Health Policy and Planning Advance Access published March 14, 2014

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), which permits
unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Health Policy and Planning 2014;1–9
Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine doi:10.1093/heapol/czu013

REVIEW

Price subsidies increase the use of private

sector ACTs: evidence from a systematic review
Alexandra Morris,1* Abigail Ward,1 Bruno Moonen,1 Oliver Sabot2 and Justin M. Cohen1
1
Clinton Health Access Initiative, 383 Dorchester Avenue, Suite 400, Boston, MA, USA and 2Slingshot, 67 Riverside Dr. Nairobi Kenya 00100
London, UK
*Corresponding author. Clinton Health Access Initiative, 383 Dorchester Avenue, Suite 400, Boston, MA 02127, USA.
E-mail: alix.morris@gmail.com

Accepted 4 February 2014

Downloaded from http://heapol.oxfordjournals.org/ by guest on March 16, 2014

Background Although artemisinin combination therapies (ACTs) are the recommended first-
line treatment for uncomplicated malaria in most endemic countries, they have
been prohibitively expensive in the retail sector where many suspected malaria
cases purchase treatment. ACT subsidies seek to stimulate consumer demand for
the drugs over cheaper but often ineffective alternatives by reducing their prices.
Recent evidence from eight regions implementing such subsidies suggests that
they are generally successful in improving availability of the drugs and
decreasing their retail prices, but it remains unclear whether these outcomes
translate to improved use by patients with suspected malaria.

Methods
and findings A systematic literature review was conducted to identify reports of experimental
or programmatic ACT subsidies to assess the impact of subsidies on consumer
use. Relationships between price, use and potential confounding factors were
examined using logistic and repeated measures binomial regression models, and
approximate magnitudes of associations were assessed with linear regression. In
total, 40 studies, 14 peer-reviewed and 26 non-peer-reviewed, were eligible for
inclusion in the analysis. The reviewed studies found a substantial increase in
private sector ACT use following the introduction of a subsidy. Overall, each $1
decrease in price was linked to a 24 percentage point increase in the fraction of
suspected malaria cases purchasing ACTs (R2 ¼ 0.302). No significant differences
were evident in this relationship when comparing the poorest and richest groups,
rural vs urban populations or children vs adults.

Conclusions These findings suggest that ACT price reductions can increase their use for
suspected malaria, even within poorer, more remote populations that may be
most at risk of malaria mortality. Whether a subsidy is appropriate will depend
upon local context, including treatment-seeking behaviours and malaria preva-
lence. This review provides an initial foundation for policymakers to make
evidence-based decisions regarding ACT price reductions to increase use of
potentially life-saving drugs.
Keywords Subsidy, artemisinin combination therapies, malaria, systematic review

1
2 HEALTH POLICY AND PLANNING
KEY MESSAGES
 Subsidies increased private sector ACT usage by children under the age of five in all of the studies identified, from
7 percentage points in Nigeria to 49 percentage points in Uganda (both at national scale).
 On average, a 1 dollar reduction in the price of ACTs was associated with a 24 percentage point increase in ACT usage
among those of any age, and a 32 percentage point increase among children under five.
 The observed increases in ACT use in the private sector following subsidization appear equitable; subsidies substantially
increased ACT use even among the poorest individuals studied, and no significant differences were apparent in ACT use
increases between those living in rural vs urban areas.
Introduction
Increasing the use of artemisinin combination therapies (ACTs),
the recommended first-line treatment for uncomplicated mal-
aria in most endemic countries, is a central pillar of the global
effort to reduce malaria mortality (WHO 2010). Over the past
8 years, substantial international donor investment has focused
on making ACTs widely available in the public sector in many
countries in sub-Saharan Africa. However, ACT use is greatly
limited by the medicine’s high cost in the private sector, where
roughly half of malaria patients in sub-Saharan Africa, and an
even higher proportion in Southeast Asia, purchase treatment
(WHO 2011). In recent years, ACTs have been found to be more
than 10 times more expensive than common but potentially
ineffective anti-malarials such as chloroquine and sulphadoxine
pyrimethamine (White 2008). As a result, many patients who
obtain treatment in the private sector continue to use sub-
optimal medicines.
To address this financial barrier and increase access to
affordable ACTs, several countries have introduced programmes
to subsidize the price of ACTs in the private sector. Most
recently, in 2009, the Global Fund to Fight AIDS, Tuberculosis
and Malaria (the Global Fund) launched the Affordable
Medicines Facility—malaria (AMFm), an international finan-
cing mechanism piloted in eight territories—such as Nigeria,
Niger, Madagascar, Uganda, Ghana, Kenya, Tanzania and
Zanzibar. The initial phase of the AMFm was comprised of
three elements: (1) price reductions through negotiations with
ACT manufacturers, (2) a buyer subsidy offered through a co-
payment at the manufacturer level of the supply chain and (3)
supporting interventions to promote appropriate use of ACTs.
The AMFm provided a roughly 90% co-payment for ACTs
ordered by both public and private sector buyers in an effort to
reduce the price to consumers.
A policy review published in 2011 suggested ACT subsidies
were successful in improving availability of the drugs and
decreasing their retail prices (Schäferhoff and Yamey 2011).
However, little data were available on whether these inter-
mediary outcomes translated to an improvement in the fraction
of suspected or confirmed malaria patients who actually
received an ACT. Following the launch of the AMFm, an
independent evaluation was commissioned to assess changes in
ACT price, availability, market share and use (Arnold et al.
2012). An extensive analysis was conducted to measure the
changes in these outcomes at national scale across the pilot
countries before and after the AMFm intervention. Due to cost
and complexity, however, the Global Fund decided to not
conduct measurement of ACT use as part of that evaluation.
Instead, the evaluation considered existing household surveys
available through Demographic Health Surveys (DHS) and
other sources, including surveys conducted by ACTwatch Group
of Population Services International (ICF Macro and LSHTM
2012). The report on ACT use concluded that three of the five
countries with available evidence demonstrated at least a five
Downloaded from http://heapol.oxfordjournals.org/ by guest on March 16, 2014
percentage point increase in ACT use pre- and post-subsidy
introduction.
This investigation seeks to provide additional evidence on
how ACT use changes when prices are subsidized, drawing
on additional data from operational research measuring the
impact of the AMFm, as well as from prior experimental and
programmatic subsidies. A comprehensive review of ACT use
data in all areas where an ACT subsidy has been introduced
and evaluated was conducted, and these data were used to
assess the empirical relationship between ACT use and the
proximal factors of ACT affordability and availability. This
additional evidence can inform policymakers and funders on if
and how interventions to reduce the price of ACTs can enable
malaria endemic countries to reach the Roll Back Malaria
Partnership goals of achieving universal access in the private
sector (RBM 2011).
Methods
Literature review
A systematic literature review was conducted to identify studies
that examined subsidies for artemisinin-based drugs in the
private sector according to PRISMA reporting protocol guide-
lines (Liberati et al. 2009). The private sector was defined as
private healthcare facilities, pharmacies and over-the-counter or
retail shops. PubMed/Medline and EMBASE were searched in
May 2013 for the key words ‘subsidy’ (or ‘subsidi*’) and a
MESH heading for ‘malaria’. Non-peer-reviewed evaluations
were identified through targeted outreach to investigators of
relevant articles and key institutions. Studies were included if
they met the following eligibility criteria: (1) the study assessed
a financial subsidy in the private sector for an ACT between
1990 and May 2013, (2) study participants were anti-malarial
drug customers or users and (3) the study measured whether or
not a suspected malaria case received or an anti-malarial
customer purchased an ACT. Studies were excluded if no ACT
subsidy intervention was introduced, the subsidy was imple-
mented only in the public sector or if there was no attempt to
measure ACT use as part of the assessment.
 SUBSIDIES INCREASE ACT USE: A SYSTEMATIC REVIEW
All ACT subsidies described in the reviewed literature were
categorized as either corresponding to (1) the AMFm, (2) an
experimental subsidy or (3) a non-AMFm, non-experimental
programmatic subsidy, typically implemented by the govern-
ment or an non-governmental organizations. These categories
were selected to examine the effect of the AMFm specifically, to
observe rigorously measured impact of subsidization in research
studies that compared changes against controls and to review
evidence on the impact of prior programmatic subsidies. Any
information provided on anti-malarial use or purchasing
behaviour, patient characteristics and ACT stocking and pricing
in retail outlets or health facilities was recorded. The ACT
stocking and pricing data were used to look for associations
between these measures and use rates.
ACT use has been defined as both (1) the proportion of
patients with fevers (or suspected malaria illness) who receive
an ACT (Kangwana et al. 2011) and (2) the proportion of anti-
malarial recipients who receive an ACT (Talisuna et al. 2012).
The second definition is more appropriate for contexts where
anti-malarial use rates may be changing as a result of factors
unrelated to the drug subsidy and may also be more appropri-
ate for evaluating the impact of subsidies given that their stated
goal is typically to replace ineffective alternative treatment with
ACTs rather than to increase the use of anti-malarials in
general (Institute of Medicine 2004). If, however, a subsidy
does cause changes in the overall envelope of anti-malarial
use, this change would not be captured by the second metric.
The second narrower definition is primarily used here, although
the first is also presented when it is the only metric provided by
the reviewed sources.
Two principal methods to measure the fraction of anti-
malarial users who take ACTs are used in the literature. The
first involves surveys of the community at the household level.
This approach is useful because it allows understanding of ACT
use from any source, whether public or private, and it allows
population-representative measures to be estimated. However,
these household surveys are subject to recall biases (Boerma
et al. 1991), and as drug choice is not directly observed, ACT use
rates may not be represented accurately. The second approach
involves interviewing customers as they leave drug-dispensing
outlets. These ‘exit interviews’ are typically conducted only on a
few days outside a select set of outlets, and may or may not be
truly representative of the entire region. Since each of these
approaches has advantages and limitations in providing a
robust measurement of the target indicator, both are reviewed
here; clear differentiation is made between the two throughout
this review so that the results can be evaluated accordingly.
Data analysis
Evidence for the impact of subsidies on use was reviewed for
each of the three categories of subsidy. Most studies reported
the results of cross-sectional surveys conducted at time points
before, during and/or after subsidy implementation, permitting
examination of how ACT use changed over time. Experimental
subsidies additionally provided measurements of use in control
groups, allowing attribution of impact to the subsidy specific-
ally. These trends were examined both overall and, where
possible, specifically in the private sector. In the public sector,
ACT scale up has occurred in many countries through
3
implementation of Global Fund grants and other interventions
(WHO 2011). However, no known private sector interventions
at national scale were implemented during this period other
than the subsidies under consideration.
Where possible, trends were examined within specific sub-
groups of interest, including children under five, those of lowest
socioeconomic status (SES) and those living in rural areas. SES
was measured by comparing the highest and lowest wealth
index quintiles, generated from principal components analysis of
assets by ACTwatch using demographic and health survey
methodology. Rurality was considered as a dichotomous rural/
urban variable defined separately by each study. Differences in
use and increases in use during subsidies between high- and
low-SES groups and those in urban and rural locations were
compared statistically using logistic regression models.
In a second set of analyses, all studies reporting not only use
but also stocking of ACTs and sales prices were compiled to
specifically examine the linkage between commonly measured
Downloaded from http://heapol.oxfordjournals.org/ by guest on March 16, 2014
indicators at the retail level and the important but less
commonly evaluated impact on use. The relationship between
the proportion of outlets stocking ACTs and the average price of
ACTs at those outlets was explored using repeated measures
logistic regression models with the GENMOD procedure in SAS
v9.2. The outcome of this model was whether or not an outlet
stocked ACTs. Separate models were run for ACTs for children
under five and ACTs for any age. The predictor variable for
these models was the average price of ACTs for children under
five and for any age, respectively. An exchangeable covariance
structure was specified in each model to account for within-
survey correlation. Because the resulting odds ratios (ORs) from
these models are difficult to interpret (Davies et al. 1998),
ordinary least squares regression models were also fit to the
same data, regressing the fraction of outlets stocking ACTs on
the mean ACT price. The resulting linear fit provided a readily
interpretable estimate of the average magnitude of the rela-
tionship between the two variables. The same approach was
then used to examine the relationships between the stocking of
ACTs, ACT price and the proportion of anti-malarial customers
choosing to purchase them.
Results
The literature search yielded 370 published, peer-reviewed
studies. A total of 206 appeared relevant and were screened
by abstract review, of which 97 studies were then selected for
full-text review. In addition to the studies identified through
the database search, 36 studies were included that were
identified through non-peer-reviewed literature or targeted
outreach to investigators. In total, 40 studies were eligible for
inclusion in the analysis and qualitative synthesis. Of these,
14 were peer-reviewed publications, and 26 were non-peer-
reviewed literature (Figure 1). These reports described 10
experimental subsidies in eight countries, non-AMFm program-
matic subsidies in nine countries and AMFm subsidies in eight
pilots. However, only four experimental subsidies, three pro-
grammatic subsidies and five of the eight AMFm pilot subsidies
measured ACT use and were therefore included in the quan-
titative analysis (Supplementary Table S1).
4 HEALTH POLICY AND PLANNING

through database
screening (n=370)
PubMed (n=134)
sources (n=36)
EMBASE (n=215)
Cochrane (n=21)

168 duplicate records

(n=238)

32 records excluded based on

Commentary, proceeding, clinical

Screening

Records screened based on abstract review if:

Downloaded from http://heapol.oxfordjournals.org/ by guest on March 16, 2014

an ACT or;

or ACT usage rate

46 full-text records from database 11 records from other sources

search excluded because: excluded because:
Eligibility

Full-text records assessed for

eligibility (n=97) No full text available (n=2)
Review with no original data measured/presented (n=5)
(n=10)
ACT usage not measured or pre- study (n=2)
sented (n=20) Data in a published paper (n=2)
Public sector only (n=2)
(n=1)
Refers to another study (n=3)
Randomized ACT user groups (n=2)
Public sector only (n=6)
Pre-ACT (n=1)

40 records included in
Included

14 records included 26 records included

synthesis from database synthesis from other

search (12 studies) sources (28 studies)

Figure 1 Literature flowchart of ACT subsidies measuring use.

Experimental subsidies Non-AMFm programmatic subsidies

Four studies included data on ACT use after subsidization in
experimental groups compared with unsubsidized controls
(Figure 2). Two of these experiments compared household
survey responses regarding the fraction of febrile episodes
treated with ACTs (Figure 2A) (Kangwana et al. 2011; Cohen
et al. 2012a,b), while the other two compared the fraction of
anti-malarial purchasers receiving ACTs vs other anti-malarials
(Figure 2B) (Sabot et al. 2009a,b; Cohen et al. 2010; Talisuna
et al. 2012). All four studies found the fractions reporting
receiving or purchasing ACTs were significantly higher when
subsidies were implemented.
Nine programmatic subsidies implemented prior to the AMFm
were identified; however, only three subsidies included data on
ACT use. In Cambodia, access to subsidized ACTs was scaled up
at private retail outlets in 2002. Before the national scale up,
household survey results indicated that only 8% of anti-malarial
recipients received ACTs (Yeung et al. 2011); surveys unrelated
specifically to the subsidy found 13% receiving ACTs in 2006
(PSI 2006), 65% in 2009 (ACTwatch Group, Population Services
International 2009) and 68% in 2011 (ACTwatch Group and
Population Services International 2011). In 2007 and 2008, the
Tanzanian government expanded access to subsidized ACTs
 SUBSIDIES INCREASE ACT USE: A SYSTEMATIC REVIEW 5

A B

Downloaded from http://heapol.oxfordjournals.org/ by guest on March 16, 2014

Figure 2 Summary of experimental studies with controlled use data. The fraction of suspected malaria patients receiving ACTs over time in the
subsidized and non-subsidized study groups are shown according to (A) household surveys of all febrile household members in the rural districts of
Busia, Mumias and Samia Kenya (dashed), and febrile children under five in rural Busia, Butere-Mumias and Teso Districts, Kenya (solid). (B) Exit
interviews with anti-malarial purchasers for children under five in three districts of Tanzania (dashed) and five districts of Uganda (solid). Studies
in (A) represent the combined private and public sectors (B) received them in private outlets only.

through private but accredited drug dispensing outlets. Under
the subsidy, use of ACTs in the private sector increased from 0%
of anti-malarial drugs purchased for children under five to 10%
almost a year after subsidy implementation (Simba and Kakoko
2012). In Kenya, a subsidy was launched in 2007 to make ACTs
available to Community and Family Wellness shops in three
rural districts. In 2006, household surveys found that 15% of
patients receiving anti-malarials in either the private shops or
public health facilities received ACTs. After 15 months of the
intervention, this fraction increased to 42%, with 9% of those
receiving ACTs obtaining them from the subsidized shops
(Sabot et al. 2009a,b).
The AMFm
The ACTwatch Group collected comparable data on ACT use in
Madagascar, Nigeria and Uganda, with surveys conducted
1–2 years before implementation of the subsidy and again
after almost 2 years of subsidization in Nigeria and Madagascar
and 1 year in Uganda (Figure 3) (ACTwatch Group and
ABMS/Benin et al. 2012). In addition, reports were identified
reporting ACT use in Ghana, Tanzania, Uganda and Madagascar
following implementation of the AMFm (Quakyi et al. 2012;
P. Yadav et al., unpublished data). All reports described
increases in ACT use after implementation of the subsidy. At
the time of analysis, no reports were available on the subsidy’s
effect in Zanzibar, Kenya or Niger.
The increases in use observed by ACTwatch household
surveys of children under five in Madagascar, Nigeria and
Uganda varied within SES groups and within urban and rural
strata between baseline and endline survey rounds (Figure 4).
These classifications reflect categorizations made by the
ACTwatch Group in their reports on each country (ACTwatch
Group and ABMS/Benin et al. 2012).
Relationships between ACT use, stocking and price
Figure 5 depicts all reviewed reports that presented data on
both the fraction of private sector outlets found to be stocking
ACTs and the price of the ACTs in those locations at proximal
time periods. Accounting for correlation between outlets
measured in the same survey, the price of ACTs for any age
group was negatively but not significantly associated with ACT
stocking (OR ¼ 0.870, Z ¼ 1.66, P ¼ 0.098). However, each $1
increase in the price of a paediatric ACT was associated with an
OR for stocking of 0.223 (Z ¼ 2.62, P ¼ 0.009). Ignoring
correlation, linear regression output suggests that this $1
increase in price is associated with a 3.49% decrease in the
fraction of outlets stocking any ACT (R2 ¼ 0.051), and a 23.5%
decrease in the fraction of outlets stocking a paediatric ACT
specifically (R2 ¼ 0.532).
In Figure 6, all reviewed studies are depicted that recorded
data on both ACT stocking in the private sector and their use in
the same regions at proximal time periods. A repeated measures
regression model suggests that, on average, each 10% increase
in stocking is associated with an OR for ACT use of 1.27
(Z ¼ 3.82, P < 0.001) among all ages. For children, the OR is
1.46 (Z ¼ 3.63, P < 0.001). Ignoring correlation between meas-
urements from the same surveys, this 10% increase in stocking
is associated with a 5% increase in the fraction using ACTs
(R2 ¼ 0.363) among all ages, and an 8% increase in the fraction
of children using ACTs (R2 ¼ 0.462).
Finally, a negative association was found between mean ACT
price and use in the same regions (Figure 7). The overall trend
was strongly influenced by a single outlier from Madagascar,
where price was >5 dollars. Excluding this point from analysis
and accounting for correlation, on average a $1 increase in ACT
price was associated with an OR for use of 0.30 (Z ¼ 3.02,
P < 0.001) among all ages, and 0.153 (Z ¼ 6.29, P < 0.001)
among children. Ignoring correlation and again excluding the
6 HEALTH POLICY AND PLANNING

Downloaded from http://heapol.oxfordjournals.org/ by guest on March 16, 2014

Figure 3 Of those obtaining anti-malarials from the private sector, the proportion that obtained an ACT by country before and during
implementation of the AMFm. Data are from ACTwatch surveys in Madagascar, Nigeria, and Uganda and contemporary surveys in Tanzania and
Ghana. The timing of each survey with respect to the beginning of the ACT subsidy is given in months in the numbers above each bar. All results are
from household surveys except in Tanzania and Ghana, where figures represent the results of exit interviews with anti-malarial purchasers.

A B

Figure 4 Changes in ACT use in AMFm countries by rurality and SES. (A) Increases in use of ACTs in the private sector by children under five in
the lowest and highest SES groups reported in ACTwatch household surveys before and after implementation of the AMFm in Madagascar, Nigeria
and Uganda. (B) Increases in use of ACTs by children under five in urban and rural areas as reported in ACTwatch household surveys before and
after implementation of the AMFm in Madagascar, Nigeria and Uganda.

outlier, a $1 decrease in price was associated with a 24 percent-
age point increase in ACT use among those of any age
(R2 ¼ 0.302). Restricting only to surveys measuring price for
paediatric drugs and use by children under five, a $1 decrease
in ACT price was associated with a 32 percentage point increase
in ACT use (R2 ¼ 0.594).
Discussion
The results of this review suggest that reducing the retail price
of ACTs can successfully increase ACT use, although evidence
from rigorous studies remains scant. Available results from
experimental studies, programmatic subsidies and the AMFm
all consistently demonstrate a negative association between
ACT use and price. In most non-experimental examples, the
lack of controls for comparison means it is impossible to assess
the attributable impact precisely, but in the absence of other
major interventions aimed at increasing use among private
sector treatment-seekers, it is likely that subsidy implementa-
tion plays an important role.
The influence of subsidies on ACT use appears consistent
within subgroups such as children under five and more remote
and poorer populations. Contrary to some stated concerns
(Oxfam 2012), this finding suggests that ACT price reductions
may have substantial impact on populations that have been
 SUBSIDIES INCREASE ACT USE: A SYSTEMATIC REVIEW
Figure 5 ACT prices and stocking in the private sector. The dashed
curve line is the logistic trend for all ages, and the solid line is the
logistic trend for under-fives only. Both lines represent the trend
without the outlier point, a region in Madagascar with high-ACT
stocking and prices.
Figure 6 ACT stocking and use in the private sector. Relationship
between stocking of ACTs in private sector or outlets and the fraction of
anti-malarial users that receive ACTs, by all ages and by children under
five only. The dashed line is the trend for all ages, and the solid line is
the trend for under-fives only.
identified as being most at risk of malaria mortality (Barat et al.
2004). With as many as 60% of young children in areas with
the highest burden of malaria obtaining treatment through
private channels (Cohen et al. 2012a,b), these findings suggest
that subsidies may be a useful tool for increasing overall ACT
coverage among vulnerable populations not currently reached
by the public health system.
This review empirically demonstrates significant relationships
between lower ACT prices and their increased availability at
retail outlets, between increased ACT availability and a higher
fraction of consumers who receive them, and, transitively,
7
Downloaded from http://heapol.oxfordjournals.org/ by guest on March 16, 2014
Figure 7 ACT prices and use in the private sector. The dashed line is
the logistic curve for all ages, and the solid line is the logistic trend for
children under 5 years of age. Both trend lines do not include the high-
priced region of Madagascar.
between lower ACT prices and their increased use. It is not
possible to derive causality from these associations, and it is
possible that increased stocking of ACTs is driven by greater
consumer demand, or that price reductions are driven by
increased availability of the drugs at competing outlets.
Nevertheless, these associations, in combination with the
available evidence from controlled experimental studies and
historical experience, should suggest to policymakers that
subsidizing ACTs is likely to result in a higher fraction of
treatment seekers receiving them.
This analysis did not examine whether increases in ACT use
were specifically among individuals with true malaria infection,
although understanding this relationship will be critical for
evaluating the true public health significance of increased ACT
use (Cohen et al. 2012a,b). One of the primary concerns with
implementing a subsidy is that low-cost ACTs may be given
to patients whose symptoms are not truly caused by malaria
infection (Moon et al. 2009; Kamal-Yanni et al. 2012). To
mitigate this risk, ACT subsidies might be considered along
with other interventions such as distribution of subsidized
diagnostic tests, as has been conducted in Cambodia (Yeung
et al. 2011). Alternatively, ACT subsidies may be reserved for
places where nearly all treatment seekers have a high prob-
ability of being infected with malaria (Cohen et al. 2012a,b).
It will be important for policymakers to weigh the costs
and benefits of implementing a subsidy against those of other
potential interventions, particularly in light of constrained
resources for malaria programmes (WHO 2013). The results of
this investigation provide an initial basis for evaluating the
impact of subsidizing ACTs on the fraction of suspected malaria
cases who receive them. Results, though based on a limited
evidence base, suggest that a 1 dollar reduction in ACT price
may increase the fraction of suspected malaria cases seeking
treatment in the private sector who receive ACTs instead of
other drugs by 24 percentage points. Evaluating whether such
an investment is worthwhile requires weighing context-specific
factors including the fraction of suspected malaria cases who
8 HEALTH POLICY AND PLANNING
treat outside the formal health sector and the prevalence of
malaria infection in this population (Cohen et al. 2012a,b).
This review has important limitations. Each of the types of
studies reviewed has specific weaknesses: the experimental
studies were in limited geographies and controlled environ-
ments that may not be fully representative of programmatic
conditions, while the programmatic findings tended not to have
controls nor proximal baselines, making it impossible to
understand the counterfactual scenario of how ACT usage
might have changed in the absence of a subsidy. Several of the
AMFm evaluations included in the analysis were conducted
shortly after the implementation of the subsidy, making it
difficult to ascertain the total magnitude of the effect of
the intervention. In some cases, such as Uganda, the imple-
mentation of the subsidy was significantly delayed, and the
evaluation period covered just 11 months. Finally, treatment
outcomes are the product of a complex series of factors
including social norms, supply chains and interactions between
sellers and customers, and it is likely that results of a given
subsidy intervention will vary significantly from place to place.
Despite these limitations, the preponderance of evidence
reviewed here is suggestive that subsidies are likely an effective
tool available to policymakers for increasing ACT use outside
the formal public health sector.
The Roll Back Malaria Partnership has set a global target of
ensuring 100% of confirmed cases receive effective malaria
treatment through the private sector by 2015. This review of the
available literature suggests ACT subsidies may be an effective
tool for equitable progress towards that goal. These results
provide an initial foundation for local and global policymakers
to make evidence-based decisions regarding whether to imple-
ment future ACT price subsidies for reduction of malaria
mortality in appropriate contexts.
Supplementary Data
Supplementary data are available at HEAPOL online.
FUNDING
This work was supported by the Bill and Melinda Gates
Foundation.
Acknowledgements
The authors would like to thank ACTwatch Group (Population
Services International), in particular Tanya Shewchuk and
Stephen Poyer, as well as Catherine Goodman, Kara Hanson
and Prashant Yadav for reviewing earlier versions of this paper.
References
ACTwatch Group and ABMS/Benin, ASF/DRC, PSI/Madagascar et al.
2012. Household Survey Report Benin, Democratic Republic of the Congo,
Madagascar, Nigeria, Uganda and Zambia.
ACTwatch Group, Population Services International. 2009. Household
Survey Report Kingdom of Cambodia. http://www.actwatch.info/
countries/evidence.php?id_country¼29. accessed 22 October 2012.
ACTwatch Group, Population Services International. 2011. Household
Survey Report Kingdom of Cambodia. Nairobi, Kenya: Population
Services Internationalhttp://www.actwatch.info/countries/evidence.
php?id_country¼29, accessed 22 October 2012.
Arnold F, Ye Y, Ren R et al. 2012. Independent Evaluation of the
Affordable Medicines Facility—malaria (AMFm). Multi-country
Independent Evaluation Report: Final Report. Calverton, MD: ICF
International.
Barat LM, Palmer N, Basu S et al. 2004. Do malaria control interventions
reach the poor? A view through the equity lens. American Journal of
Tropical Medicine and Hygiene 71: 174–8.
Boerma JT, Black RE, Sommerfelt AE, Rutstein SO, Bicego GT. 1991.
Accuracy and completeness of mothers’ recall of diarrhoea occur-
rence in pre-school children in Demographic and Health Surveys.
International Journal of Epidemiology 20: 1073–80.
Cohen J, Dupas P, Shaner S. 2012a. Price Subsidies, Diagnostic Tests, and
Targeting of Malaria Treatment: Evidence from a Randomized Controlled
Trial. [Manuscript in progress]. http://www.povertyactionlab.org/pub-
lication/price-subsidies-diagnostic-tests-and-targeting-malaria-
Downloaded from http://heapol.oxfordjournals.org/ by guest on March 16, 2014
treatment-evidence-randomized-con, accessed 3 June 2013.
Cohen J, Woolsey A, Sabot O et al. 2012b. Optimizing investments in
malaria treatment and diagnosis. Science 338: 612–4.
Cohen JM, Sabot O, Sabot K et al. 2010. A pharmacy too far? Equity and
spatial distribution of outcomes in the delivery of subsidized
artemisinin-based combination therapies through private drug
shops. BMC Health Services Research 10(Suppl. 1):S6.
Davies H, Crombie I, Tavakoli M. 1998. When can odds ratios mislead?
BMJ 316: 989.
ICF Macro and London School of Hygiene and Tropical Medicine. 2012.
Independent Evaluation of Phase 1 of the Affordable Medicines Facility—
malaria (AMFm). http://www.theglobalfund.org/en/amfm/indepen
dentevaluation/, accessed 22 May 2013.
Institute of Medicine. 2004. Saving Lives, Buying Time: Economics of Malaria
Drugs in an Age of Resistance, National Academies Press, 388p.
Kamal-Yanni M, Potet J, Saunders P. 2012. Scaling-up malaria treat-
ment: a review of the performance of different providers. Malaria
Journal 11: 414.
Kangwana BP, Kedenge SV, Noor AM et al. 2011. The impact of retail-
sector delivery of artemether-lumefantrine on malaria treatment of
children under five in Kenya: a cluster randomized controlled trial.
PLoS Medicine 8: 5.
Liberati A, Altman D, Tetzlaff J et al. 2009. The PRISMA statement for
reporting systematic reviews and meta-analyses of studies that
evaluate healthcare interventions: explanation and elaboration.
BMJ 339: b2700.
Moon S, Perez Casas C, Kindermans JM, de Smet M, von Schoen-
Angerer T. 2009. Focusing on quality patient care in the new global
subsidy for malaria medicines. PLoS Medicine 6: e1000106.
Oxfam. 2012. Salt, Sugar, and Malaria Pills: How the Affordable Medicine
Facility—Malaria Endangers Public Health. Oxfam Briefing Paper. http://
www.oxfam.org/sites/www.oxfam.org/files/bp163-affordable-medi
cine-facility-malaria-241012-en.pdf, accessed 24 October 2012.
Population Services International. 2006. Cambodia Malaria TRaC Survey
among Populations Living in Malaria Endemic Areas. Washington, DC:
Population Services International Research Division.
Quakyi I, Winch P, Adjei A, Sullivan D. 2012. Summary Report: Improving
Correct Diagnosis and Treatment of Malaria in Rural Ghana. Baltimore,
MD: Johns Hopkins School of Public Health.
Roll Back Malaria Partnership. 2011. GMAP—the Global Malaria Action
Plan. Refined/Updated GMAP Objectives, Targets, Milestones and Priorities
Beyond 2011. http://www.rbm.who.int/gmap/gmap2011update.pdf,
accessed 24 October 2012.
 SUBSIDIES INCREASE ACT USE: A SYSTEMATIC REVIEW
Sabot O, Yeung S, Pagnoni F et al. 2009a. Distribution of artemisinin-
based combination therapies through private sector channels:
lessons from four country case studies. Resources for the Future
Discussion Paper: Prepared for the Consultative Forum on the
AMFm—the Affordable Medicines Facility—malaria, September
27–28, 2008, Washington, D.C. http://www.rff.org/RFF/Documents/
RFF-DP-08-43_FINAL.pdf, accessed 7 September 2012.
Sabot OJ, Mwita A, Cohen JM et al. 2009b. Piloting the global subsidy:
the impact of subsidized artemisinin-based combination therapies
distributed through private drug shops in rural Tanzania. PLoS One 4:
e6857.
Schäferhoff M, Yamey G. 2011. Estimating Benchmarks of Success in the
Affordable Medicines Facility—Malaria Phase 1. San Francisco:
Evidence-to-Policy Initiative. http://www.theglobalfund.org/docu
ments/amfm/E2PI_EstimatingBenchmarksInAMFm_Report_en/,
accessed 7 September 2012.
Simba D, Kakoko D. 2012. Access to subsidized artemether-lumefantrine
from the private sector among febrile children in rural setting in
Kilosa, Tanzania. Tanzania Journal of Health Research 14: 2.
9
Talisuna A, Daumerie PG, Balyeku A et al. 2012. Closing the access
barrier for effective anti-malarials in the private sector in rural
Uganda: consortium for ACT private sector subsidy (CAPSS) pilot
study. Malaria Journal 11: 356.
White NJ. 2008. Qinghaosu (Artemisinin): the price of success. Science
320: 330–4.
World Health Organization. 2010. Guidelines for Treatment of Malaria—
Second edition. http://www.who.int/malaria/publications/atoz/
9789241547925/en/index.html, accessed 15 October 2012.
World Health Organization. 2011. World Malaria Report 2011. http://www.who.
int/malaria/world_malaria_report_2011/en/, accessed 7 September
2012.
World Health Organization. 2013. Financing Malaria Control—Allocating
Limited Resources. MPAC Presentation. http://www.who.int/malaria/
mpac/resource_allocation_mpac_presentation_march_2013.pdf,
accessed 20 May 2013.
Yeung S, Patouillard E, Allen H, Socheat D. 2011. Socially-marketed
rapid diagnostic tests and ACT in the private sector: ten years of
experience in Cambodia. Malaria Journal 10: 243.
Downloaded from http://heapol.oxfordjournals.org/ by guest on March 16, 2014