1·3–4·4), with no increased
hazard of death in the Middle East (1·1; 0·6–1·9), or
China (0·4; 0·2–0·8), compared with the South American
cohort. For patients recruited as outpatients, findings
were similar, with highest hazard of death in Africa (4·2;
2·3–7·5) and India (4·1; 2·2–7·6), followed by southeast
Asia (3·3; 1·9–5·9), the Middle East (1·7; 1·0–3·0), and
China (1·2; 0·8–2·2).
Of the 5823 participants in INTER-CHF, 3189 (55%)
had all required variables measured for the Boston
Criteria for heart failure. Of these 3189 participants,
2597(81%) had either definite or probable heart failure. In
this stratum patients with either definite of probable heart
failure, regional patterns in hazard ratios noted for the
overall cohort remained: participants in Africa (HR 2·8;
95% CI 1·7–4·3) and India (2·7; 1·6–4·3) had the highest
hazard of death, followed by those in southeast Asia
(2·2;1·3–3·9), the Middle East (0·9; 0·6–1·5), and China
(0·6; 0·3–1·1), compared with South American patients.
We used the explained-risk statistic15 to determine the
relative contributions of each of demographic and clinical
variable, medications, socioeconomic variable and region
to the risk of death within 1 year in the overall cohort.
These variables, together, explained 46% of the risk of
death. Of the explained risk, clinical variables alone
accounted for 53%, and region for 35%, with lesser
contributions from demographic variables, medications
,and socioeconomic variables (2% for each; table 3).
Discussion
We have observed substantial variations in mortality
between regions in patients with heart failure—rates
were higher in poorer regions and lower in richer regions,
and dierences persisted after multivariable modelling.
Although data from international studies in clinic
outpatients (that is, patients with chronic heart failure)
are few, they do also suggest regional dierences in heart
failure outcomes. In an analysis of regional dierences
observed in the PARADIGM-HF study19 (a randomised
trial of sacubitril/valsartan versus enalapril, in which
10 521 patients from 1043 centres in 47 countries entered
the run-in period), Kristensen and colleagues analysed
data from 8399 patients with chronic heart failure with
reduced LVEF in six geographical regions. They showed
Africa India Southeast Asia Middle East South America China Overall
0
Mortality at 1 year (%)
Region
10
20
30
40
50
60
70 Outpatients
Inpatients
Figure 3: Unadjusted mortality at 1 year, by admission status at recruitment and region
Value (%) Standard
error (%)
Demographic variables 0·2 0·4
Demographic + clinical variables 35 2·7
Only demographic variables 0·1 0·1
Only clinical variables 35 2·7
Demographic + clinical variables +
medications
36 2·7
Only demographic variables 0 0·1
Only clinical variables 33 2·8
Only medications 2·1 1
Demographic + clinical + medications +
socioeconomic variables
39 2·6
Only demographic variables 0·2 0·3
Only clinical variables 29 2·8
Only medications 1·7 0·9
Only socioeconomic variables 3·4 1·2
Demographic + clinical + medications +
socioeconomic variables + region
46 2·4
Only demographic variables 0·7 0·6
Only clinical variables 25 2·6
Only medications 1·1 0·6
Only socioeconomic variables 0·7 0·5
Only region 12 3·0
Demographic, clinical, medication, socioeconomic variables and region account
for 46% of explained relative risk of death in the overall cohort. The remainder of
risk is unexplained. N=4347; 565 events.
Table 3: Explained risk analysis for independent variables included in
Cox proportional hazards (model 1) of all-cause mortality in the overall
cohort at 1 year
Downloaded for Anonymous User (n/a) at Saint Louis University from ClinicalKey.com by Elsevier on September 30,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.

Articles
www.thelancet.com/lancetgh Vol 5 July 2017 e
671
that the adjusted risk of cardiovascular death was greater
in Latin America and Asia-Pacific than in North America
and Europe; patients from the Asia-Pacific region were
also significantly younger than patients from western
countries. A pooled analysis of 53 local and regional
studies also reported that the mean age of patients with
heart failure in LMIC was a decade younger than in high-
income countries.8 In INTER-CHF, patients in Africa,
India, and southeast Asia were approximately 10 years
younger than patients in South America and China, but
had much higher mortality. This finding may be related
to patients with heart failure presenting later for medical
care (that is, when they are sicker) in low-income com-
pared with high-income regions, and late presen tation
being associated with a worse prognosis.20 This finding
may also be related to the overall lower life expectancies
in LMIC compared with high-income countries.21
Study participants with acute heart failure enrolled as
inpatients in clinical trials have shown that patients from
North America and Europe had lower mortality rates
than did patients from South America and Asia, with few
data from the Middle East, Africa, South Asia or
China.10,22,23 In a retrospective analysis of the ASCEND-HF
trial24 (a randomised trial of nesiritide versus placebo in
7141 hospitalised patients with acute heart failure from
398 sites in 30 countries), patients from Latin America
had the highest adjusted 180-day mortality (17·3%),
followed by western Europe, North America, Asia-Pacific,
with Central Europe (9·3%) having the lowest mortality.
Local registries have also reported data on mortality in
inpatients with heart failure. A study of 1009 patients
with heart failure admitted to hospitals in sub-Saharan
Africa reported an unadjusted 6-month mortality of
17·8%, which is probably an underestimate because of
the substantial losses to follow-up.25 In a registry of
1205patients hospitalised with heart failure in India, the
90-day mortality rate was 24·3%.26 In a hospital registry
of 5005 patients with heart failure from Gulf states,
mortality at 1 year was 20%.27 Therefore, data from other
studies accord with our findings of variations in
outcomes between regions and also high mortality by
region. Our study also provides new data on outcomes in
outpatients with chronic heart failure, and in several
countries and regions not previously studied.
In INTER-CHF, both cardiac and non-cardiac factors
were associated with all-cause mortality, consistent with
data from western populations.28 However, in our study,
multivariable adjustment including these factors only
explained about half the mortality risk, the remainder
being unexplained. Importantly, when compared with
other regions in this study, patients in Africa were much
younger, more symptomatic, more often treated with
digoxin, had little education, low rates of health-
insurance, and were more often from a rural area. Similar
patterns were observed in India. These were the countries
with highest mortality. Therefore, country-level, environ-
mental, wealth, and healthcare infrastructure, quality and
access factors would need to be addressed if these
dierences in outcomes were to be further explained.
Indeed, in a systematic review by Callender and
colleagues8 of earlier regional heart failure studies,
higher mortality in lower-income regions was thought to
be partly the result of diering health systems, quality of
care, and variations in health-care access. In addition,
genetic and genomic factors may also influence regional
variations in heart failure outcomes, but were not
recorded in INTER-CHF. Therefore, measuring and
understanding health systems, environmental and
societal issues, and genetic and genomic markers in
future international heart failure studies could help in
understanding the causes of the regional variations in
outcomes. Information on health systems could assist in
developing strategies to further reduce mortality in
patients with heart failure.
The 5823 patients with heart failure enrolled in this
study from 108 centres in 16 countries constitute a large
study of heart failure in Africa, Asia, the Middle East and
South America. However, some randomised trials in
patients with heart failure that included dierent regions
were larger.19,24 By using a standardised protocol and
approach, comparisons of patient characteristics and
outcomes between regions are possible. Furthermore,
the high follow-up rate of 98% gives confidence in our
findings. As with most registries, we were unable to
randomly sample clinical sites or populations, for
practical reasons. However, data from INTER-CHF are
similar to data collected elsewhere in other studies.8,19
Although representativeness and potential ascertainment
bias are inherent challenges in data collection for
registries, the consistent data from other studies from
specific regions validate our findings. Furthermore,
variability in decisions about the causes of heart failure
and death are possible, but are a reality of clinical
practice, not only in LMIC but also in western countries.
Unpublished data from the PURE study, which is
community based, also show that mortality after a
diagnosis of heart failure was lowest in high-income
countries, intermediate in middle-income countries and
highest in low-income countries.
In this heart failure cohort study in 16 countries,
mortality was highest in Africa and India, intermediate
in southeast Asia, and lowest in China, South America,
and the Middle East. These regional dierences persisted
after multivariable adjustment including both cardiac
and non-cardiac factors. Therefore, the large regional
dierences in death in patients with heart failure could
be influenced by unmeasured variables, requiring
further study in large, global cohorts of patients with
heart failure.
Contributors
HD contributed to study design, literature search, data analysis, data
interpretation, figures, and writing. KT contributed to study design, data
interpretation, and writing. JZ, AR, KFA, AE, LP-V, PL-J, KK, KY, AO, KS,
CM, FL, DP, AB, ME, AD, and KT contributed to study design, data
collection, and data interpretation. EB-C contributed to data
Downloaded for Anonymous User (n/a) at Saint Louis University from ClinicalKey.com by Elsevier on September 30,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.

Articles
e672
www.thelancet.com/lancetgh Vol 5 July 2017
interpretation and writing. KB and SI contributed to data analysis, data
interpretation, and figures. MHY,MDH, and KH contributed to study
design and data interpretation. AG contributed to study design, and data
collection. RMcontributed to study design, and data interpretation. SIB
contributed to data analysis, data interpretation, and writing.
SYcontributed to study design, data analysis, data interpretation, and
writing.
Declaration of interests
JZ has received grants from Bayer. DP has received grants from
Novartis. MH has received grants from the World Heart Federation.
Allother authors declare no competing interests.
Acknowledgments
Novartis sta provided assistance and support for the study. We thank
the Saudi Heart Association, and the Deanship of Scientific Research at
King Saud University, Riyadh, Saudi Arabia (Research group number:
RG -1436-013) for research coordination and assistance.
References
1 Ambrosy AP, Fonarow GC, Butler J, et al. The global health and
economic burden of hospitalizations for heart failure: lessons
learned from hospitalized heart failure registries. J Am Coll Cardiol
2014; 63: 1123–33.
2 Cook C, Cole G, Asaria P, Jabbour R, Francis DP. The annual global
economic burden of heart failure. Int J Cardiol 2014; 171: 368–76.
3 Moran AE, Oliver JT, Mirzaie M, et al. Assessing the Global Burden
of Ischemic Heart Disease: Part 1: Methods for a Systematic Review
of the Global Epidemiology of Ischemic Heart Disease in 1990 and
2010. Glob Heart 2012; 7: 315–29.
4 Khatibzadeh S, Farzadfar F, Oliver J, Ezzati M, Moran A. Worldwide
risk factors for heart failure: a systematic review and pooled
analysis. Int J Cardiol 2013; 168: 1186–94.
5 Crespo-Leiro MG, Anker SD, Maggioni AP, et al. European Society
of Cardiology Heart Failure Long-Term Registry (ESC-HF-LT): 1-year
follow-up outcomes and dierences across regions: Eur J Heart Fail
2016; 18: 613–25.
6 Howlett JG, Ezekowitz JA, Podder M, et al. Global variation in
quality of care among patients hospitalized with acute heart failure
in an international trial: findings from the acute study clinical
eectiveness of nesiritide in decompensated heart failure trial
(ASCEND-HF). Circ Cardiovasc Qual Outcomes 2013; 6: 534–42.
7 Kristensen SL, Kober L, Jhund PS, et al. International geographic
variation in event rates in trials of heart failure with preserved and
reduced ejection fraction. Circulation 2015; 131: 43–53.
8 Callender T, Woodward M, Roth G, et al. Heart failure care in
low- and middle-income countries: a systematic review and
meta-analysis. PLoS Med 2014; 11: e1001699.
9 Dokainish H, Teo K, Zhu J, et al. Heart failure in low- and middle-
income countries: Background, rationale, and design of the
INTERnational Congestive Heart Failure Study (INTER-CHF).
AmHeart J 2015; 170: 627–34.
10 Greene SJ, Fonarow GC, Solomon SD, et al. Global variation in
clinical profile, management, and post-discharge outcomes among
patients hospitalized for worsening chronic heart failure: findings
from the ASTRONAUT trial. Eur J Heart Fail 2015; 17: 591–600.
11 Yancy CW, Jessup M, Bozkurt B, et al. 2013 ACCF/AHA guideline
for the management of heart failure: a report of the American
College of Cardiology Foundation/American Heart Association Task
Force on Practice Guidelines. J Am Coll Cardiol 2013; 62: e147–239.
12 The Criteria Committee of the New York Heart Association.
Nomenclature and Criteria for Diagnosis of Diseases of the Heart
and Great Vessels, 9th ed. Boston: Little, Brown & Co, 1994: 253–56.
13 Di Bari M, Pozzi C, Cavallini MC, et al. The Diagnosis of Heart
Failure in the Community. Comparative Validation of Four Sets of
Criteria in Unselected Older Adults: The ICARe Dicomano Study.
JAm Coll Cardiol 2004; 44: 1601–08.
14 Pocock SJ, Ariti CA, McMurray JJV, et al. Predicting survival in
heart failure: a risk score based on 39 372 patients from 30 studies.
EurHeart J 2012; 34: 1404–13.
15 Heller G. A measure of explained risk in the proportional hazards
models. Biostatistics 2012; 13: 315–25.
16 Dokainish H, Teo K, Zhu J, et al. Heart Failure in Africa, Asia, the
Middle East and South America: The INTER-CHF study.
IntJCardiol 2016; 204: 133–41.
17 Peduzzi P, Concato J, Feinstein AR, Holford TR. Importance of
events per independent variable in proportional hazards regression
analysis. II. Accuracy and precision of regression estimates.
JClinEpidemiol 1995; 48: 1503–10.
18 Concato J, Peduzzi P, Holford TR, Feinstein AR. Importance of
events per independent variable in proportional hazards analysis.
I.Background, goals, and general strategy. J Clin Epidemiol 1995;
48: 1495–501.
19 Kristensen SL, Martinez F, Jhund PS, et al. Geographic variations
in the PARADIGM-HF heart failure trial. Eur Heart J 2016;
37:3167–74.
20 Ambrosy AP, Gheorghiade M, Chioncel O, Mentz RJ, Butler J.
Global perspectives in hospitalized heart failure: regional and
ethnic variation in patient characteristics, management, and
outcome. Curr Heart Fail Rep 2014; 11: 416–27.
21 Wang H, Dwyer-Lindgren L, Lofgren KT, et al. Age-specific and
mortality in 187 countries, 1970–2010: a systematic analysis for the
Global Burden of Disease Study 2010. Lancet 2012; 380: 2071–94.
22 Blair JE, Zannad F, Konstam MA, et al. Continental dierences in
clinical characteristics, management, and outcomes in patients
hospitalized with worsening heart failure results from the
EVEREST program. J Am Coll Cardiol 2008; 52: 1640–48.
23 Greene SJ, Fonarow GC, Solomon SD, et al. Global variation in
clinical profile, management, and post-discharge outcomes among
patients hospitalized for worsening chronic heart failure: findings
from the ASTRONAUT trial. Eur J Heart Fail 2015; 17: 591–600.
24 Metra M, Mentz RJ, Hernandez AF, et al. Geographic dierences
in patients in a global acute heart failure clinical trial (from the
ASCEND-HF Trial). Am J Cardiol 2016; 117: 1771–78.
25 Sliwa K, Davison BA, Mayosi BM, et al. Readmission and death
after an acute heart failure event: predictors and outcomes in sub-
Saharan Africa: results from the THESUS-HF registry. Eur Heart J
2013; 34: 3151–59.
26 Harikrishnan S, Sanjay G, Anees T, et al. Clinical presentation,
management, in-hospital and 90-day outcomes of heart failure
patients in Trivandrum, Kerala, India: the Trivandrum Heart Failure
Registry. Eur J Heart Fail 2015; 17: 794–800.
27 Sulaiman K, Pandurange P, Al-Zakwani, et. al. Clinical
characteristics, management, and outcomes of acute heart failure
patients: observations from the Gulf acute heart failure registry
(GulfCARE). Eur J Heart Fail 2015; 17: 374–84.
28 Ather S, Chan W, Bozkurt B, Aguilar D, Ramasubbu K, Zachariah
AA, Wehrens XH, Deswal A. Impact of noncardiac comorbidities
on morbidity and mortality in a predominantly male population
with heart failure and preserved versus reduced ejection fraction.
JAm Coll Cardiol 2012; 59: 998–1005.
Downloaded for Anonymous User (n/a) at Saint Louis University from ClinicalKey.com by Elsevier on September 30,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.