new england

The
journal of medicine
established in 1812 August 24, 2017 vol. 377 no. 8

Global, Regional, and National Burden of Rheumatic Heart

Disease, 1990–2015
David A. Watkins, M.D., M.P.H., Catherine O. Johnson, Ph.D., M.P.H., Samantha M. Colquhoun, Ph.D.,
Ganesan Karthikeyan, M.D., D.M., Andrea Beaton, M.D., Gene Bukhman, M.D., Ph.D.,
Mohammed H. Forouzanfar, M.D., Ph.D., Christopher T. Longenecker, M.D.,
Bongani M. Mayosi, M.B., Ch.B., D.Phil., George A. Mensah, M.D., Bruno R. Nascimento, M.D., Ph.D.,
Antonio L.P. Ribeiro, M.D., Ph.D., Craig A. Sable, M.D., Andrew C. Steer, Ph.D.,
Mohsen Naghavi, M.D., M.P.H., Ph.D., Ali H. Mokdad, Ph.D., Christopher J.L. Murray, M.D., D.Phil.,
Theo Vos, M.D., Ph.D., Jonathan R. Carapetis, M.B., B.S., Ph.D., and Gregory A. Roth, M.D., M.P.H.​​

a bs t r ac t

BACKGROUND
Rheumatic heart disease remains an important preventable cause of cardiovascular The authors’ affiliations are listed in the
death and disability, particularly in low-income and middle-income countries. We Appendix. Address reprint requests to
Dr. Watkins at the Division of General In-
estimated global, regional, and national trends in the prevalence of and mortality due ternal Medicine, Department of Medi-
to rheumatic heart disease as part of the 2015 Global Burden of Disease study. cine, University of Washington, 325 9th
Ave., Box 359780, Seattle, WA 98104, or at
METHODS ­davidaw@​­uw​.­edu.
We systematically reviewed data on fatal and nonfatal rheumatic heart disease for the N Engl J Med 2017;377:713-22.
period from 1990 through 2015. Two Global Burden of Disease analytic tools, the DOI: 10.1056/NEJMoa1603693
Cause of Death Ensemble model and DisMod-MR 2.1, were used to produce estimates Copyright © 2017 Massachusetts Medical Society.

of mortality and prevalence, including estimates of uncertainty.

RESULTS
We estimated that there were 319,400 (95% uncertainty interval, 297,300 to 337,300)
deaths due to rheumatic heart disease in 2015. Global age-standardized mortality due
to rheumatic heart disease decreased by 47.8% (95% uncertainty interval, 44.7 to 50.9)
from 1990 to 2015, but large differences were observed across regions. In 2015, the
highest age-standardized mortality due to and prevalence of rheumatic heart disease
were observed in Oceania, South Asia, and central sub-Saharan Africa. We estimated
that in 2015 there were 33.4 million (95% uncertainty interval, 29.7 million to 43.1
million) cases of rheumatic heart disease and 10.5 million (95% uncertainty interval,
9.6 million to 11.5 million) disability-adjusted life-years due to rheumatic heart dis-
ease globally.
CONCLUSIONS
We estimated the global disease prevalence of and mortality due to rheumatic heart
disease over a 25-year period. The health-related burden of rheumatic heart disease
has declined worldwide, but high rates of disease persist in some of the poorest
regions in the world. (Funded by the Bill and Melinda Gates Foundation and the
Medtronic Foundation.)

n engl j med 377;8 nejm.org August 24, 2017 713

The New England Journal of Medicine
Downloaded from nejm.org on October 29, 2023. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

R
heumatic heart disease is a sequela We identified rheumatic heart disease–specific
of acute rheumatic fever,1 which is usu- deaths from vital registration systems using codes
ally a disease of poverty associated with from the International Classification of Diseases, 9th
overcrowding, poor sanitation, and other social Revision (ICD-9) and 10th Revision (ICD-10) (Table
A Quick Take is
determinants of poor health.2,3 The near elimina- S1 in the Supplementary Appendix). In total,
available at tion of acute rheumatic fever and reduction in the 10,049 site-years of vital registration data from
NEJM.org rates of rheumatic heart disease in high-income 132 countries were used. Deaths attributed to
countries during the late 20th century was attrib- ill-defined or nonspecific causes (e.g., “heart
uted in part to improvements in socioeconomic disease, unspecified” [ICD-10 code I51.9]) or
conditions and the widespread use of penicillin G intermediate causes (i.e., causes, such as “heart
benzathine to treat streptococcal pharyngitis.4,5 failure” [ICD-10 code I50], that are not the under-
The remaining burden of rheumatic heart dis- lying disease that initiated the chain of events
ease is found mostly in low-income and middle- leading to death) were reassigned to accepted
income countries and among immigrants and causes of death, including rheumatic heart dis-
older adults in high-income countries.6,7 ease, with the use of algorithms developed for
Guidelines for the prevention and treatment GBD 2015.20 We performed a sensitivity analysis
of acute rheumatic fever and rheumatic heart in which we evaluated uncertainty in the reas-
disease were originally released by the World signment to rheumatic heart disease of deaths
Health Organization (WHO) more than 60 years that had originally been coded to left heart fail-
ago.8 Many countries have had striking reduc- ure, the ICD-10 code most commonly reassigned
tions in mortality related to acute rheumatic fever to rheumatic heart disease.
and rheumatic heart disease; these reductions The GBD 2015 Cause of Death Ensemble
can be credited to the implementation of control model was used to produce estimates of the frac-
programs and improvements to health systems.9,10 tion of deaths caused by rheumatic heart disease
Despite these improvements, high prevalences of according to age, sex, and location for each year
and mortality due to rheumatic heart disease from 1980 through 2015. Separate Ensemble
continue to be reported in many regions, including models were run for each sex and for two levels
Africa, South Asia, and the Pacific Islands.7,11-13 of data availability. Country-level covariates asso-
There is increasing interest in the burden of ciated with rheumatic heart disease were included
rheumatic heart disease, driven in part by the to inform the models. These covariates were the
availability of echocardiography-based screening proportion of the population under 30 years of
in areas in which the condition is endemic and age, years of education per capita, income per
a growing need to meet benchmarks in cardio- capita, the proportion of children under 5 years
vascular health.14,15 The WHO and World Heart of age with low body weight for age (i.e., >2
Federation have called for a 25% reduction in standard deviations below the WHO standard
mortality due to cardiovascular causes, including weight-for-age curve), access to health care (a
rheumatic heart disease, by the year 2025.16,17 As summary variable based on principal-component
part of the 2015 Global Burden of Disease study analysis of several health services indicators), the
(GBD 2015), we estimated the global, regional, proportion of the population with access to im-
and national burden of rheumatic heart disease proved water sources (as defined by the WHO–
for the years 1990 through 2015. UNICEF Joint Monitoring Program for Water
Supply and Sanitation [JMP]), the proportion of
the population with access to improved sanita-
Me thods
tion (as defined by the JMP), sociodemographic
Strategy for Estimating Mortality Due index (a summary indicator derived from mea-
to Rheumatic Heart Disease sures of income per capita, educational attain-
The overall objectives, methods, and organiza- ment, and fertility), and a summary exposure
tion of GBD 2015 have been reported previous- variable for rheumatic heart disease (a measure
ly.18-20 Methods relevant to rheumatic heart dis- of risk-weighted prevalence of exposure).
ease are described briefly in this section and in The results obtained with the ensemble models
detail in the Supplementary Appendix, available were then adjusted to account for secular trends
with the full text of this article at NEJM.org. in mortality due to human immunodeficiency

714 n engl j med 377;8 nejm.org August 24, 2017

The New England Journal of Medicine

Downloaded from nejm.org on October 29, 2023. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
 Burden of Rheumatic Heart Disease, 1990 –2015

virus–acquired immunodeficiency syndrome (HIV– of disease on the basis of studies that showed a
AIDS), which biases death estimates in countries high prevalence of childhood rheumatic heart
with a high HIV–AIDS burden. Finally, the disease in these countries. Estimates of preva-
model results were adjusted by scaling them lence on a global level were based on a combina-
within the fraction of deaths due to all cardio- tion of the endemic and nonendemic models,
vascular diseases and all deaths. Age-standard- under the assumption that very few cases of
ized mortality was calculated with the use of the asymptomatic rheumatic heart disease exist among
direct method and a 2015 world reference popu- young people in countries with a nonendemic
lation based on United Nations Population Divi- pattern.
sion data updated for GBD 2015. Years of life We separately modeled the prevalence of symp-
lost were calculated by multiplying the number tomatic heart failure due to rheumatic heart dis-
of deaths due to rheumatic heart disease in each ease. We estimated the prevalence of heart failure
age group by the global standard remaining life due to any cause for each location, age, sex, and
expectancy at the mean age at death for persons year, then assigned cases to 20 specific causes,
who die in each age group.18 relying on published and administrative data on
the causes of heart failure and rates of mortality
Strategy for Estimating the Prevalence due to these causes. Heart failure was estimated
of Rheumatic Heart Disease as mild, moderate, or severe with the use of
We performed a systematic literature review for Medical Expenditure Panel Survey data on patient-
data on rheumatic heart disease incidence, prev- reported quality of life among persons with heart
alence, and case fatality rate. Data were identi- failure.19
fied primarily from community-based cross- All data were analyzed with the use of a
sectional and cohort studies and nationally Bayesian mixed-effects meta-regression tool (des-
representative hospital administrative data sets. ignated DisMod-MR 2.1) that was developed for
Our case definition was rheumatic heart disease the GBD study.19 DisMod-MR 2.1 is a compart-
identified by a clinician, with or without echo- mental model that consists of three states —
cardiographic confirmation, that would require susceptible, diseased, and dead — with state
antibiotic prophylaxis or medical or surgical transitions determined by the rates of incidence,
treatment.21 We excluded studies that reported remission, excess mortality, and other-cause mor-
only the results of echocardiographic screening tality.23 Differential equations with appropriate
without clinical confirmation or expert interpre- boundary conditions ensure consistency among
tation. We did not use estimates of rates of all disease parameters in the model. The tool
“borderline” rheumatic heart disease (i.e., minor uses an offset log-normal model with fixed ef-
abnormalities revealed by echocardiography that fects for study characteristics (i.e., design factors)
could represent normal variation in the structure that deviate from a predetermined reference and
of the aortic or mitral valve).22 for location-specific covariates (income and the
We considered countries to have one of two summary exposure variable).
patterns of rheumatic heart disease: endemic, To make predictions for all countries, esti-
with high mortality and prevalence among chil- mates were made in an analytical cascade from
dren, and nonendemic, with low mortality and the world to 7 super-regions, then to 21 world
prevalence among children and predominance at regions, and then to 195 countries and territo-
older ages, when the delayed sequelae of rheu- ries. This cascade took advantage of the assump-
matic heart disease occur (Fig. 1). Because of the tion that geographic proximity influences pat-
differences between these disease patterns, coun- terns of disease prevalence for rheumatic heart
tries with each pattern were modeled separately. disease. Information from higher levels in the
Endemicity was defined on the basis of esti- cascade were used as prior distributions at the
mates of mortality due to rheumatic heart dis- next level. Uncertainty intervals were taken as
ease from GBD 2015, with a threshold of 0.15 the 2.5th and 97.5th percentiles of the posterior
deaths per 100,000 population among children distribution.
5 to 9 years of age in 2015. After expert review Years lived with disability were estimated by
of country assignments, Kenya and Nicaragua multiplying the number of cases by disability
were reclassified as having an endemic pattern weights developed for the GBD studies.19 For

n engl j med 377;8 nejm.org August 24, 2017 715

The New England Journal of Medicine
Downloaded from nejm.org on October 29, 2023. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Endemic
Nonendemic

ATG VCT BRB COM MHL KIR

W. Africa E. Med.

SLB FSM
DMA GRD MDV MUS
MLT

VUT WSM
Caribbean LCA TTO TLS SYC Balkan Peninsula
Persian SGP
Gulf
FJI TON

Figure 1. Classification of Countries as Having an Endemic or Nonendemic Pattern of Rheumatic Heart Disease.
A country was classified as having an endemic pattern of disease if its estimated childhood mortality due to rheu-
matic heart disease was greater than 0.15 deaths per 100,000 population among children 5 to 9 years of age. ATG
denotes Antigua and Barbuda, BRB Barbados, COM Comoros, DMA Dominica, E. Med. eastern Mediterranean
­region, FJI Fiji, FSM Federated States of Micronesia, GRD Grenada, KIR Kiribati, LCA Saint Lucia, MDV Maldives,
MHL Marshall Islands, MLT Malta, MUS Mauritius, SGP Singapore, SLB Solomon Islands, SYC Seychelles, TLS
Timor-Leste, TON Tonga, TTO Trinidad and Tobago, VCT Saint Vincent and the Grenadines, VUT Vanuatu, W. Africa
West Africa, and WSM Samoa.

asymptomatic rheumatic heart disease, we used nonfatal cases were available from most coun-
a disability weight that represented a healthy tries. For sub-Saharan Africa, data were available
person with the need for long-term medication from only 14 countries. We also relied on country-
use (prophylactic antibiotic therapy). For heart specific covariates from all countries and geo-
failure, we used disability weights representing spatial modeling, as described above, to develop
New York Heart Association class II, III, or IV estimates of prevalence and mortality for coun-
symptoms. Years of life lost and years lived with tries without data on rheumatic heart disease.
disability were summed to obtain the number of
disability-adjusted life-years due to rheumatic R e sult s
heart disease.18
Mortality Due to Rheumatic Heart Disease
Data Availability Figure 3 shows the raw numbers of global deaths
The availability of data on fatal and nonfatal that were coded to rheumatic heart disease and
cases of rheumatic heart disease varied widely to indeterminate or intermediate cause-of-death
across countries and regions. Figure 2 shows the codes that were reassigned to rheumatic heart
types of data (on fatal cases, nonfatal cases, or disease, according to year. The increase in the
both) available according to country. Figure S2 in number of deaths in 2008 is due to the addition
the Supplementary Appendix shows the amount of data from the China Mortality Registration
of available data for both modeling processes and Reporting System.
according to region and year. Data on fatal or The cause-of-death codes that were most com-

716 n engl j med 377;8 nejm.org August 24, 2017

The New England Journal of Medicine

Downloaded from nejm.org on October 29, 2023. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
 Burden of Rheumatic Heart Disease, 1990 –2015

Data on fatal and nonfatal cases

Data on fatal cases only
Data on nonfatal cases only
No data

ATG VCT BRB COM MHL KIR

W. Africa E. Med.

SLB FSM
DMA GRD MDV MUS
MLT

VUT WSM
Caribbean LCA TTO TLS SYC Balkan Peninsula
Persian SGP
Gulf
FJI TON

Figure 2. Classification of Countries According to the Availability of Data on Fatal and Nonfatal Cases of Rheumatic
Heart Disease.

monly reassigned to rheumatic heart disease were Patterns of mortality due to rheumatic heart
left heart failure and right heart failure, which disease varied significantly according to world
accounted for 25.5% and 5.3%, respectively, of region in 2015. The largest number of deaths
deaths from rheumatic heart disease after reas- occurred in East Asia and South Asia. The highest
signments had been made. Detailed results of the age-standardized death rates occurred in Oceania,
sensitivity analyses performed to assess the un- South Asia, and central sub-Saharan Africa, the
certainty in reassignment of deaths due to left only regions where the 95% uncertainty intervals
heart failure are provided in the Supplementary in 1990 and 2015 overlap (Fig. 4A).
Appendix. In 2015, the countries with the highest esti-
On the basis of results derived from the en- mated numbers of deaths due to rheumatic heart
semble models, we estimated that there were disease were India (119,100 deaths), China
347,500 deaths (95% uncertainty interval, 328,300 (72,600), and Pakistan (18,900). The highest es-
to 367,100) from rheumatic heart disease in 1990 timated age-standardized death rates — more
and 319,400 deaths (95% uncertainty interval, than 10 deaths per 100,000 population — were in
297,300 to 337,300) in 2015, a decrease of 8.1% the Solomon Islands, Pakistan, Papua New
(95% uncertainty interval, 2.7 to 13.5). Global Guinea, Kiribati, Vanuatu, Fiji, India, Federated
age-standardized mortality from rheumatic heart States of Micronesia, Marshall Islands, Central
disease decreased from 9.2 deaths per 100,000 African Republic, and Lesotho.
population (95% uncertainty interval, 8.7 to 9.7)
in 1990 to 4.8 deaths per 100,000 population Prevalence of Rheumatic Heart Disease
(95% uncertainty interval, 4.4 to 5.1) in 2015, a We estimated that in 2015 a total of 33,194,900
decrease of 47.8% (95% uncertainty interval, cases (95% uncertainty interval, 29,466,400 to
44.7 to 50.9). An estimated 77% and 82% of the 42,905,600) of rheumatic heart disease occurred
deaths in 1990 and 2015, respectively, occurred in countries with an endemic pattern of disease
in locations with an endemic disease pattern. and 221,600 cases (95% uncertainty interval,

n engl j med 377;8 nejm.org August 24, 2017 717

The New England Journal of Medicine
Downloaded from nejm.org on October 29, 2023. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Intermediate or indeterminate codes

reassigned to rheumatic heart disease
Codes assigned to rheumatic heart disease
150,000

100,000
No. of Deaths

50,000

0
90
91
92
93
94
95
96
97
98
99
00
01
02
03
04
05
06
07
08
09
10
11
12
13
14
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
Figure 3. Total Reported Deaths Assigned to Rheumatic Heart Disease and Intermediate or Nonspecific Causes
of Death Reassigned to Rheumatic Heart Disease, 1990–2014.
Rheumatic heart disease–specific deaths were identified from vital registration systems with the use of codes from
the International Classification of Diseases, 9th Revision (ICD-9) and 10th Revision (ICD-10) (Table S1 in the Supple-
mentary Appendix). Deaths attributed to ill-defined or nonspecific causes (e.g., “heart disease, unspecified” [ICD-10
code I51.9]) or intermediate causes (e.g., “heart failure” [ICD-10 code I50]) were reassigned to accepted causes of
death, including rheumatic heart disease, with the use of algorithms developed for the Global Burden of Disease study
for 2015. The increase in the number of deaths in 2008 is due to the inclusion of the China Mortality Registration
and Reporting System starting in 2008. The decrease in intermediate or indeterminate coded deaths in 2014 is due
to a delay in the receipt of data from vital registration data systems that had higher proportions of indeterminate or
intermediate death codes.

205,800 to 238,300) occurred in countries with a Republic of the Congo (805,000), together ac-
nonendemic pattern. The estimated age-standard- counting for 73% of global cases. Twenty coun-
ized prevalence of rheumatic heart disease in tries with an endemic pattern of disease had an
2015 was 444 cases per 100,000 population for age-standardized prevalence exceeding 1%.
countries with an endemic pattern and 3.4 cases
per 100,000 population for countries with a non- Number of Cases of Heart Failure among
endemic pattern. Between 1990 and 2015, the Cases of Rheumatic Heart Disease
age-standardized prevalence declined significant- We estimated that there were 156,900 cases
ly in several regions (Fig. 4B). In 2015, the age- (95% uncertainty interval, 103,400 to 212,500) of
standardized prevalence remained highest in mild heart failure, 129,500 cases (95% uncer-
Oceania, followed by central sub-Saharan Africa tainty interval, 93,700 to 170,300) of moderate
and South Asia. In 2015, the countries with the heart failure, and 352,400 cases (95% uncertainty
largest estimated numbers of cases of rheumatic interval, 302,300 to 405,300) of severe heart
heart disease were India (13.17 million cases), failure due to rheumatic heart disease in 1990.
China (7.07 million), Pakistan (2.25 million), For 2015, our estimates were 295,300 cases (95%
Indonesia (1.18 million), and the Democratic uncertainty interval, 194,100 to 401,400) of mild

718 n engl j med 377;8 nejm.org August 24, 2017

The New England Journal of Medicine

Downloaded from nejm.org on October 29, 2023. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
 Burden of Rheumatic Heart Disease, 1990 –2015

Figure 4. Age-Standardized Mortality Due to and Prevalence A Mortality

of Rheumatic Heart Disease According to World Region 1990 2015
in 1990 and 2015.
I bars represent 95% uncertainty intervals.

Age-Standardized Mortality (per 100,000)

heart failure, 243,700 cases (95% uncertainty
interval, 176,600 to 320,900) of moderate heart 20

failure, and 663,000 cases (95% uncertainty in-

terval, 566,800 to 763,900) of severe heart fail-
ure, which represents an 88% increase in the
number of cases overall.

Summary Measures of Health 10

The number of disability-adjusted life-years due
to rheumatic heart disease in 2015 was
10,513,200 (95% uncertainty interval, 9,611,000
to 11,514,500), accounting for 0.43% of global
disability-adjusted life-years due to any cause.
The global rate of disability-adjusted life-years
0
due to rheumatic heart disease in 2015 was
as sia
Ce Oce sia
h- E ntr ral nia
co te E ia
As u e
ig u W A -Pa pe
in er st tr ific
e tin u ia
th e e
Ce an C Am ica
or r ra ti ib ca
Af ica at Am an
an in e a
M m a
ut rn b-S So le ica
es n -S ar A t
rn b- a A a
b- ha A a
ha n A ica
n ica
a
W her sub ah uth Eas
e E p

or m p

a at Am ric
d A ric

te su ah an si
su Sa ran fric

ric
in as al As

m La n E as

nt La ar eri
m rn uro
ia ro

N A ro

th op l L n be
he A
tA

id er

Sa ra fr
ra fr
h- th e us c
t a
142.6 per 100,000 population (95% uncertainty

ric l L in e

Af
co n er al
ut ast

d
So E

interval, 130.4 to 156.2). The highest age-stan- C e n

dardized rates were found in Oceania, South

So ste su
de
An

l
Ea ra
Asia, and Africa (Fig. 5). Most disability-adjusted
H So

T
ig

nt
H

Ce
life-years due to rheumatic heart disease were

N
the result of years of life lost (84.9%), which in- B Prevalence
dicated that premature death was a larger driver 1990 2015
of total health loss from rheumatic heart disease
than was years of life lived with disability.
Age-Standardized Prevalence (per 100,000)

2000
Discussion
We used multiple sources of data and epidemio-
logic modeling techniques to estimate the global 1500
prevalence of and mortality due to rheumatic
heart disease over a 25-year period. Over this
interval, the health-related burden of rheumatic
heart disease declined in most countries, but the 1000
condition persisted in some of the poorest re-
gions in the world. We estimate that 10 persons
per 1000 population living in South Asia and
500
central sub-Saharan Africa and 15 persons per
1000 population in Oceania were living with
rheumatic heart disease in the year 2015.
Rheumatic heart disease is a consequence of 0
untreated streptococcal pharyngitis, and its ma-
as sia
Ce Oce sia
h- E ntr ral nia
co te E ia
As u e
ig u W A -Pa pe
in er st tr ific
e tin u ia
th e e
Ce an C Am ica
or r ra ti ib ca
Af ica at Am an
an in e a
M m a
ut rn b-S So le ica
es n -S ar A t
rn b- a A a
b- ha A a
ha n A ica
n ica
a
W her sub ah uth Eas
e E p

or m p

a at Am ric
d A ric

te su ah an si
su Sa ran fric

ric
in as al As

m La n E as

nt La ar eri
m rn uro
ia ro

N A ro

th op l L n be
he A
tA

id er

Sa ra fr
ra fr
h- th e us c
t a

ric l L in e

Af
co n er al
ut ast

jor antecedents are the factors that influence the

d
So E

transmission of this infection, including access

C e

So ste su
de

to high-quality health care and social determi-

An

l
Ea ra
H So

T
ig

nt

nants of health.2,3 At the national level, progress

H

Ce
N

— or lack thereof — in addressing social deter-

n engl j med 377;8 nejm.org August 24, 2017 719

The New England Journal of Medicine
Downloaded from nejm.org on October 29, 2023. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

0–49 per 100,000

50–99 per 100,000
100–149 per 100,000
150–199 per 100,000
≥200 per 100,000

ATG VCT BRB COM MHL KIR

W. Africa E. Med.

SLB FSM
DMA GRD MDV MUS
MLT

VUT WSM
Caribbean LCA TTO TLS SYC Balkan Peninsula
Persian SGP
Gulf
FJI TON

Figure 5. Age-Standardized Disability-Adjusted Life-Years Due to Rheumatic Heart Disease per 100,000 Population, 2015.

minants such as education and income has were actually the result of underlying rheumatic
tracked closely with mortality due to rheumatic heart disease.20
heart disease.24 Our estimates of disease prevalence are simi-
In addition to impeding the effective preven- lar to those in a recent meta-analysis of screen-
tion of acute rheumatic fever, social and eco- ing studies in which the overall prevalence of
nomic factors may also make the management rheumatic heart disease in low-income and
of chronic rheumatic heart disease more diffi- middle-income countries was shown to range
cult. Lifelong treatment options for rheumatic from 2.7 cases per 1000 population (for “clini-
heart disease, although effective, place large de- cally manifest” disease) to 21.1 cases per 1000
mands on health systems.25 Major shortfalls in population (for “clinically silent” disease).7 Among
medical and surgical care for rheumatic heart subclinical cases of rheumatic heart disease that
disease have been documented in countries where are detected through echocardiographic screen-
the condition is endemic, even at tertiary centers.26 ing (termed “borderline” rheumatic heart dis-
We adjusted our mortality input data by reas- ease), some may progress to definite rheumatic
signing codes for intermediate or indeterminate heart disease, whereas others may regress. To
causes of death, including heart failure, and this date, only a few small prospective studies have
adjustment substantially increased the estimates evaluated the progression of borderline disease.27,28
of the number of deaths due to rheumatic heart Our prevalence estimates would have been higher
disease. Advances in methods for handling cause- if we had included borderline cases in our
of-death codes are an important component of model; however, current data do not support this
improved estimates of mortality due to rheumatic approach, because it is unclear how this condi-
heart disease. At the same time, it is likely that tion should be managed clinically.29
some deaths from stroke and endocarditis are It is possible that our estimates for some loca-
miscoded, so we cannot estimate how many of tions were biased upward by the use of studies
the 6.3 million cases of stroke and 85,000 deaths of prevalence that were conducted in subna-
from endocarditis that were estimated for 2015 tional areas with an endemic pattern of disease.

720 n engl j med 377;8 nejm.org August 24, 2017

The New England Journal of Medicine

Downloaded from nejm.org on October 29, 2023. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
 Burden of Rheumatic Heart Disease, 1990 –2015

Yet most of these studies focus on schoolchildren, tion, prevalence among adults in low-income and
among whom rheumatic heart disease might be middle-income countries, and rates of nonfatal
less common than in the total population.30 To outcomes and excess mortality in longitudinal
clarify these issues, future prevalence studies studies involving persons with rheumatic heart
should sample more broadly and screen persons disease. Improvements in the measurement of
beyond school-aged children. It is also possible the burden of rheumatic heart disease will assist
that some middle-income countries (e.g., in Latin in planning for its control and will help identify
America, where our estimates are comparatively countries where further investments are needed.
low) will have subpopulations that differ from In summary, we estimated the global disease
the national average in their patterns of disease prevalence of and mortality due to rheumatic
(i.e., endemic vs. nonendemic).31 Future work on heart disease over a 25-year period. The health-
disease burden at the state or provincial level will related burden of rheumatic heart disease has
be required to address this discrepancy. Finally, declined worldwide, but the condition persists in
our analysis was limited to English-language some of the poorest regions in the world.
studies. The views expressed in this document are those of the authors
Better data for low-income and middle-income and do not necessarily represent the views of the National Heart,
Lung, and Blood Institute, National Institutes of Health, or the
countries are needed to guide policies for the Department of Health and Human Services.
control of rheumatic heart disease. In our analy- Supported by a grant from the Bill and Melinda Gates Foun-
sis, we used epidemiologic modeling techniques dation and by the Medtronic Foundation (to Dr. Watkins).
Disclosure forms provided by the authors are available with
to provide estimates for countries for which data the full text of this article at NEJM.org.
were insufficient. However, further improvements We thank Tahiya Alam, Christine Pinho, Megha Arora, Michael
in estimates of the burden of rheumatic heart Kutz, Katya Shackelford, Caitlyn Steiner, Martin Pletcher, Ben
Zipkin, and Minh Nguyen at the Institute for Health Metrics and
disease will require new research in three areas: Evaluation, University of Washington, Seattle, for assistance in
the extent of misclassification in death certifica- producing earlier versions of the tables and figures.

Appendix
The authors’ affiliations are as follows: the Division of General Internal Medicine, Department of Medicine (D.A.W.), the Institute for
Health Metrics and Evaluation, Department of Global Health (C.O.J., M.H.F., M.N., A.H.M., C.J.L.M., T.V., G.A.R.), and the Division
of Cardiology, Department of Medicine (G.A.R.), University of Washington, Seattle; the Department of Medicine, Groote Schuur Hos-
pital and University of Cape Town, Cape Town, South Africa (D.A.W., B.M.M.); the Murdoch Children’s Research Institute and the
Centre for International Child Health, University of Melbourne, Melbourne, VIC (S.M.C., A.C.S.), and Telethon Kids Institute, Univer-
sity of Western Australia and Princess Margaret Hospital for Children, Perth, WA (J.R.C.) — both in Australia; the Department of
Cardiology, All India Institute of Medical Sciences, New Delhi (G.K.); Children’s National Health System, Washington, DC (A.B.,
C.A.S.); Program in Global NCDs and Social Change, Department of Global Health and Social Medicine, Harvard Medical School, and
the Cardiovascular Division, Department of Medicine, Brigham and Women’s Hospital — both in Boston (G.B.); the Division of Cardi-
ology, Department of Medicine, Case Western Reserve University and University Hospitals Cleveland Medical Center, Cleveland (C.T.L.);
the Center for Translation Research and Implementation Science and Division of Cardiovascular Sciences, National Heart, Lung, and
Blood Institute, National Institutes of Health, Bethesda, MD (G.A.M.); and the School of Medicine and Telehealth Center, Universidade
Federal de Minas Gerais, Belo Horizonte, Brazil (B.R.N., A.L.P.R.).

References
1. Marijon E, Mirabel M, Celermajer DS, decrease in morbidity and mortality from 9. Bach JF, Chalons S, Forier E, et al. 10-
Jouven X. Rheumatic heart disease. Lancet rheumatic fever in the United States. N Engl Year educational programme aimed at rheu-
2012;​379:​953-64. J Med 1988;​318:​280-6. matic fever in two French Caribbean is-
2. Meira ZM, Goulart EM, Colosimo EA, 6. Doukky R, Abusin SA, Bayissa YA, lands. Lancet 1996;​347:​644-8.
Mota CC. Long term follow up of rheu- Kelly RF, Ansari AH. Rheumatic heart dis- 10. Nordet P, Lopez R, Dueñas A,
matic fever and predictors of severe rheu- ease in modern urban America: a cohort Sarmiento L. Prevention and control of
matic valvar disease in Brazilian children study of immigrant and indigenous pa- rheumatic fever and rheumatic heart dis-
and adolescents. Heart 2005;​91:​1019-22. tients in Chicago. Int J Cardiol 2014;​175:​ ease: the Cuban experience (1986-1996-
3. Longo-Mbenza B, Bayekula M, Ngiyulu 178-80. 2002). Cardiovasc J Afr 2008;​19:​135-40.
R, et al. Survey of rheumatic heart disease 7. Rothenbühler M, O’Sullivan CJ, Stor­ 11. Günther G, Asmera J, Parry E. Death
in school children of Kinshasa town. Int J tecky S, et al. Active surveillance for rheu- from rheumatic heart disease in rural
Cardiol 1998;​63:​287-94. matic heart disease in endemic regions: Ethiopia. Lancet 2006;​367:​391.
4. Gordis L. The virtual disappearance a systematic review and meta-analysis of 12. Colquhoun SM, Condon JR, Steer AC,
of rheumatic fever in the United States: prevalence among children and adolescents. Li SQ, Guthridge S, Carapetis JR. Dispar-
lessons in the rise and fall of disease — Lancet Glob Health 2014;​2(12):​e717-e726. ity in mortality from rheumatic heart dis-
T. Duckett Jones Memorial Lecture. Cir- 8. Rheumatic diseases:​first report of ease in indigenous Australians. J Am Heart
culation 1985;​72:​1155-62. the Expert Committee. WHO technical re- Assoc 2015;​4(7):​e001282.
5. Massell BF, Chute CG, Walker AM, port series no. 78. Geneva:​World Health 13. Parks T, Kado J, Miller AE, et al.
Kurland GS. Penicillin and the marked Organization, 1954. Rheumatic heart disease-attributable mor-

n engl j med 377;8 nejm.org August 24, 2017 721

The New England Journal of Medicine
Downloaded from nejm.org on October 29, 2023. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
 Burden of Rheumatic Heart Disease, 1990 –2015

tality at ages 5-69 years in Fiji: a five-year, 310 acute and chronic diseases and inju- 26. Zühlke L, Engel ME, Karthikeyan G,
national, population-based record-linkage ries, 1990-2015: a systematic analysis for et al. Characteristics, complications, and
cohort study. PLoS Negl Trop Dis 2015;​ the Global Burden of Disease Study 2015. gaps in evidence-based interventions in
9(9):​e0004033. Lancet 2016;​388:​1545-602. rheumatic heart disease: the Global Rheu-
14. Maurice J. Rheumatic heart disease back 20. GBD 2015 Mortality and Causes of matic Heart Disease Registry (the REMEDY
in the limelight. Lancet 2013;​382:​1085-6. Death Collaborators. Global, regional, and study). Eur Heart J 2015;​36:​1115-22a.
15. Marijon E, Ou P, Celermajer DS, et al. national life expectancy, all-cause and 27. Beaton A, Okello E, Aliku T, et al. Latent
Prevalence of rheumatic heart disease cause-specific mortality for 249 causes of rheumatic heart disease: outcomes 2 years
detected by echocardiographic screening. death, 1980-2015: a systematic analysis for after echocardiographic detection. Pediatr
N Engl J Med 2007;​357:​470-6. the Global Burden of Disease Study 2015. Cardiol 2014;​35:​1259-67.
16. Global action plan for the prevention Lancet 2016;​388:​1459-544. 28. Zühlke L, Engel ME, Lemmer CE, et al.
and control of noncommunicable diseases 21. Rheumatic fever and rheumatic heart The natural history of latent rheumatic
2013-2020. Geneva:​World Health Organi- disease. Technical report series no. 923. heart disease in a 5 year follow-up study:
zation, 2013. Geneva:​World Health Organization, 2004. a prospective observational study. BMC
17. Remenyi B, Carapetis J, Wyber R, Tau- 22. Reményi B, Wilson N, Steer A, et al. Cardiovasc Disord 2016;​16:​46.
bert K, Mayosi BM. Position statement of World Heart Federation criteria for echo- 29. Zühlke L, Mayosi BM. Echocardio-
the World Heart Federation on the preven- cardiographic diagnosis of rheumatic graphic screening for subclinical rheu-
tion and control of rheumatic heart dis- heart disease — an evidence-based guide- matic heart disease remains a research
ease. Nat Rev Cardiol 2013;​10:​284-92. line. Nat Rev Cardiol 2012;​9:​297-309. tool pending studies of impact on prog-
18. GBD 2015 DALYs and HALE Collab­ 23. Barendregt JJ, Van Oortmarssen GJ, nosis. Curr Cardiol Rep 2013;​15:​343.
orators. Global, regional, and national Vos T, Murray CJ. A generic model for the 30. Roberts K, Maguire G, Brown A, et al.
disability-adjusted life years (DALYs) for assessment of disease epidemiology: the Echocardiographic screening for rheumat-
315 diseases and injuries and healthy life computational basis of DisMod II. Popul ic heart disease in high and low risk Aus-
expectancy (HALE) for 195 countries and Health Metr 2003;​1:​4. tralian children. Circulation 2014;​ 129:​
territories, 1990-2015: a systematic analy- 24. Social determinants of health visual- 1953-61.
sis for the Global Burden of Diseases, In- ization. Seattle:​Institute for Health Met- 31. Nascimento BR, Beaton AZ, Nunes
juries, and Risk Factors (GBD) 2015 Study. rics and Evaluation, 2015 (http://vizhub​ MC, et al. Echocardiographic prevalence
Lancet 2016;​388:​1603-58. .healthdata​.org/​sdh). of rheumatic heart disease in Brazilian
19. GBD 2015 Disease and Injury Inci- 25. Wyber R, Zühlke L, Carapetis J. The schoolchildren: data from the PROVAR
dence and Prevalence Collaborators. Glob- case for global investment in rheumatic study. Int J Cardiol 2016;​219:​439-45.
al, regional, and national incidence, prev- heart-disease control. Bull World Health Copyright © 2017 Massachusetts Medical Society.
alence, and years lived with disability for Organ 2014;​92:​768-70.

ARTICLE METRICS NOW AVAILABLE

Visit the article page at NEJM.org and click on the Metrics tab to view
comprehensive and cumulative article metrics compiled from multiple sources,
including Altmetrics. Learn more at www.nejm.org/page/article-metrics-faq.

722 n engl j med 377;8 nejm.org August 24, 2017

The New England Journal of Medicine

Downloaded from nejm.org on October 29, 2023. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.