Original Article

Primary Prevention of Acute Rheumatic Fever and

Rheumatic Heart Disease With Penicillin in South African
Children With Pharyngitis
A Cost-Effectiveness Analysis
James Irlam, MPhil; Bongani M. Mayosi, DPhil; Mark Engel, MPH; Thomas A. Gaziano, MD, MSc

BackgroundAcute rheumatic fever and subsequent rheumatic heart disease remain significant in developing countries.
We describe a cost-effective analysis of 7 strategies for the primary prevention of acute rheumatic fever and rheumatic
heart disease in children presenting with pharyngitis in urban primary care clinics in South Africa.
Methods and ResultsWe used a Markov model to assess the cost-effectiveness of treatment with intramuscular penicillin
using each of the following strategies: (1) empirical (treat all); (2) positive throat culture (culture all); (3) clinical decision
rule (CDR) score 2 (CDR 2+); (4) CDR score 3 (CDR 3+); (5) treating those with a CDR score 2 plus those with
Downloaded from http://circoutcomes.ahajournals.org/ by guest on January 8, 2017

CDR score <2 and positive cultures (CDR 2+, culture CDR negatives); (6) treating those with a CDR score 3 plus
those with CDR score <3 and positive cultures (CDR 3+, culture CDR negatives); and (7) treat none. The strategies
ranked in order from lowest cost were treat all ($11.19 per child), CDR 2+ ($11.20); the CDR 3+ ($13.00); CDR 2+,
culture CDR negatives ($16.42); CDR 3+, culture CDR negatives ($23.89); and culture all ($27.21). The CDR 2+ is the
preferred strategy at less than $150/quality-adjusted life year compared with the treat all strategy. A strategy of culturing
all children compared with the CDR 2+ strategy costs more than $125000/quality-adjusted life year gained.
ConclusionsTreating all children presenting with pharyngitis in urban primary care clinics in South Africa with intramuscular
penicillin is the least costly. A strategy of using a clinical decision rule without culturing is overall the preferred strategy.
A strategy of culturing all children may be prohibitively expensive.(Circ Cardiovasc Qual Outcomes. 2013;6:343-351.)
Key Words:cost-benefit analysis prevention rheumatic heart disease

A cute rheumatic fever (ARF) remains the most important

cause of acquired heart disease in children and young
adults in the world.1 ARF itself is rarely fatal, yet up to 16
years of life and 3 years of quality-adjusted life attributable to
disability are lost per case of rheumatic heart disease (RHD).2
For this reason, the World Heart Federation has made the
elimination of ARF and control of RHD 1 of the 6 main goals
in its strategic plan through 2015.
In developed countries, the incidence of ARF during the past
50 years declined from several hundred to <10 per 100000
per year. However, new evidence suggests that outbreaks can
still occur even in a well-developed country area, such as in
Italy,3 and former gains can be lost when resources are with-
drawn. This occurred with the fall of the former Soviet Union,
where in several Central Asian republics, the ARF incidence
rates increased from 80 per 100000 in the 1980s to >500 per
100000 in 2007.4
In the absence of a vaccine against group A streptococ-
cus (GAS) infection, primary prevention of ARF and RHD
depends on preventing the initial attacks of ARF by means
of short-term oral or intramuscular (IM) penicillin treatment
of patients presenting with acute sore throat (pharyngitis)
caused by GAS infection. Yet primary prevention has been
less widely adopted in developing countries. This is because
of both barriers to its implementation and a concern about its
cost-effectiveness.5
Barriers to primary prevention in developing countries include
poor access to primary care, a shortage of skilled personnel, the
expense of microbiological diagnosis, poor public awareness
about the diagnosis and prompt treatment of suspected GAS
pharyngitis, and a high incidence of ARF without sore throat.
Yetthere is no consensus on the role of primary prevention of
ARF as a public health strategy. School-based programs or
mass screenings have not been recommended because of their
significant costs, mainly because of high staffing and lab diag
nostic costs required for active screening in such large programs.6
What has not been assessed is what one should do in the
setting of symptomatic presentation of sore throat to the clinic

Received December 6, 2012; accepted March 14, 2013.

From the Primary Health Care Directorate, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa (J.I.); Department of
Medicine, Groote Schuur Hospital and University of Cape Town, Cape Town, South Africa (B.M.M., M.E.); and Cardiovascular Division, Brigham and
Womens Hospital, Boston, MA (T.A.G.).
Correspondence to Thomas A. Gaziano, MD, MSc, Division of Cardiovascular Medicine, Brigham and Womens Hospital, 75 Francis St, Boston, MA
02115. E-mail tgaziano@partners.org
2013 American Heart Association, Inc.
Circ Cardiovasc Qual Outcomes is available at http://circoutcomes.ahajournals.org DOI: 10.1161/CIRCOUTCOMES.111.000032

343
 344 Circ Cardiovasc Qual Outcomes May 2013
WHAT IS KNOWN
Acute rheumatic fever remains the most impor-
tant cause of acquired heart disease in children and
young adults in the world.
Secondary prevention of rheumatic heart disease has
been shown to be cost-effective. However, contro-
versy still exists on primary prevention.
WHAT THE STUDY ADDS
Using a Markov model to assess different treatment
strategies, this study concluded that treatment with
penicillin was cost-effective in all pharyngitis cases.
The results also suggest that use of a clinical deci-
sion rule aids in the cost-effective management of
pharyngitis, but use of throat cultures in all cases
Downloaded from http://circoutcomes.ahajournals.org/ by guest on January 8, 2017
of pharyngitis was not cost-effective in low- and
middle-income countries.
in a high endemic area with relatively low labor costs and low
screening costs because of the opportunistic nature of presen-
tation. In the South African setting, a marginal cost of an IM
penicillin injection of R300 ($46) per case of ARF prevented
has been estimated, suggesting that it may be cost-effective.5,7
However, diagnostic strategies are not 100% sensitive or spe-
cific and can be costly or unavailable in resource-limited set-
tings. In addition, penicillin reactions can be both mild and
severe, even though they are rare. Therefore, the challenge is
to find a relatively sensitive strategy that ascertains the great-
est proportion of GAS infections, limits costly testing, and
minimizes unwarranted antibiotic treatment.
We describe a cost-effective analysis of 7 diagnostic and
treatment strategies for the primary prevention of ARF in chil-
dren presenting with pharyngitis in urban primary care set-
tings in South Africa.
Methods
Decision Analysis Model
The decision of whom to treat with pharyngitis is complex. Providers
must weigh the health benefits of preventing ARF and subsequent
RHD, which has considerable morbidity and mortality, against the
small risks of potentially significant drug reactions. Furthermore, the
costs of diagnosis and initial management of GAS pharyngitis with
IM penicillin must be weighed against the costs of later treatment for
ARF and RHD. This set of complex choices with both clinical and di-
agnostic uncertainty is best facilitated by a medical decision analysis.
We therefore developed a Markov decision analysis cohort model to
evaluate common strategies for managing children (315 years) pre-
senting with an acute sore throat (pharyngitis) and no history of ARF
in an urban primary care clinic setting in South Africa (Figure1).
The model incorporates annual recurrent rates of pharyngitis in this
pediatric population and thus does not reflect the risk of RHD from 1
episode of pharyngitis but the possibility of multiple infections. The
model follows the cohort through the annual probability of having
GAS, developing ARF or peritonsillar abscess (PTA) acutely, devel-
oping complications from penicillin treatment, developing RHD on
an annual basis, developing congestive heart failure requiring valve
surgery, and ultimately premature death from RHD or other causes.
Furthermore, the model incorporates the sensitivity and specificity
of the various screening tools for GAS infection. The probability of
developing ARF or complications from treatment and incurring costs
are affected by a patients treatment status, which in turn is affected
by the likelihood of treatment based on the diagnostic strategies de-
scribed below.
Strategies Compared
We compared 7 strategies for the evaluation and management of
children between the ages of 3 to 15 years who present to an urban
primary care clinic with a complaint of sore throat and no history of
ARF. Given the resource constraints in many parts of South Africa,
the following 7 strategies were compared in our model. The strate-
gies regarding screening and initiating treatment for possible GAS
are based on a clinical decision rule, throat culture results, or a com-
bination of the 2 in multistage screening strategies. We also compared
these screening strategies to a treat all strategy with no screening and
compared all strategies to no treatment. Several symptom-based
clinical decision rules (CDRs) of varying sensitivity and specificity
have been developed and evaluated to assist in diagnosis in resource-
limited settings, which typically have limited access to lab culture and
Figure1. Schematic of 7-strategy decision
analysis model. ARF indicates acute
rheumatic fever; CDR, clinical decision
rule; GAS, group A streptococcus; PTA,
peritonsillar abscess; and RHD, rheumatic
heart disease.
 Irlam et al Primary Prevention of Acute Rheumatic Fever 345

Table 1. Components* and Performance of 3-Variable Data Sources

Clinical Decision Rule10
Cumulative Score of Components (Range 03) Sensitivity, % Specificity, %
1 99 3 pharyngitis or rheumatic fever or rheumatic heart disease and
2 92 38
3 38 83
*Clinical components of clinical decision rule: enlarged cervical nodes, no
rhinitis, no rash.
rapid testing and higher incident rates of ARF and RHD.8 The CDR
used in the model is a modification of the World Health Organization
(WHO)-recommended CDR9 that uses a 3-variable cumulative score
(ie, presence of enlarged anterior cervical lymph nodes, absence of
rash, and absence of rhinitis) to guide decisions about whether to
treat with antibiotics. This rule has been tested and validated in Egypt
in a population with health and socioeconomic status similar to that
in sub-Saharan Africa.10 In the Egyptian study, children presenting
with pharyngitis were assigned a point for each clinical variable for
Downloaded from http://circoutcomes.ahajournals.org/ by guest on January 8, 2017
The probabilities, costs, and utilities used in the decision model
were derived from the published literature and local sources where
available. The MEDLINE database was searched using the terms
cost-effectiveness. The references of relevant articles and evidence-
based clinical guidelines were examined for additional studies.
Epidemiology
Table2 lists the probabilities of GAS pharyngitis and its progres-
sion to RHD with and without treatment, as well as the associated
complications. Anywhere from 1.6% to 30% of children with sore
throat have GAS as the cause.2 The prevalence of GAS pharyngitis
was 15.3% (95% confidence interval, 11.2%20.4%) among 255
children presenting with sore throat at 3 urban primary care clinics in
Cape Town between June 2008 and June 2010 (unpublished data). We
have therefore used a baseline probability of 15% with a sensitivity
range of 1.6% to 30% to reflect the above estimates. The incidence of
pharyngitis in the same population is 870 per 100000 patient years.
Published estimates of the incidence of ARF in school-aged

a cumulative score with a range of 0 to 3 points. The sensitivity and
specificity of diagnosing GAS infection with 1, 2, or 3 of the vari-
ables are presented in Table1. We used scores 2 or 3 as part of
our strategies.
We included 5 single-stage strategies and two 2-stage screening
strategies: (1) empirical treatment with IM penicillin (treat all); (2)
treatment based on a positive throat culture (culture all); (3) treat-
ment based on a CDR score 2 (CDR 2+); (4) treatment based on
a CDR score 3 (CDR 3+); (5) treating those with a CDR score
2, culturing those with CDR scores <2 and then treating positive
cultures (CDR 2+, culture CDR negatives); (6) treating those with a
CDR score 3, culturing those with CDR scores <3 and treating pos-
itive cultures (CDR 3+, culture CDR negatives); and (7) observation
only (treat none). The 2-stage strategies were included to improve
the sensitivity of the screening process and to decrease the costs of
culturing. Rapid streptococcus antigen tests are not presently used
in government-supported primary care centers in South Africa and
thus were not evaluated. National guidelines on primary prevention
include both oral and IM penicillin but encourage IM penicillin be-
cause of problems with patient adherence to oral regimens and its
greater effectiveness in preventing ARF.11,12
children after untreated GAS pharyngitis range from 0.3% to 5% per
year depending on crowding conditions,2,1316 although some studies
have used much lower probabilities for nonendemic areas.20,30 We
have used a base case estimate of 0.3% and a wide range from 0.1%
to 5% to explore in sensitivity analyses, given the large uncertainty
in this variable. We have used an estimate of the annual risk of death
from ARF of 1% (range, 0.5%2%) based on prior studies.14,1719 The
PTA risk was estimated to be 1.5%.14,2022
On the basis of a review of epidemiological data, there is a life-
time risk of about 50% to 75% of developing RHD among those with
untreated ARF, with either mild, moderate, or severe cardiomegaly
(24%) or congestive heart failure leading to death.23,24 Therefore, as-
suming a constant hazard with an average life expectancy after infec-
tion, we estimated an annual probability of developing RHD of 1.8%
with a range from 1.4% to 2.7%. Mortality from RHD has been esti-
mated in our model as 1.5% per year, ranging from 1% to 2%, based
on an analysis by Carapetis et al.23 The annual probability of death
from natural causes in South Africa is derived from the age-specific
mortality rates in South Africa for the year 2006.31,32 We also adjust
the annual risk of developing RHD based on treatment with monthly
IM penicillin injections among those known to have ARF.

Table 2. Probability Variables*

Variable Name Base Case Range Sources
Epidemiology
GAS prevalence 0.15 0.0160.3 Michaud,2 Carapetis1
ARF given GAS+ 0.003 0.0010.05 Michaud,2 Stollerman,13 Tsevat,14 Denny,15 Robertson16
Risk of death from ARF 0.01 0.0050.02 Neuner,17 Tsevat,14 Feinstein,18 Hillner19
Risk of PTA 0.015 0.000.03 Van Howe,20 Tsevat,14 Lieu,21 Dippel22
Annual incidence of RHD, untreated ARF 0.018 0.0140.027 Carapetis,23 Michaud24
Annual risk of death from RHD 0.015 0.010.02 Carapetis1
Probability of patients with ARF receiving secondary prevention 0.12 0.000.90 RHD registry at GSH, Kumar,25 Grayson26
Annual incidence of pharyngitis 0.0087 0.00010.017 ARF registry in Vanguard community
Treatment effects
RR of IM penicillin vs ARF among GAS+ 0.2 0.110.5 Robertson27
RR of monthly penicillin vs RHD if ARF+ 0.45 0.080.78 Padmavati28
RR of IM penicillin vs PTA 0.15 0.050.47 del Mar29
Probability of penicillin-induced rash 0.015 0.0050.04 Neuner17
Incidence of penicillin-induced anaphylaxis 0.0001 00.0005 Tsevat14
Probability of death from anaphylaxis 0.10 0.0000.25 Neuner,17 Tsevat14
*ARF indicates acute rheumatic fever; GAS, group A streptococcus; IM, intramuscular; PTA, peritonsillar abscess; RHD, rheumatic heart disease; and RR, risk
reduction.
 346 Circ Cardiovasc Qual Outcomes May 2013

Table 3. Diagnostic Parameters* and subsequent ARF and RHD. The analysis includes health provider
costs related to the diagnosis and treatment of GAS pharyngitis with
Variable Base Case Range Sources IM penicillin, the treatment of any penicillin-induced rash and ana-
Sensitivity of CDR (score of 2+) 0.92 0.501.00 Steinhoff10 phylaxis or suppurative complications of GAS, and the ongoing inpa-
tient and outpatient treatment of rheumatic fever and RHD (Table4)
Specificity of CDR (score of 2+) 0.38 0.000.50 Steinhoff10 incurred by the health system and the patient. Costs are reported in
Sensitivity of CDR (score of 3+) 0.38 0.000.50 Steinhoff10 2010 $US, and discounting of future costs and outcomes at 3% per
year was applied, with a sensitivity analysis around a range of 0% to
Specificity of CDR (score of 3+) 0.83 0.501.00 Steinhoff10
10% to reflect local inflation rates.
Sensitivity of throat culture 0.90 0.901.00 Webb,30 Kellogg33 The provider costs of outpatient clinic visits and inpatient days at
Specificity of throat culture 0.95 0.901.00 Webb,30 Kellogg33 different levels of public hospital care in South Africa are derived
from a model of the estimated resource requirements of alternative
*CDR indicates clinical decision rule. healthcare financing reforms in South Africa,34 which used utilization
and recurrent expenditure data from the South African District Health
Treatment Effects Information System and the National Treasury. Costs of throat cul-
A 2005 fixed-effects meta-analysis of 9 trials found a protective effect of ture diagnosis and medical treatment were sourced from the National
IM penicillin against ARF of 80% (risk ratio, 0.20; 95% confidence in- Health Laboratory Service and the hospital pharmacy, respectively,
terval, 0.110.36),27 whereas a 2006 meta-analysis of 14 penicillin trials and for surgical procedures from the hospital case managers. Costs
found a 73% reduction in ARF (risk ratio, 0.27; 95% confidence interval, to patients for subsistence and transport to attend the health facili-
0.140.50).29 We use the risk ratio estimate of 0.2 with a range from ties are based on questionnaires that were administered to parents or
0.11 to 0.5 for IM penicillin. Penicillin effectiveness against PTA was guardians of children with a sore throat attending a primary care clinic
Downloaded from http://circoutcomes.ahajournals.org/ by guest on January 8, 2017

estimated to be 85%.29 The probabilities of allergic17 and anaphylactic14
reactions to penicillin were derived from previously published estimates.
Test Characteristics
Table3 displays the sensitivity and specificity of the diagnostic throat
cultures and CDRs for GAS pharyngitis. The CDR we used10 per-
formed the best (with a sensitivity of 92% and specificity of 38%
using a cutoff score of 2) in comparison with 6 other CDRs in pre-
dicting children with and without GAS in an Egyptian data set with
a GAS prevalence of 24.6% (n=410),8 under conditions most similar
to South African population. As expected, a CDR score of 3+ had
a lower sensitivity (38%) but much higher specificity (83%) than a
score of 2+. Throat culture sensitivity has been estimated to range
from 75% to 95% and specificity from 90% to 100%.30,33 We used
90% sensitivity and 95% specificity as a base case and conduct sensi-
tivity analyses around the ranges above.
Costs
The model uses a societal perspective to include all healthcare costs
related to the identification and management of those with sore throat
(n=49) and to adult patients with RHD (n=48) attending the cardiac
clinic at Groote Schuur Hospital in Cape Town, South Africa, in 2010.
The standard of care for treating adverse effects of penicillin, PTA,
rheumatic fever, and RHD is based on several sources, as detailed in
Table4. These sources include the National Guidelines on Primary
Prevention of Rheumatic Fever and RHD at the primary level,12
the South African Medicines Formulary,35 the Standard Treatment
Guidelines and Essential Drugs List,36 local hospital treatment pro-
tocols, and expert opinion. Patient data from Groote Schuur Hospital
were used in some of the cost calculations: inpatients with RHD on
the CLINICOM information system of the hospital department of
Medical Informatics, outpatients with RHD on the REMEDY regis-
try,37 and patients on the database of the Division of Cardiothoracic
Surgery at Groote Schuur Hospital who had valve surgery.
Utilities
Utility values were derived from the literature (Table5), with the ex-
ception of the utility of RHD, which was derived from a local survey
of 48 adult patients with RHD attending the Cardiac Clinic at Groote
Schuur Hospital during September to December 2010. Patients were
asked 5 questions related to the dimensions of mobility, self-care,

Table 4. Costs (2010 $US)

Sensitivity Range
Variable Name Base Case Low High Sources/Base Case Assumptions
OPV at primary care clinic 15 7 29 McIntyre 34

OPV at primary care clinic (to patient) 2 1 4 Patient survey

OPV at L2/L3 hospital (to patient) 6 6 6 Patient survey at GSH September to December, 2010 (n=48)
Single IM penicillin vial (1.2 MU) 1 0 13 Hospital pharmacy 2010
Throat culture and serotyping 13 6 26 NHLS price schedule 2010
Symptomatic treatment of penicillin-induced rash 36 19 70 Assume return visit for check-up
Treatment of penicillin-induced anaphylaxis 302 302 594 1 IP day (SAMF 2010)
Treatment of PTA 465 226 705 2 OP visits and 1 IP day
ARF admission 2958 916 3064 Daily IM penicillin + 1 set of lab tests/ECG/echo/CXR
Secondary prevention 226 226 453 4 weekly IM penicillin (SA DoH guideline 1999)
RHD admission 1597 976 2840 7 days admission for heart failure, GSH
RHD valve surgery stay 10967 7532 14803 Assume weighted average cost per year
RHD medical management 769 469 1070 Assume IM penicillin and anticoagulation at L2/3
ARF indicates acute rheumatic fever; CXR, chest x-ray; echo, echocardiogram; GSH, Groote Schuur Hospital; IM, intramuscular; L2, level 2; L3, level 3; IP, inpatient;
MU, million units; NHLS, National Health Laboratory Service; OP, outpatient; OPV, outpatient patient visit; PTA, peritonsillar abscess; RHD, rheumatic heart disease; SA
DoH, South Africa Department of Health; and SAMF, South African Medicines Formulary.
Vanguard Community Health Center, Cape Town, South Africa (dates September to December 2010 [n=49]).
GSH, Cape Town, South Africa.
 Irlam et al Primary Prevention of Acute Rheumatic Fever 347

Table 5. Utility Variables Used in the Decision Tree Model CDR 3+ strategy ($13.00), treat none strategy ($14.39), the
CDR 2+, culture CDR negatives strategy ($16.42), and the
Variable Base Case Range Sources
CDR 3+, culture CDR negatives strategy ($23.89). The culture
Utility of ARF 0.995 0.90000.999 Neuner17 all strategy was the most costly at $27.21 per case of initial
Utility of RHD without surgery 0.848 0.75001.000 Patient survey pharyngitis. These costs include the cost of the initial visit, as
Utility of RHD postsurgery (adults) 0.881 0.80000.950 Milne6 well as any costs associated with future pharyngitis episodes,
Disutility of ARF 0.208 0.0250 to 2.04 Neuner17 hospitalizations, or outpatient treatments because of ARF or
Disutility of anaphylaxis 0.025 0.0080 to 0.050 Neuner17 RHD or suppurative complications, as well as complications
Disutility of PTA 0.014 0.0050 to 0.027 Neuner17
from a penicillin reaction.
In evaluating the ICERs, each strategys costs and effects were
Disutility of rash 0.002 0.0004 to 0.004 Neuner17
evaluated compared with the least costly intervention. Compared
ARF indicates acute rheumatic fever; PTA, peritonsillar abscess; and RHD, with the treat all strategy, the CDR 2+ strategy yielded greater
rheumatic heart disease.
discounted QALYs, which resulted in an ICER of $136/QALY
Groote Schuur Hospital September to December 2010 (n=48).
Disutility measured in years lost with 1 year a maximum if death occurs. compared with the treat all strategy and thus is the preferred
strategy with a WTP of $30000/QALY. Compared with the
usual activities, pain/discomfort, and anxiety/depression, using the CDR 2+ strategy, the CDR 3+ strategy was dominated because
standardized EQ-5D instrument38 to obtain their quantitative health it both costs more and leads to fewer QALYs (0.000041). The
Downloaded from http://circoutcomes.ahajournals.org/ by guest on January 8, 2017

states. These health states were then converted into a mean single sum-
mary index for the sample by using a value set obtained from a repre-
sentative sample of the general population in the United Kingdom.39
Data Analyses
All analyses were conducted using TreeAge Pro Version 2011.
Incremental cost-effectiveness ratios (ICERs) are calculated as the
incremental cost per quality-adjusted life year (QALY) gained.
One-way sensitivity analyses were performed on all variables
with sensitivity ranges as tabulated in Tables2 to 5. In addition to
the absolute ICERs, we also highlight where strategies would not
be acceptable at various willingness to pay (WTP) thresholds. The
WHO Commission on Macroeconomics and Health suggested that
interventions that add 1 QALY for less than the annual income per
capita are very cost-effective and those that cost <3 times the per capita
income are still cost-effective. Interventions above this WTP of 3
times the per capita income are thus not considered cost-effective. For
South Africa, the gross domestic product per capita was approximately
$10000 for the year 2010,40 and thus using the WHO criteria, $30000
per QALY could serve as an upper bound for being cost-effective.
Results
Base Case Analysis
Table6 displays the costs, effects, and ICERs for the 7 strategies
for the primary prevention of ARF and RHD in children with
GAS pharyngitis in urban primary care settings in South
Africa. The strategies are listed in order of increasing cost. The
treat all strategy has the lowest average cost of $11.19 per case
of pharyngitis, followed by the CDR 2+ strategy ($11.20), the
treat none strategy was eliminated because it both costs more
and leads to fewer QALYs than the treat all strategy and was thus
dominated. The CDR 2+, culture CDR negatives and the CDR
3+, culture CDR negatives strategies are eliminated by extended
dominance by a blend of the CDR 2+ and the culture all strat-
egy. Extended dominance is defined as those strategies with a
higher ICER than a more effective option. The next strategy that
was on the efficiency frontier was the culture all strategy, but at
$127600/QALY, it is well above the WTP for South Africa.
Sensitivity Analyses
In the base case analysis, the prevalence of GAS was estimated
at 15%. However, the results were sensitive to the prevalence
across a plausible range of 1.6% to 30%. At a GAS prevalence
of 1.6%, the least costly strategy was CDR 3+ alone, with cul-
ture all being the most effective but with an ICER of $1300000/
QALY when compared with the CDR 3+ strategy. All other
strategies were eliminated. This value is well above the 3 times
gross domestic product/capita that the WHO recommends for a
cost-effective intervention. Once the prevalence exceeds 2.9%,
however, the CDR 2+ strategy becomes the lowest cost strat-
egy, with the CDR 3+ strategy being the next best strategy with
an ICER of $198/QALY. From 2.9% until 7.3% prevalence,
the CDR 2+ strategy remains the lowest cost strategy and the
ICER for CDR 3+ below the $30000/QALY threshold of 3
times gross domestic product/capita. Beyond a prevalence of
12.9% up to 30%, the treat all strategy has the lowest cost, with

Table 6. Costs, Effects, and Incremental Cost-Effectiveness of Strategies for Primary Prevention of Acute Rheumatic Fever/
Rheumatic Heart Disease*
Strategy Cost, $US Incremental Cost Effect (QALY) Incremental Effect ICER ($/QALY)
Treat all 11.19 22.939936
CDR 2+ 11.20 0.01 22.940002 0.000066 136
CDR 3+ 13.00 1.80 22.939961 0.000041 (Dominated)
Treat none 14.39 3.19 22.939899 0.000103 (Dominated)
CDR 2+, culture CDR negatives 16.42 5.22 22.940017 0.000015 (Dominated)*
CDR 3+, culture CDR negatives 23.89 12.69 22.940101 0.000099 (Dominated)*
Culture all 27.21 16.01 22.940128 0.000126 127600
CDR indicates clinical decision rule; ICER, incremental cost-effectiveness ratio; and QALY, quality-adjusted life year.
*Eliminated by extended dominancethose strategies that are not eliminated by any 1 strategy but by a mixed strategy of 2 other programs and have higher ICERs
than a more effective option.
 348 Circ Cardiovasc Qual Outcomes May 2013
the CDR 2+ strategy remaining preferred and having an ICER
below $30000/QALY.
We assumed in our base case that 0.3% of those with GAS
will develop ARF if untreated. In our sensitivity analysis, we
varied the probability from 0.1% to 5%. At a probability of
0.1% to 0.29% and a WTP of $30000/QALY, the CDR 2+
is the least costly ($10$12 per patient) and the preferred
strategy. Above 0.29%, the treat all strategy is the lowest cost
strategy. However, from 0.29% to 1%, the CDR 2+ is still
the preferred strategy, with ICERs ranging from $5/QALY to
$29000/QALY using the standard WHO WTP threshold of
$30000/QALY. Beyond 1%, the treat all strategy is preferred.
The culture all strategy never has an ICER of below $120000/
QALY, which is found when the risk of developing ARF from
GAS is 0.3%, and the ICER rises as the probability increases.
The results were also sensitive to the probability of anaphy-
laxis from treatment. Using a WTP of $30000, both 2-stage
Downloaded from http://circoutcomes.ahajournals.org/ by guest on January 8, 2017
strategies were preferred once the risk of anaphylaxis reached
>3.4 per 10000. However, the culture all strategy was never
preferred unless the WTP was increased to $100000, and this
only occurred as the risk of anaphylaxis approached 4.1 of
10000 or 4 times the baseline assumption.
When we varied the lifetime risk of developing RHD sub-
sequent to ARF from 50% to 75%, there was no significant
change in the results. Similarly, the results were not sensitive
to a change in the probability of patients with ARF receiv-
ing secondary prevention from 0% to 90%. Using a very con-
servative WTP of $200000, the results were also insensitive
to the effectiveness of antibiotic treatment to prevent PTA,
ARF (primary), or RHD (secondary) across the range of val-
ues tested. Furthermore, the results were not sensitive to the
whole range of estimates for the probability of dying from
ARF, inducement of rash, and dying from anaphylaxis.
We further evaluated the sensitivity to changes in the test
characteristics. At no level below our base case estimate (0.92)
for the sensitivity of the CDR 2+ screening test were there
significant changes in the results. However, with a CDR 2+
sensitivity of 0.97, the CDR 2+ strategy would be the preferred
strategy. Regardless of the specificity of the CDR 2+, the treat
all strategy remained dominant, with no changes in the results
using a WTP of $30000. The results were not changed in the
various ranges for the sensitivity or specificity of the CDR
3+ either. When testing the range of values for the sensitivity
of the throat culture from 90% to 100%, the treat all strategy
remained the lowest cost strategy, and there were no changes in
the results until the sensitivity of throat culture exceeded 98%.
In this case, it made both 2-stage strategies using both CDR
and cultures acceptable only when using a WTP of $200000.
Only at a sensitivity of >98% does the culture all strategy have
an ICER at $186000/QALY. The results were insensitive to
changes across the range of specificity values we tested. The
results did not change significantly across the whole range of
sensitivity and specificity values we tested for the CDR 3+
criterion. All results assumed a rate of 100% return for culture
results. If the follow-up rate declines, however, then the ICERs
increase even more for the culture all strategy.
We assumed that the cost of throat culture requires a
return visit. Even when we allowed for a low-cost ($1) value
for phone follow-up, there is still a patient cost to return to
the clinic to get the injection, so the cost is only minimally
reduced. Even under these assumptions, the culture all strat-
egy was not acceptable compared with the clinical scores or
treat all strategies. The results were not sensitive to the cost
of the IM injection, the clinic visit cost, the cost of manag-
ing ARF, throat cultures, valve surgery, managing RHD long
term, discount rate, or any other costs tested across the ranges
reported in Table4. The only utility for which the results
were sensitive was for those who survived a prior ARF attack.
Because the utility was lowered from our base case of 0.995,
the ICER for the CDR 2+, culture CDR negatives increased;
and at a threshold of 0.91, all strategies were dominated by the
treat all strategy.
Our probabilistic sensitivity findings were similar to the
above findings (Figure2). At a WTP of $10000/QALY, the
CDR 2+ strategy was the preferred strategy in 60% of the iter-
ations and 20% in treat all, with the remainder of the itera-
tions being made up of CDR 3+ (10%) or no treatment. At
$30000/QALY, the results were similar, with a slight decrease
in preference of the CDR 2+. At $150000/QALY, the CDR 2+
and the treat all strategy were still the preferred strategy >50%
of the time. Culture all is only a preferred strategy 22% of the
time at this very high ratio. The analyses were not sensitive to
changes in the range of utilities for RHD, anaphylaxis, or the
disutility of ARF, anaphylaxis, rash, or PTA.
Discussion
ARF and RHD have a relatively high incidence in South
Africa and are a significant health concern in developing
countries.41,42 Our analyses show that in an urban South Afri-
can setting, a strategy of treating all children with IM penicil-
lin who present to primary care centers with pharyngitis is
the least costly strategy, costing about $11 per case of phar-
yngitis. The most cost-effective strategy, however, was using
a CDR with 2 or more clinical criteria of enlarged cervical
nodes, absence of rhinitis, or absence of rash. It was mini-
mally more expensive, resulting in a very good return of about
$150 per QALY gained. This value is well below the recom-
mended threshold of 3 times the GDP/capita recommended by
the WHO for ICER.43 The remaining strategies either did not
result in long-term QALYs gained or resulted in too high an
incremental cost per QALY gained. Culturing everyone com-
pared with the clinical score was prohibitively expensive at
more than $125000 per QALY. Our analysis is sensitive to
2 key variables: the likelihood of developing ARF with GAS
pharyngitis, also known as the attack rate; and the prevalence
of GAS. The literature suggests a fairly broad range of esti-
mates for the attack rate from as high as 5%44 in epidemics to
as low as 0.3%45 and possibly lower in stable endemic situ-
ations. When the attack rate is closer to the epidemic rate,
treating everyone may be a preferred strategy, especially if
skin testing for drug allergy is possible. When the attack rate
is closer to the endemic rates reported in the United States,
then clinical scoring is a preferred strategy. Given that it is
unlikely that studies to evaluate attack rates will be con-
ducted for ethical reasons, a region may not be able to assess
its attack rate and will thus have to assume the lower attack
rate unless an epidemic is occurring. When the prevalence of
GAS is <13%, then the clinical rules with either 2 or 3 positive
 Irlam et al Primary Prevention of Acute Rheumatic Fever 349
Downloaded from http://circoutcomes.ahajournals.org/ by guest on January 8, 2017
findings become preferred strategies, and treating everyone is
no longer a preferred option. Therefore, it remains essential
for countries or regions to know their local GAS prevalence.
Treating everyone is a low-cost strategy because it essen-
tially has a sensitivity of 100%, and the potential downstream
costs of ARF and RHD that are prevented per GAS case out-
weigh the costs of the relatively cheap IM injection of about
$1. Overall, the incremental costs of the program to treat all
cases of pharyngitis during the lifetime of children from ages
3 to 15 years would cost about $580 per 100000 children per
year. Furthermore, the health benefits gained from preventing
ARF or RHD outweigh the relatively minimal risk of severe
reactions from anaphylaxis in aggregate, provided the inci-
dence of ARF after untreated pharyngitis remains >1%. The
same was found in a US study30 that reported that treating all
children with antibiotics is the least costly strategy. Lieu et
al21 also found that a treat all strategy leads to the greatest
reduction in potential GAS complications (suppurative and
nonsuppurative) but led to many minor drug reactions and
rare fatal reactions. They did not calculate the QALYs, how-
ever, but only the cost per ARF, RHD, or suppurative compli-
cation, which was lowest with the treat all strategy. Because
they did not examine overall mortality, they could not make an
overall comparison of the strategies. Tsevat et al14 found the
culture strategy less costly, but they did not include the long-
term costs of secondary prophylaxis or the cost of RHD and
its adverse health consequences, which are some of the major
motivations for ARF prevention. Van Howe et al20 included
RHD complications and costs but grossly underestimated
the probability of GAS pharyngitis. They used one fiftieth of
Siegels estimate of 0.3% as the likely probability of untreated
GAS converting to ARF, citing more recent evidence suggest-
ing that the incidence of ARF has decreased 50-fold since
1961.46 This value represents incidence in the US popula-
tion, however, which has markedly increased treatment rates
for pharyngitis and different crowding and socioeconomic
Figure2. Cost-effectiveness acceptability
curve (probabilistic sensitivity analysis).
CDR indicates clinical decision rule.
conditions. It may not reflect the natural history of the disease
in a relatively untreated population. We therefore used the
estimate of 0.3% to reflect the relatively low treatment rates
and more crowded conditions in South African urban town-
ships. Michaud et al,2 using the estimated attack rate of ARF
from GAS of 3%, estimated the cost-effectiveness of primary
prevention at $22000 to $33000 per disability-adjusted life
year averted for treatment of all GAS, including asymptom-
atic cases. This assumed that 50% of patients never come to
the clinic and that 30% are asymptomatic. Because we only
evaluated symptomatic patients that present, this would be
about $3500 per disability-adjusted life year averted, accord-
ing to their estimates in high endemic settings.
The authors of the 2 studies,21,30 which show that a treat all
strategy leads to fewer total deaths, still did not recommend
this strategy. Their resistance stems from 2 concerns. The
first concern is about unnecessary severe reactions from the
treatment as part of the desire to fulfill the Hippocratic oaths
primum, non nocere. Certainly, there are negative health
consequences to treat false positives, as well as not treating
false negatives. However, we are also reminded of Benjamin
Franklins lament of not inoculating his son, whom he lost to
smallpox in 1736:
I bitterly regretted that I had not given it to him by inocu-
lation. This I mention for the sake of parents who omit that
operation, on the supposition, that they should never forgive
themselves if a child died under it. My example shows the
regret may be the same either way, and that therefore the safer
should be chosen.47
The second concern relates to the potential risk of drug
resistance with the overuse of antibiotics. There have been
several instances of large-scale use of penicillin to treat sus-
pected streptococcal pharyngitis in Puerto Rico,48 Guadeloupe
and Martinique,49 and Cuba50 for the primary prevention of
ARF. There were no reports of drug resistance to penicillin in
GAS in these countries, and to the best of our knowledge, all
 350 Circ Cardiovasc Qual Outcomes May 2013
GAS strains remain sensitive to penicillin, and even the mini-
mum inhibitory concentrations have not changed in >70 years.
Therefore, the concern about drug resistance to penicillin in
GAS is not supported by the present evidence. Therefore, we
were unable to calculate the potential health or medical costs
of drug resistance, and neither have any other cost-effective-
ness analyses of primary prevention.
If a society wishes to value death averted from anaphy-
laxis greater than death averted from ARF, suppurative com-
plications, or from RHD, then this valuation should be made
explicit and quantifiable so that appropriate decisions could be
made. However, we are not sure whether this rationale would
not also need to be applied to all interventions where there
is some acceptable up-front risk for some long-term reduc-
tion in overall risk, as is seen with procedures such as car-
diac surgery for ischemic heart disease. Similarly, if there is
a concern about antibiotic resistance, then models to evaluate
Downloaded from http://circoutcomes.ahajournals.org/ by guest on January 8, 2017
this harm should be developed. If either of these concerns is
heavily weighted, a strategy using a clinical rule first and then
culturing only those who do not meet clinical thresholds could
be a more palatable but perhaps more expensive alternative.
Our analysis differs from previous models in other ways
as well. Lieu et al21 do not place utilities on different health
states or follow a full life course, and thus they did not report
a cost/QALY, only a cost per case averted of ARF. Tsevat et
al14 report costs per life years saved, but no utility adjustments
are made for the various health states, and the long-term costs
of secondary prevention or RHD are not included. Our model
considers the full set of downstream events and costs associ-
ated with untreated GAS infection, including increased risk
of ARF, development of RHD, heart failure, and the need for
valve surgery. Fewer missed diagnoses (false negatives) in a
setting with a relatively high incidence of post-GAS cardiac
sequelae minimize the costs of later treatment of ARF and
RHD, weighed against the comparatively minor expense and
possible antibiotic resistance arising from the treatment of
false-positive cases.
We did not consider the use of rapid antigen testing in this
analysis for several reasons. First, this study was designed to
assess the cost-effectiveness of diagnostic approaches that are
in use in a South African urban setting. Rapid antigen testing
is not used in the South African public health service, which
caters to >80% of the population. Second, this study was
designed to delineate the optimal strategy within the available
resources, which is likely to be rapidly translated into policy
and practice. Finally, the performance of the rapid antigen test
is more appropriate for countries with low endemicity of ARF
and RHD, given the high specificity and low sensitivity of this
screening test. Our study shows that approaches with a high
sensitivity are preferable in settings with a high incidence of
ARF. An additional limitation is that we report on the cost-
effectiveness only on those who report to the clinic for care.
We recognize that children do not always have access or are
unable to present with pharyngitis in low-income settings or
that many cases of GAS are asymptomatic. Wider efforts to
reduce ARF and RHD must therefore not be neglected if the
burden is to be controlled.
In conclusion, we demonstrate that a strategy of treating
all patients presenting in urban primary care clinics with
symptomatic pharyngitis and at least 2 additional features is
the most cost-effective intervention for the prevention of ARF
and RHD in populations with a high burden of these diseases,
such as among children living in sub-Saharan Africa and other
developing countries. The findings of this study have implica-
tions for public health policy.
Acknowledgments
We acknowledge the assistance of the following in this study by pro-
viding information on costs of care, standards of treatment, quality
of life with rheumatic disease, or frequency of treatments for acute
rheumatic fever and rheumatic heart disease: Motasim Badri, Wendy
Bryant, Patrick Commerford, Alexia Joachim, Wendy Mathiassen,
Lauren Watkins, Kathie Walker, and Liesl Zuhlke.
Disclosures
Dr Gaziano is a faculty member of the Brigham and Womens
Hospital. J. Irlam is employed in the Primary Health Care Directorate
of the University of Cape Town. Drs Mayosi and Engel are employed
in the Department of Medicine of the Groote Schuur Hospital.
References
1. Carapetis JR, McDonald M, Wilson NJ. Acute rheumatic fever. Lancet.
2005;366:155168.
2. Michaud C, Rammohan R, Narula J. Cost-effectiveness analysis of inter-
vention strategies for reduction of the burden of rheumatic heart disease.
In: Narula J, Virmani R, Reddy KS, Tandon R, eds. Rheumatic Fever.
Washington, WA: American Registry of Pathology; 1999:485497.
3. Pastore S, De Cunto A, Benettoni A, Berton E, Taddio A, Lepore L. The
resurgence of rheumatic fever in a developed country area: the role of
echocardiography. Rheumatology (Oxford). 2011;50:396400.
4. Omurzakova NA, Yamano Y, Saatova GM, Mirzakhanova MI, Shukurova
SM, Kydyralieva RB, Jumagulova AS, Seisenbaev ASh, Nishioka K,
Nakajima T. High incidence of rheumatic fever and rheumatic heart
disease in the republics of Central Asia. Int J Rheum Dis. 2009;12:
7983.
5. Karthikeyan G, Mayosi BM. Is primary prevention of rheumatic fe-
ver the missing link in the control of rheumatic heart disease in Africa?
Circulation. 2009;120:709713.
6. Milne RJ, Lennon D, Stewart J, Scuffham P, Hoorn SV, Cooke J, Remeny
B, Finucane K, Wilson N, Nicholson R. Economic evaluation of a school
intervention to reduce the risk of rheumatic fever. A report to the Ministry
of Health, 2011.
7. Gaziano TA. Economic burden and the cost-effectiveness of treatment of
cardiovascular diseases in Africa. Heart. 2008;94:140144.
8. Fischer Walker CL, Rimoin AW, Hamza HS, Steinhoff MC. Comparison
of clinical prediction rules for management of pharyngitis in settings with
limited resources. J Pediatr. 2006;149:6471.
9. Rimoin AW, Hamza HS, Vince A, Kumar R, Walker CF, Chitale RA,
da Cunha AL, Qazi S, Steinhoff MC. Evaluation of the WHO clinical
decision rule for streptococcal pharyngitis. Arch Dis Child. 2005;90:
10661070.
10. Steinhoff MC, Walker CF, Rimoin AW, Hamza HS. A clinical decision
rule for management of streptococcal pharyngitis in low-resource settings.
Acta Paediatr. 2005;94:10381042.
11. Manyemba J, Mayosi BM. Penicillin for secondary prevention of rheu-
matic fever. Cochrane Database Syst Rev. 2002;3:CD002227.
12. The National Department of Health, South Africa. National guidelines on
primary prevention and prophylaxis of rheumatic heart fever and rheu-
matic heart disease for health professionals at primary level. South African
Medical Journal. 1999;89(Suppl 2):C91C94.
13. Stollerman GH. Penicillin for streptococcal pharyngitis: has anything
changed? Hosp Pract (Off Ed). 1995;30:8083.
14. Tsevat J, Kotagal UR. Management of sore throats in children: a cost-
effectiveness analysis. Arch Pediatr Adolesc Med. 1999;153:681688.
15. Denny FW. T. Duckett Jones and rheumatic fever in 1986. T. Duckett
Jones memorial lecture. Circulation. 1987;76:963970.
16. Robertson KA, Volmink JA, Mayosi BM. Lack of adherence to the
National Guidelines on the Prevention of Rheumatic Fever. S Afr Med J.
2005;95:5256.
 Irlam et al Primary Prevention of Acute Rheumatic Fever 351
17. Neuner JM, Hamel MB, Phillips RS, Bona K, Aronson MD. Diagnosis
and management of adults with pharyngitis: a cost-effectiveness analysis.
Ann Intern Med. 2003;139:113122.
18. Feinstein AR, Stern EK, Spagnuolo M. The prognosis of acute rheumatic
fever. Am Heart J. 1964;68:817834.
19. Hillner BE, Centor RM. What a difference a day makes: a deci-
sion analysis of adult streptococcal pharyngitis. J Gen Intern Med.
1987;2:244250.
20. Van Howe RS, Kusnier LP II. Diagnosis and management of pharyngitis
in a pediatric population based on cost-effectiveness and projected health
outcomes. Pediatrics. 2006;117:609619.
21. Lieu TA, Fleisher GR, Schwartz JS. Cost-effectiveness of rapid latex
agglutination testing and throat culture for streptococcal pharyngitis.
Pediatrics. 1990;85:246256.
22. Dippel DW, Touw-Otten F, Habbema JD. Management of children with
acute pharyngitis: a decision analysis. J Fam Pract. 1992;34:149159.
23. Carapetis JR, Currie BJ, Mathews JD. Cumulative incidence of rheumatic
fever in an endemic region: a guide to the susceptibility of the population?
Epidemiol Infect. 2000;124:239244.
24. Michaud CM, Trejo-Gutierrez J, Cruz C, Pearson T. Rheumatic heart
disease. In: Jamison DT, Mosely WH, Measham AR, Bobadilla JL, eds.
Disease Control Priorities in Developing Countries. New York: Oxford
Downloaded from http://circoutcomes.ahajournals.org/ by guest on January 8, 2017
University Press; 1993:221232.
25. Kumar R, Raizada A, Aggarwal AK, Ganguly NK. A community-based
rheumatic fever/rheumatic heart disease cohort: twelve-year experience.
Indian Heart J. 2002;54:5458.
26. Grayson S, Horsburgh M, Lennon D. An Auckland regional audit of the
nurse-led rheumatic fever secondary prophylaxis programme. N Z Med J.
2006;119:U2255.
27. Robertson KA, Volmink JA, Mayosi BM. Antibiotics for the primary
prevention of acute rheumatic fever: a meta-analysis. BMC Cardiovasc
Disord. 2005;5:11.
28. Padmavati S, Sharma KB, Jayaram O. Epidemiology and prophylaxis
of rheumatic fever in Delhia five year follow-up. Singapore Med J.
1973;14:457461.
29. Del Mar CB, Glasziou PP, Spinks AB. Antibiotics for sore throat.
Cochrane Database Syst Rev. 2006:CD000023.
30. Webb KH. Does culture confirmation of high-sensitivity rapid strep-
tococcal tests make sense? A medical decision analysis. Pediatrics.
1998;101:E2.
31. Statistics South Africa. Statistics South Africa annual report 2007/8.
Statistics South Africa. 2008.
32. World Health Organization. The World Health Report 2006. Working to-
gether for health. 2006.
33. Kellogg JA. Suitability of throat culture procedures for detection of group
A streptococci and as reference standards for evaluation of streptococcal
antigen detection kits. J Clin Microbiol. 1990;28:165169.
34. McIntyre D. Shield work package 5 report: modelling the estimated re-
source requirements of alternative health care financing reforms in South
Africa. Health Economics Unit, University of Cape Town. 2010.
35. Health and Medical Publishing Group of the South African Medical
Association. Division of Clinical Pharmacology, Faculty of Health
Sciences, University of Cape Town. South African medicines formulary.
2010.
36. The National Department of Health, South Africa. Standard treatment
guidelines and essential drugs list for South Africa; hospital level (adults).
2006.
37. Karthikeyan G, Zhlke L, Engel M, Rangarajan S, Yusuf S, Teo K, Mayosi
BM. Rationale and design of a Global Rheumatic Heart Disease Registry:
the REMEDY study. Am Heart J. 2012;163:535540.e1.
38. The EuroQol Group. Euroqol: a new facility for the measurement of
health-related quality of life. Health Policy. 1990;16:199208.
39. Dolan P. Modeling valuations for EuroQol health states. Med Care.
1997;35:10951108.
40. World Bank. World Development Indicators. Washington, DC: The World
Bank; 2012.
41. Sliwa K, Carrington M, Mayosi BM, Zigiriadis E, Mvungi R, Stewart S.
Incidence and characteristics of newly diagnosed rheumatic heart disease
in urban African adults: insights from the heart of Soweto study. Eur Heart
J. 2010;31:719727.
42. Tibazarwa KB, Volmink JA, Mayosi BM. Incidence of acute rheumatic
fever in the world: a systematic review of population-based studies. Heart.
2008;94:15341540.
43. World Health Organization. Who-choice. Choosing interventions that are
cost-effective. Geneva, Switzerland; 2003.
44. Catanzaro FJ, Stetson CA, Morris AJ, Chamovitz R, Rammelkamp CH Jr,
Stolzer BL, Perry WD. The role of the streptococcus in the pathogenesis
of rheumatic fever. Am J Med. 1954;17:749756.
45. Siegel A, Johhson E, Stollerman G. Controlled studies of streptococcal
pharyngitis in a pediatric population. NEJM. 1961;265:559566.
46. Land MA, Bisno AL. Acute rheumatic fever: vanishing disease in subur-
bia. JAMA. 1983;249:895898.
47. Schmidt WM. Health and welfare of Colonial American children. Am J
Dis Child. 1976;130:694701.
48. Arguedas A, Mohs E. Prevention of rheumatic fever in Costa Rica. J
Pediatr. 1992;121:569572.
49. Bach JF, Chalons S, Forier E, Elana G, Jouanelle J, Kayemba S, Delbois
D, Mosser A, Saint-Aime C, Berchel C. 10-year educational programme
aimed at rheumatic fever in two French Caribbean islands. Lancet.
1996;347:644648.
50. Nordet P, Lopez R, Dueas A, Sarmiento L. Prevention and control of
rheumatic fever and rheumatic heart disease: the Cuban experience (1986-
1996-2002). Cardiovasc J Afr. 2008;19:135140.
 Primary Prevention of Acute Rheumatic Fever and Rheumatic Heart Disease With
Penicillin in South African Children With Pharyngitis: A Cost-Effectiveness Analysis
James Irlam, Bongani M. Mayosi, Mark Engel and Thomas A. Gaziano
Downloaded from http://circoutcomes.ahajournals.org/ by guest on January 8, 2017

Circ Cardiovasc Qual Outcomes. 2013;6:343-351; originally published online May 7, 2013;
doi: 10.1161/CIRCOUTCOMES.111.000032
Circulation: Cardiovascular Quality and Outcomes is published by the American Heart Association, 7272
Greenville Avenue, Dallas, TX 75231
Copyright 2013 American Heart Association, Inc. All rights reserved.
Print ISSN: 1941-7705. Online ISSN: 1941-7713

The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circoutcomes.ahajournals.org/content/6/3/343

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Circulation: Cardiovascular Quality and Outcomes can be obtained via RightsLink, a service of the
Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for
which permission is being requested is located, click Request Permissions in the middle column of the Web
page under Services. Further information about this process is available in the Permissions and Rights
Question and Answer document.

Reprints: Information about reprints can be found online at:

http://www.lww.com/reprints

Subscriptions: Information about subscribing to Circulation: Cardiovascular Quality and Outcomes is online
at:
http://circoutcomes.ahajournals.org//subscriptions/