The private retail sector plays a central role in the provision of malaria treatment in Tanzania. In rural areas, 68% of the population live within 5 km of a health centre or a dispensary (98% in urban areas) . Yet, poor quality of care, shortage of skilled providers, stock-outs of essential drugs, and long waiting times [16, 17] are challenges which may drive patients to seek care (or at least buy drugs) from more expensive non-governmental facilities, or from drug stores. The private retail sector may complement health facility services where the facilities are unable to deliver .
In the studies presented here, drug stores were more frequently visited for drugs than general shops. In an earlier study, general shops have been described as being important treatment sources for fever/malaria, with 29% of fever cases using this source of treatment. Yet, in terms of drug volumes, general shops accounted for only 6–7% of all antimalarial doses dispensed in the two study districts . However, general shops are important first contact points of patients with a network of treatment providers. They are numerous even in small villages and often more easily accessible than drug stores or health facilities . While not being legally allowed to dispense antimalarial drugs, they are recognised in the national policy as one component of the health care delivery structure [25, 26]. Yet, their relatively poor knowledge of malaria and its appropriate treatment supports the ban of antimalarial drugs from these outlets. Surprisingly, only 60% of general shopkeepers mentioned homa (fever) as a symptom of malaria. In part, this may be explained by the parallel use of homa as a term to describe a less severe febrile illness or general malaise [27, 28]. Knowing the correct treatment was clearly a function of the shopkeeper's education, which in general shops was lower than in drug stores. However, general shopkeepers did not seem to be completely unaware of their limitations, as 34% of them mentioned referral to a health facility as the correct action for a child with malaria.
Drug stores on the other hand are the lowest level of providers which is generally tolerated to dispense prescription-only antimalarial drugs. Unfortunately, they often do not reach out into small villages or remote areas . Shopkeepers in drug stores were more knowledgeable about malaria-related symptoms and malaria treatment than their counterparts in general shops. This was correlated with basic medical or health-related training, a prerequisite for shopkeepers of licensed part II drug stores . Nevertheless, their performance was not satisfactory, with only 52% mentioning SP in the correct dosage as recommended treatment for children.
Knowledge vs. practice
In order to get a realistic picture of drug-sellers' performance, we used mystery shoppers; an approach which has been applied frequently in market research, but rarely in a public health context [30, 31]. The main challenge of applying this methodology in a rural setting, which is to find capable mystery shoppers within a certain village, was tackled with the help of knowledgeable village-based DSS field staff.
Daily shopkeepers' practices clearly reflected their level of understanding of appropriate treatment, the current drug regulations, as well as the low antimalarial availability in general shops . Antipyretics were frequently sold in both, drug stores and general shops. Most drug stores (88%) also sold antimalarials to the mystery shoppers. In contrast, during a study conducted elsewhere in Tanzania in which shopkeepers were under observation, only 17% of febrile patients had received an antimalarial . In general shops, 19% of the mystery shoppers were sold an antimalarial, which was more than expected based on the shop census in which 8% of all general shops that had drugs in stock also stocked an antimalarial .
While many shopkeepers in drug stores knew that SP was the recommended treatment for children and adults, in practise, amodiaquine and quinine were sold as often as SP. This may to some extent reflect that amodiaquine was slightly more readily available in drug stores and, according to anecdotal evidence, quinine was popular as it was often regarded a strong and powerful medicine . Overall, it was more likely that a mystery shopper received an antimalarial or even SP in a drug store. However, drug stores did not adhere better to the guidelines than general shops. In part, this may be attributed to the larger choice of products in drug stores. Mere non-availability may also be a reason why no other antimalarials than SP, amodiaquine and quinine were sold, along with the fact that with the cash provided by the researchers, the mystery shoppers would not have been able to purchase expensive drugs such as artemisinin mono therapies or ACT .
Altogether, adults would more readily be dispensed an antimalarial than children. This is interesting in the light of findings from a cross-sectional community-survey in which adults would be treated more frequently with shop bought drugs while children were more often brought to a health facility . This may give some indications of provider-side influences on treatment-seeking behaviour.
Treatments for adults were 25% cheaper than treatments obtained for very young children and drug stores were more expensive than general shops. The latter was also found in another study in the same area, where more expensive treatments were obtained from non-governmental organisation (NGO) facilities and drug stores, usually by people from the better-off socio-economic stratum .
Private retailers may commonly be perceived as being mainly business-driven in their behaviour. In this study we found that in theory, more than half of all shopkeepers said they would refer severely ill patients and general shopkeepers commonly regarded referral as best option for young children. In practice, 15% (3/20) of drug stores and 31% (25/82) of general shops did not sell any medicines but referred the simulated patients to a higher level of care – although they would have had drugs in their shops. The awareness of shopkeepers that certain cases need to be dealt with at a higher level may be a good entry point for interventions targeted at the retailer level. Several projects targeting private drug retailers, have already counted on the ability and willingness of shopkeepers to refer severe or complicated cases to an appropriate facility [13, 34].
Implications for policy and interventions
The importance of the retail sector as a source of malaria treatment and care complementary to health facility has been recognised internationally  and within Tanzania . However, the major concern regarding the private sector has been inadequacy of the treatments offered by often untrained (or not sufficiently trained) shopkeepers [3, 34, 36]. This issue has re-emerged in the discussions about appropriate delivery channels for ACTs. Defining the role of each type of retailer present in a health system within the frame of their capabilities and the given legal context is an important first step in improving quality and access.
Fully-fledged pharmacies only reach 17% of the Tanzanian population and are hence not sufficient to meet the demand for essential drugs . Part II drug stores which are the largest network of licensed drug-retailers in Tanzania  are licensed to sell only OTC drugs, to which none of the recommended antimalarials belongs. Kachur et al. showed that patients at drug stores are as likely to be infected with malaria as patients seeking care at health facilities . Considering this demand for antimalarial treatments, there is a need to make efficacious antimalarial drugs available in drug stores. In reality this is usually tolerated by the authorities who recognise the lack of alternatives. In order to improve the quality of services in drug stores, specialised training for drug vendors may be a valid option for improving management of malaria-cases, as has been shown in other areas . The mere definition of educational prerequisites as currently the case for part II shops may only lure health workers away from health facilities to a more profitable business in the retail sector. Yet, training alone is unlikely to improve performance if not coupled with appropriate means of rewarding the shopkeepers for good practices [36, 38]. These approaches are combined in a project that upgrades part II shops and potentially general shops to Accredited Drug Dispensing Outlets (ADDO) and that is currently being implemented in selected districts in Tanzania [13, 37].
The role of general shops should not be the dispensing of prescription medicines. Yet, due to their importance as easily accessible first contact point for malaria patients, they should not be completely left aside when targeting the private sector. There are several options to strengthen their role in the health sector. Firstly, they could be upgraded to drug retailers (e.g. ADDOs) if appropriately trained, thereby increasing the population coverage with antimalarial providers. Secondly, general shopkeepers could be trained on the appropriate first aid for malaria cases with OTC medicines and subsequent referral to a higher level. Considering that general shops may manage malaria cases only with antipyretics, particularly in places where they are the nearest provider, targeted information or training may decrease the number of inappropriately managed cases at the lowest level. The social pressure exerted on shopkeepers by communities' expectations on their performance should not be under-estimated. In our study, a considerable number of shopkeepers did without business in favour of referring the patient to a drug store or a health facility.
Including all levels of formal and informal health care providers is feasible within the existing legal framework and guided by the national malaria control policy. Alternative approaches including lowest level shops may be a step forward in improving access for people living in remote areas or deprived villages which so far lack any provider of antimalarial medicines .