Background
In many low-income countries, private commercial providers play an important role in the provision of malaria treatment [
1‐
6]. Patient choice of private over public sector providers is influenced by several factors including the wider range of drugs stocked in private outlets, proximity to patients’ homes, less frequent stock outs and greater responsiveness to consumer preferences. However, the quality of care they provide is often poor, with commercial providers often lacking relevant qualifications and adequate knowledge of malaria national treatment guidelines [
7,
8].
One of the cornerstones of malaria control is parasitological confirmation of all suspected malaria cases by either microscopy or rapid diagnostic tests (RDTs) and treatment of confirmed Plasmodium falciparum cases only with artemisinin-based combination therapy (ACT). By 2009, most countries with P. falciparum had switched to ACT as their first-line medicine, with the choice of combination drugs based on their efficacy in specific countries.
The therapeutic life of ACT is, however, threatened by resistance to artemisinins confirmed on the Cambodia-Thailand border and more recently detected in two other countries in the Greater Mekong subregion, Myanmar and Vietnam [
9]. This is of major concern to the international community, particularly as resistance may spread to countries with much higher disease burdens in Africa, as has been the case in the past for older anti-malarial drugs [
10‐
14]. The loss of ACT to resistance would be catastrophic for malaria control strategies as no other treatment with the same efficacy and tolerability is currently available [
15].
Containing resistance to areas where it exists is, therefore, paramount. Factors believed to encourage the emergence and spread of anti-malarial resistance include lack of confirmed diagnosis through blood testing, the consumption of oral artemisinin monotherapies (AMT), poor consumer adherence to treatment regimens and poor quality drugs [
16,
17]. Increasing access to parasitological diagnosis of all cases with suspected malaria and rational treatment with affordable ACT for
P. falciparum confirmed cases, including in the private sector, are key recommended responses for the containment of artemisinin resistance [
9]. Complementary actions include the removal of oral AMT and substandard and counterfeit drugs from the market, and educative communication campaigns targeting consumers and providers [
18].
Cambodia finds itself at the heart of these issues, given its malaria burden, resistance profile and large private sector. Malaria transmission is seasonal (from May to December) and is concentrated in forested areas. Around 15 % of the population (2.2 m people) is at risk of infection, predominantly male adults. The national malaria treatment policy specifies that all people with suspected malaria should first receive a blood test, either using microscopy or RDTs [
19]. At the time of the study in 2009, first-line treatment of confirmed uncomplicated adult
P. falciparum cases consisted of the ACT artesunate and mefloquine (ASMQ) for three days while first-line treatment of
Plasmodium vivax cases was chloroquine, also for three days [
19]. The sale and distribution of oral AMT had been prohibited. In 2009, the country could be roughly divided into two strata: the Western part where resistance was confirmed or suspected, and the North and Northeast, where
P. falciparum remained susceptible to ACT.
In 2009, the commercial sector supplied about 75 % of all malaria treatments in Cambodia [
20,
21]. Commercial anti-malarial providers included pharmacies or clinical pharmacies, drug shops, mobile providers, grocery stores and village shops. ACT and RDTs have been the focus of a social marketing programme since 2002, which includes consumer behaviour change communication, provider training, and the sale of subsidized ACT and RDTs. In 2009, one adult pack of ACT was sold to commercial providers for US$0.42 with a recommended retail consumer price (RRP) of US$0.61 printed on each pack. In the public sector, parasitological diagnosis and anti-malarials were available free of charge from government-owned outlets, including health centres and hospitals and a network of trained volunteers (referred to as Village Malaria Workers (VMWs)) [
20] in areas with high malaria risk and with no access to health facilities.
Despite these initiatives, evidence shows that in 2009 clinical practices rarely followed national treatment guidelines and that the quality of fever management was often low both at public and private outlets [
20]. Overall, only 47 % of positive malaria cases received an anti-malarial and 35 % an ACT [
22] whilst AMT, which are banned, accounted for 15 % of all anti-malarials dispensed [
20]. “Cocktail therapies”, a mix of several different drugs, were the most common treatment for malaria fevers received in the commercial sector, with 47 % of respondents receiving a cocktail containing no anti-malarial and 11 % a cocktail with an anti-malarial drug [
20]. These poor treatment practices are related to ACT availability and consumers’ perceptions of product quality. Whilst knowledge about the recommended first line treatment for
P. falciparum was generally high amongst anti-malarial drug sellers [
20], ACT availability was low, ranging between 1 % and 49 % across different commercial provider types [
20]. It has been reported that provision of cocktail treatments may be perceived by consumers as reflecting provider knowledge and expertise [
21,
23] whilst oral AMT may be seen as causing fewer side effects than the ACT ASMQ [
21,
24].
Despite the subsidy programme, retail price is also still likely to have been an important determinant of ACT uptake. For instance, in 2009 the median retail price of one ACT adult equivalent treatment dose (AETD) was US$1.18, roughly two times the RRP. The structure of the market for anti-malarial drugs is likely to have an important impact on private sector prices. A market is commonly defined as the set of sellers and buyers of a good or service whose interactions determine the quantity, quality and price of the good or service [
25]. Market structure is defined as the characteristics of the market, including the range of sellers and products available, the number of sellers and their relative importance in terms of sales volumes and values (also referred as the degree of market concentration), the presence and intensity of barriers to market entry and exit, the structure of the retail distribution chain and the regulatory system. The structure-conduct-performance (SCP) framework predicts that the structure of the market determines the way firms behave which, in turn, affects market performance [
26]. In particular, providers facing weak competition may have the opportunity to raise their prices above levels seen in more competitive markets. Weak competition is traditionally characterized by more concentrated markets, where fewer providers account for larger market shares [
27]. This is compounded if there is limited competition from potential new providers (termed limited contestability), due to high barriers to entering the market such as costly retail outlet license fees.
Few studies have investigated the association between market structure and price setting decisions in health care markets in low-income countries [
28‐
33]. One study in rural Tanzania found that higher prices were statistically associated with higher market concentration [
32]. The aim of this paper is to analyse the determinants of retailers’ price setting decisions for anti-malarial drugs in Cambodia, and draw policy recommendations for improved access to affordable quality malaria treatment and resistance containment.
Discussion
This is the first study to investigate the market for malaria treatment in Cambodia using an economic lens, by exploring the determinants of commercial providers’ pricing decisions for anti-malarial drugs. For the interpretation of the study results, a few issues should be taken into account. The analysis of competition requires the difficult task of defining markets on product and geographic dimensions. Outlet census and drug audit data were used to set the product definition as all anti-malarial drug types stocked by all outlets selling anti-malarials. Using the administrative boundary approach, the geographical definition was set as the commune. Ideally household data on treatment seeking behaviour (location and type of provider visited and drugs obtained) would be used for defining the product and geographic definition of the markets, but these data were not available at the time of our study. The definition of markets was therefore validated through qualitative interviews with commercial providers during which information on the provenance of their customers and location of competitors were collected. It is possible that retailers overestimated the degree of competition they faced, as has been observed elsewhere [
32], notably from government providers, and that the geographical size of markets was overestimated in the study. However, in the context of Cambodia, where mobile providers represented an important source of anti-malarial drugs [
20], it is most likely that a narrower definition of the market boundaries, such as the village, could have underestimated the size of retail markets.
Market concentration measures included both private and public anti-malarial sales, with all government outlets within each market treated as one provider on the basis that government providers were not expected to compete with one another. It is possible that this analytical approach distorted market concentration measures that were later used in the analysis of retail mark-ups, by masking the impact of the relative importance of private sales volumes on private retailers’ mark-ups. This could have been one reason for the difference in the nature of the relationship between mark-ups and concentration in markets at different accessibility levels. However, qualitative interviews with commercial providers highlighted the competitive pressure from VMWs in areas where they operated. Therefore, it was deemed important to include the volumes dispensed at government outlets in the analysis of market concentration as excluding them might have underrepresented the degree of retail competition. This analytical approach was also validated by estimating a model of mark-ups excluding government sales and found the same results as in the models presented here. Differences in wealth across markets may also have influenced retailers’ pricing decisions as, for example, they may have charged higher prices in settings where customers had greater willingness-to-pay. Data on household socio-economic status were not available for these markets at the time of analysis, but the accessibility measure may to some degree have captured this as the accessibility of markets would be expected to be positively associated with household wealth.
It is possible that commercial providers underreported stocking of banned anti-malarial drugs such as AMT, or deliberately misrepresented their pricing behaviour, perhaps by reporting lower selling prices, as they would not want to be seen as making excessive profits, which would have led to an underestimation of mark-ups. Also, the percent mark-ups measured gross margins, which include other costs of selling, and therefore higher mark-ups may not have been a sign of higher net profit margins.
Finally, commercial providers’ price setting behaviour and the extent of price competition in retail markets were analysed by following the traditional SCP approach, investigating whether more concentrated markets were associated with higher mark-ups. Limitations of this approach that hypothesizes a causal link between structure and conduct, combined with that of using HHI as the measure of market competition, are widely recognized in the industrial organization literature [
39,
40]. Market concentration as measured by the HHI may be endogenous, that is the result of providers’ pricing decision, demand or cost factors [
30,
41]. For example, firms that can produce at a lower cost have the ability to charge lower prices and are therefore likely to have higher market shares than firms with higher costs.
The study assessed the relationship between percent mark-ups for anti-malarials, market concentration measured by the HHI and a set of other market, product and outlet characteristics. In Model 1, no evidence of a significant statistical relationship between anti-malarial mark-ups and the HHI was found, which from a traditional economic theory perspective was surprising. Given the heterogeneity of the markets under study in terms of concentration, the relationship between anti-malarial mark-ups and the HHI in markets at different accessibility and malaria transmission risk levels was explored in Model 2. The expected positive association between HHI and mark-ups was observed in moderately accessible markets only. In these markets, a higher concentration may have indicated less intense competition, offering commercial providers the opportunity to charge higher mark-ups. By contrast, in remote markets an increase in the HHI led to lower mark-ups. One explanation for this difference may be that market concentration was not capturing variations in competitive pressure. With most (82 %) remote markets experiencing moderate or high risk of malaria transmission, lower mark-ups in more concentrated markets may have indicated the competitive pressure of VMWs’ supply of free treatment in areas where they operated, leading private providers to limit their pricing. It is also plausible that retailers operating in hard-to-reach, less populated communes at higher risk of malaria transmission may have put greater emphasis on socially oriented objectives relative to profit maximization. Higher mark-ups in less concentrated remote markets may also have reflected higher costs, for example the costs of transporting anti-malarial supplies. Overall, whilst the price of ACT was set above the recommended retail price, market concentration and mark-ups were not positively associated in most markets. Inadequate ACT coverage for malaria treatment in the private sector was therefore unlikely to primarily reflect excessive retail mark-ups on ACT, and instead reflect other key elements of medicine demand and supply.
Even at a subsidized price, ACT wholesale prices were much higher than those of non-artemisinin monotherapies [
21], meaning that even with relatively competitive mark-ups, ACT retail prices were more than two times higher, and therefore may have been beyond the reach of some consumers. Furthermore, while AMT were sold at substantially higher prices than ACT, it may be common for consumers to purchase incomplete doses of AMT as observed elsewhere [
42], implying that the price of one adult equivalent treatment dose did not reflect what consumers usually paid [
20]. The negative externality of increased pressure for the development of drug resistance caused by AMT use, particularly at low doses, would not be considered by consumers. Also, as noted in the introduction, purchase of drug “cocktails” is common in Cambodia for treating illnesses, including malaria and fevers in general [
23]. Although these mixtures of tablets may not contain any anti-malarial, they are often seen as more effective than pre-packaged medicines at treating multiple symptoms [
23] and may be perceived by consumers as a substitute for ACT. Overall, it is likely that information on the relative efficacy of ACT was far from perfect among consumers.
The study results indicate that provider pricing may have played an important role in the performance of the market for malaria treatment in Cambodia. At the time of the study, the subsidized ACT wholesale price was US$0.42, which was much higher than the wholesale price of chloroquine (US$0.08), the most popular monotherapy [
20,
21]. There may therefore be potential to decrease ACT wholesale and retail prices by applying larger subsidies, such as those provided through the Affordable Medicines Facility – malaria (AMFm). The AMFm implemented in seven African countries made ACT available at a highly subsidized price at the manufacturer level [
43], representing 1–20 % of the ex-factory prices [
44]. After the introduction of the AMFm, large increases in ACT availability and large decreases in ACT median retail prices were observed in the private sector of most, though not all participating countries [
44]. The pooled purchasing power within the AMFm programme also facilitated price negotiations to reduce manufacturer ACT prices [
45]. Cambodia was initially included in the group of countries eligible to participate in the Phase 1 of the AMFm pilot but later excluded due to the lack of a first line drug complying with the quality assurance policy of the Global Fund to Fight AIDS, Tuberculosis and Malaria [
46]. With a quality approved drug now available for Cambodia, an AMFm-like subsidy may have the potential to decrease wholesale and retail prices as observed in other settings.
Experiences in using RRPs on anti-malarial drugs have proved to have different effects on consumer prices across settings [
47‐
49]. In Cambodia, since the start of the national programme in 2001, the subsidized ACT always had a printed RRP though retail prices generally exceeded it [
24,
47]. Little research has been conducted in Cambodia on the level at which RRP should be set. At the time of our study, the RRP for ACT was US$ 0.61, based on a study on consumers’ willingness to pay conducted some years previously [
50]. During qualitative interviews conducted in 2009, several private retailers argued that the RRP was set too low and did not provide sufficient profits [
21]. Generating evidence on commercial providers’ overhead costs including transport, rent, staff, etc. and their relative importance should be considered when investigating the levels of profit perceived to be “sufficient” by private providers. However, during qualitative interviews, some retailers in Cambodia reported that when consumers were informed about the RRP, they were constrained to sell the subsidized ACT at that recommended price [
21], supporting the use of RRP combined with effective communications to consumers. A similar observation can be made from the experiences of countries included in the AMFm pilot. In the three countries in which timely communication campaigns about the AMFm subsidy and RRP were implemented, including Kenya, the Republic of Tanzania (mainland and Zanzibar) and Ghana, median ACT prices were at the RRP levels. By contrast, in countries where promotion activities were delayed or less intense, the median ACT price was above the RRP [
49].
Even with lower wholesale prices, retailers may be reluctant to stock ACT if they expect consumer demand to be low because of lack of information on what constitutes appropriate quality malaria treatment. Expanding and intensifying social marketing-like activities, including road shows, radio messages and TV ads combined with increased communication on the regulatory framework could improve consumer information on the need for confirmed diagnosis prior treatment, ACT efficacy compared to older anti-malarial drugs, the potential dangers of cocktail therapies and the negative social externalities of AMT in terms of drug resistance. These activities could also be expanded to non-malarial areas in the context of mobile populations with no or little immunity to malaria who travel to forested areas. Finally, active and supportive regulation has a key role to play in improving the performance of the market for malaria treatment, through measures to stop the importation and distribution of oral AMT and the marketing of cocktail therapies, and through supportive and educative activities with private providers of malaria treatment.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
EP designed the study, conducted the data analysis and wrote the manuscript. KOC and SP managed the data collection process of the outlet survey data. CG and KH provided guidance on all aspects of the study and contributed to the write-up of the paper. KOC, IK, ST, BP contributed to the study design and data analysis. All authors read and approved the final manuscript.