Malaria kills more than one million people annually, mostly children under five years of age in sub-Saharan Africa. As part of an integrated approach to malaria control, the World Health Organization promotes prompt access to effective treatment for all episodes of malaria 1. However, it can be safely assumed that of the 2.4 billion people (2005) living in low-income countries 2, only few have access to high quality health-care, including appropriate malaria-treatment 345.

Increased attention to the issue of access to treatment is required in the light of rolling-out highly efficacious artemisinin-based combination therapy (ACT). Lack of access is a complex issue that can be better understood when seen in the context of poverty, vulnerability and livelihoods 6. Which obstacles to treatment are relevant – and should be addressed by interventions – depends to a large extent on the local setting 3. As there is no one-size-fits-all solution, several strategies have been proposed and tested to improve access to malaria treatment. These include scaling-up home-based management 78, stronger involvement of the private sector 910, improving case-management in health facilities 11 as well as integrated approaches 12. In addition, it is now widely acknowledged that no malaria-control strategy can be successful and sustainable without an increased investment in the local health system through which the interventions are to be channelled 131415.

In order to develop targeted and cost-effective interventions, it is essential to add to the knowledge of treatment rates an understanding of the major obstacles to treatment access in a particular setting. For Tanzania, the 2004–5 Demographic and Health Survey (DHS) reported 58.2% antimalarial use in children under 5 years of age with a recent fever 16. Yet, how many of these children received appropriate treatment, in a timely fashion and in the correct dose – or why others did not receive an antimalarial drug – is not elaborated upon in DHS surveys.

The studies presented here assessed the impact of local knowledge of malaria and treatment seeking practices on access to malaria treatment in a rural Tanzanian community. The research was carried out in the frame of a comprehensive intervention programme which aims to improve access to malaria treatment in rural Tanzania (ACCESS Programme) 17.

This paper assessed population coverage with prompt and effective malaria treatment from the perspective of those affected. The analysis revealed some important issues for the development of future malaria control strategies. However, in order to establish a comprehensive understanding of access to health, it may be useful to combine several approaches, as elaborated by Obrist et al. 6.

Local concepts of malaria will influence successful implementation of effective case-management 32. This research demonstrated clear changes in the understanding of degedege as compared to historical data. Degedege was traditionally linked to spirits in the form of a bird or a moth and mainly herbal treatment would be administered by traditional healers 2933. Studies done between 1995 and 1997 in the same area found that concepts of degedege and malaria were fuzzy; while degedege was sometimes seen as a cause of severe malaria, only mild malaria would be related to mosquito bites 2734. Today, fevers with convulsions were in most cases labelled malaria rather than degedege. Mosquito bites were usually seen as the cause for convulsions and degedege. These findings are currently being validated through an in-depth study on degedege cases. In contrast to degedege, which was considered a severe and dangerous illness 34, homa ("fever") was often regarded as a normal stage in child growth or was attributed to weather conditions. Similar findings have been reported from other regions of Tanzania 28. While for homa and malaria different symptoms were reported in children, this was not the case in adults. It also appeared that malaria was usually associated with mosquitoes whereas homa was often attributed to other causes. In earlier studies, homa has been described as a label for general malaise and aches, sometimes even in the absence of fever 3536. To some extent this may explain why fever as a symptom was not most prominent in all homa cases. Nevertheless, local illness labels may still influence treatment-seeking behaviour. As a result, those cases labelled homa were more than 12 times (OR = 0.08) less likely to receive an antimalarial than cases labelled malaria or degedege.

An increasing overlap of the popular and biomedical concepts of malaria can be attributed partly to regular and intensive health education campaigns in the area, from the national "Mtu ni Afya" (Man is Health) campaign in the late 1970s 26 to the intensive social marketing of insecticide-treated nets in the 1990s 2737. Nevertheless, factual knowledge does not necessarily translate directly into improved behaviour 34, particularly since appropriate care-seeking depends on several factors other than illness understanding.

In contrast to what has been reported from earlier studies and other areas 383940, no fever episode remained untreated. Antimalarial administration was common, even for cases with convulsions or labelled degedege and traditional herbal medicine use was rare. This is a major improvement in treatment seeking when compared to a household survey carried out in two DSS villages in 1995–97, which reported 35% of degedege episodes were treated with herbal medicine and only 2% received an antimalarial 26. It fits well with the findings of de Savigny et al. 41 who reported 78.7% of sufferers first use biomedical care for cases of fatal malaria in Tanzania. Considering that first treatment with herbal medicine was correlated with less antimalarial use (OR 0.08), discouraging herbal treatment may help to increase treatment rates with antimalarials.

The frequent usage of antipyretics as first treatment may partly reflect a decreased availability of antimalarials in shops 23 and the Tanzanian malaria control policy, which does not actively promote home-based management with antimalarials. Most of the children were brought to a health facility, which increased the chance of receiving a recommended antimalarial. Health facility attendance is generally desirable because other severe febrile illnesses, such as pneumonia or meningitis, can not easily be managed at home. Despite high health facility usage rates, however, the proportion of patients receiving an appropriate antimalarial timely and correctly dosed (23% of all children and 11% of all adults) was low and too far from the 80% target set by the RBM Partnership for 2010 42. These figures put into perspective the 51.1% use of a recommended antimalarial on the same or next day for Tanzania and 82.7% for Morogoro Region reported by the DHS 16. At the same time, the DHS figures also validate the results of some of our indicators, except for the rates of quinine use, which were extraordinarily high in our study.

Patients, particularly children, who were treated at a health facility, were not more likely to receive an appropriately dosed antimalarial. Reported wrong dosages may be attributed to poor patient adherence or reporting errors. Yet, without any doubt, quality of care is a critical step in assuring appropriate treatment. Merely limiting antimalarial sales to health facilities and drug stores 23 has apparently not resulted in an acceptable quality of case-management. The observed effect of the number of people in a household on (timely) antimalarial treatment may be a chance finding or related to intra-household relations 43. Yet it may also reflect that in larger families, child care is often delegated from parents to older siblings who may not yet know how to handle an illness episode. In families with more children (or elderly household members), caretakers may be more busy and have less time devote to each one.

Children under 5 years of age should be treated free of charge in (government) health facilities. Yet 25% of the child episodes seen at a health facility were eventually treated with a shop-bought antimalarial. One might expected that commercial sector treatment of childhood fevers would pose an additional and unnecessary burden on poor households since in the private sector no exemption policy applies. Njau et al. reported significantly higher spending in drug stores (but not general shops) compared to government health facilities 44. However, data from this study suggests that due to a dysfunctional exemption system drug expenditures did not significantly differ between health facilities and drug stores. In this situation, it may even be cheaper to treat a child with shop-bought drugs. A shorter distance to shops than health facilities may result in lower secondary costs for time and transport. This may consequently contribute to higher facility usage rates and possibly to more use of antimalarials by richer households as reported elsewhere 45. This link, however, could not be proven in our study.

Stock-outs of SP, amodiaquine, and quinine, which were repeatedly observed in the study area 46, may have forced certain patients to purchase medicines from shops. The commercial sector therefore plays a pivotal role in providing life-saving drugs in the case of delivery-failure of formal health facilities.

The key issues resulting from this research need to be interpreted in consideration of the research methodology and the study setting. Case-selection was based on reported fever rather than lab-confirmed malaria, acknowledging that due to a lack of diagnostics in most primary health facilities episodes of fever suggestive of malaria are treated as such. The Integrated Management of Childhood Illness (IMCI) algorithms (which are implemented in the study area), as well as the national malaria treatment guidelines advocate an assessment and treatment based on clinical signs and symptoms if no microscopy is available 22. Very often however, a diagnosis is based on the patient's or caretaker's reported symptoms rather than a proper clinical assessment 4748. The study generally relied on patients' or caretakers' accounts which may result in misreporting of symptoms and treatment seeking behaviour. However, in real life, it is also the patient's perceived ill-health that triggers help-seeking action, rather than a clinical or laboratory diagnosis 4.

When making inferences to other areas one should consider the limited sample size of the studies and the long history of malaria-control and research activities in the study area. The latter would suggest that treatment and coverage rates in the area are likely to be above the Tanzanian average.

It is also worth considering that all estimates presented here did explicitly exclude the aspect of drug efficacy. Yet, actual efficacy of SP in children in nearby Mlimba village was reported to be only 65.7% (adequate clinical and parasitological response at day 28) 49. Furthermore, 24 % of SP tablets and 40 % of quinine sulphate tablets collected in the study area did not meet USP specifications for the amount of active ingredient and were mostly under-dosed 1750. Adding these factors into an effectiveness-model would result in even lower levels of community effectiveness than reported in the coverage figures above.

In the ongoing process of rolling out highly efficacious artemisinin-based combination therapy (ACT), special attention will have to be paid to the quality of prescription in formal health facilities, the ability and willingness of patients to comply with the treatment regimen, as well as to the channels through which these drugs are brought to the patients. It needs to be closely monitored whether restricting ACT to the formal health sector will result in increased community effectiveness (timely administration of correctly dosed efficacious ACT to all those in need) or in a decrease in treatment rates without an improvement in the administered drug regimens.

In the study area, the local understanding of the biomedical concept of malaria has markedly improved after continuous health education. Further education may have to focus on how, when and where to treat a febrile illness instead of reiterating illness concepts. Availability of antimalarials has been largely limited to certified providers and health facility usage was very popular. Nevertheless, the quality of case-management was far from satisfactory. Decreased drug efficacy and sub-standard drug quality may have further decreased effectiveness at community level. Exemption mechanisms aim to facilitate treatment access for poor and vulnerable groups but in our setting they did not seem to be properly implemented. Private drug retailers played a central role in the provision of timely malaria treatment, complementing the existing formal health services. All these issues can be attributed to a health system which is still too weak to deliver a growing number of increasingly complex health interventions. Such constraints need to be tackled urgently with an increased investment in the local health system in order to translate the high efficacy of newly introduced ACT into equitable community-effectiveness and health-impact.

The authors declare that they have no competing interests.

MWH was involved in the design and implementation of the studies, analysed and interpreted the data and wrote the manuscript. BO and CL conceived the studies and contributed to the writing, interpretation and discussion of the manuscript. JJM developed the EMIC together with BO and MWH, and was responsible for its implementation. RN was in charge of all DSS-related activities. AD contributed to the analysis of the data. AM and CM provided support in the field and contributed to the discussion of the findings. AS and HM contributed to the design of the study and to the discussion of the manuscript.