Background
In malaria endemic areas, the lack of specific features and adequate laboratory diagnostic facilities often limits the capacity of health workers to establish a definitive diagnosis of malaria. Given the potentially lethal consequences of a missed diagnosis of malaria, case management guidelines had adopted a presumptive approach to treatment for the febrile child in areas considered endemic for disease [
1]. Given the high disease burden, and the availability of cheap and effective anti-malarials, this was considered a cost-effective approach to care [
2] and is acknowledged to have contributed to the reduction in morbidity and mortality in preschool children [
3,
4].
With reports of declining incidence of malaria [
5‐
10] and now almost universal introduction of artemisinin-based combination therapy (ACT), there have been increasing calls for improving the way diagnosis and case management for malaria is conducted, with a flurry of diagnostic options [
11‐
14]. This follows because it has been well documented that presumptive anti-malarial treatment of febrile cases results in considerable overdiagnosis and treatment across varying transmission settings [
15‐
21]. For instance, a large scale study in north-east Tanzania revealed that 54% of hospital admitted patients treated for malaria had no evidence of
Plasmodium infection as assessed by expert slide microscopy [
19]. Since these patients were usually not investigated or treated for other potentially life-threatening diagnoses, the case fatality rate in the malaria parasite negative group presumptively diagnosed was 1.75-fold higher than in the parasite positive group.
With declining incidence of malaria, adherence to the reflex to treat any febrile case for malaria is likely to put proportionately more patients at risk of not receiving appropriate treatment. Also, the indiscriminate use of ACT has major cost implications and potentially increases the risk of an emergence of resistance [
22,
23]. To address these issues, accurate data are needed on the degree of overtreatment from presumptive diagnosis in different populations [
24,
25].
The bulk of the literature on the subject reports from studies in East Africa in areas with different degrees of malaria transmission. The aim of this study is to provide comparative evidence from a West African country with a distinctly seasonal malaria transmission pattern. The study, therefore, investigated prescription practices immediately following artemether-lumefantrine (AL) introduction as first-line treatment in the Gambia, during and outside of the annual malaria transmission season. The implications of these findings on case management and diagnostic options are discussed.
Discussion
Presumptive diagnosis of malaria previously advocated for malaria endemic areas may have always resulted in some degree of overtreatment. With the decline in malaria prevalence reported from several parts of sub-Saharan Africa [
9,
10,
28,
29], adherence to this approach is likely to increase the degree of overtreatment and inevitably a failure to treat alternative causes of fever. Indeed, studies in critically ill patients treated for "malaria" showed significantly higher mortality in those with a negative slide compared to those with a positive slide [
19,
30]. An economic layer is added to this problem with an increasing number of countries now using relatively costly ACT currently as the first line of defence against malaria. These are important reasons to develop treatment strategies targeting those who actually have malaria and requires a better understanding of the phenomenon of malaria overdiagnosis.
Overtreatment has been reported from many parts of East Africa, mainly from areas with low to moderate or high perennial transmission [
15,
16,
19,
21]. The pattern is likely to vary across different epidemiologic settings, but little has been known from areas where malaria incidence has dropped to very low levels [
31], and data are lacking from West Africa. Here, data on prescription practices in the Gambia, where malaria is highly seasonal [
32] and where incidence has declined to unprecedentedly low levels [
9] are presented.
The observations indicate that PHF workers under-utilize the available laboratory facility in making decisions on diagnosis. Despite the availability of slide reading at the health centres, a slide was only requested in about 1/3 of the cases diagnosed with malaria. Interestingly, this proportion remained stable throughout the year despite the fact that substantially more patients are seen during the wet than in the dry season and suggests that the capacity to perform slide reading is unlikely to be the limiting factor.
The low use of the laboratory facilities supports other reports [
24,
33], and may in part be explained by the mistrust in the quality of the slide reading [
34], which may be justified. While the sensitivity of the microscopic diagnosis performed at the PHF was acceptable compared to the reference slide, the specificity was unacceptably poor, resulting in a high number of "false positives". Thus, similar to reports from East Africa [
35], slide reading at the PHF has an extremely poor positive predictive value, which argues in favour of presumptive treatment.
However, as has been shown previously [
19,
31], a negative slide did not necessarily exclude a patient from receiving malaria treatment. The negative predictive value was high in both the dry season (97.8%) and the wet season (97.3%), which would have allowed exclusion of slide negative cases from anti-malarial treatment with high confidence. Adherence to the policy to withhold malaria treatment from patients with a negative slide result could, therefore, save a significant amount of overtreatment without significantly increasing the risk of missing true malaria cases but achieving compliance with this policy is obviously challenging.
The data presented here indicate that the health workers tend to target children less than five years of age for slide requests, particularly during the transmission season. While the focus on this age group probably reflects the impact of programmes such as the Integrated Management of Childhood Illnesses (IMCI), the increased use of parasitological diagnostics in this age group (for which the IMCI guideline advocated presumptive treatment), may be influenced by more recent local guidelines that promote the use of diagnostics where available. It is conceivable that the recommendation to perform parasitological diagnostics is perceived primarily as an additional layer of measures aimed to improve care in those already identified as risk groups.
However, microscopic diagnosis of blood slides from all participants at the reference laboratory show that, among those for whom a slide reading was requested, the highest prevalence of malaria parasites was in 5-15 year old children; these being 3.5 and 13.1 fold more likely to have parasitaemia in the wet and dry seasons respectively than children under the age of five years. To some extent this may be confounded by the fact that care-givers are more likely to bring younger children to the PHF when they are febrile, and that febrile disease due to a variety of causes other than malaria may simply be more prevalent in younger compared to older children. However, a cross sectional survey carried out recently in healthy Gambians similarly identified the peak parasite prevalence occurring in 10-15 year olds [
9]. It would be tempting to speculate that this may be the result of consequent implementation of malaria control strategies in under fiver year olds, but up to date information on bed net coverage in the study area is lacking. Taken together, this suggests that in areas of low malaria transmission the target group for malaria control measures should be enlarged to include children up to the age of 15 years.
The most remarkable observation, however, is that in the wet season, 87.5% had no detectable parasitaemia on the reference slide. During the dry season 4.6 times less patients received malaria treatment, but among those, the proportion of unnecessarily administered anti-malarials was even higher, being 95.3%. Several reports from East Africa have recently documented malaria overdiagnosis, and - although only few studies simultaneously measured transmission by means of entomological inoculation rates (EIR) - it appears that overdiagnosis seems to increase as transmission declines. In areas of perennial [
15] and moderate [
19] transmission in Tanzania, 62% and 54% of malaria treatments were found to be unnecessary, respectively, whereas overtreatment rates of 86% [
27] and 75% [
16] were reported from areas in East Africa where transmission was classified as low and unstable. One recent study in Tanzania measured both transmission and the degree of overtreatment at the same time and determined that where the EIR was 0.69 by mosquito sampling, 99.6% of patients receiving anti-malarials had no slide-detectable parasitaemia [
31]. Comparable data here for the Gambia [
9,
36] for the year preceding this survey, supports the notion that over-prescription may relate inversely with a decline in the level of malaria transmission.
In order to treat patients appropriately and to ensure the sustainability of ACT and to protect the "useful lifespan" of such drug combinations, the new guidelines from WHO recommend that a laboratory test should be performed before treating [
37]. Therefore, efforts are being made to improve access to parasitological diagnostics. Presently, the options available are to increase capacity to carry out microscopy or the widespread use of rapid malaria diagnostic test kits (RDTs).
Attempts to step up the capacity of slide reading are likely to be of limited success, as microscopy depends on reliable electricity supply and requires considerable skill and expertise. RDTs in a dipstick format appear to be an attractive alternative for parasitological diagnosis. RDTs do not require particular skills or electricity and can be performed at the point of care by the PHF worker. However, procurement of RDTs is probably only a modest proportion of the cost of integrating RDTs into a health system. The challenges of forecasting RDT needs, managing procurement and distribution through the supply-chain should not be underestimated.
Discussions about the best way forward have stimulated a vibrant debate with strong advocacy in favour [
38] and against [
39] the rapid implementation of RDT. While RDTs perform at least as well as microscopy in diagnosing malaria, it appears clinicians are reluctant to refrain from treating for malaria even after a negative test [
40]. Surprisingly, the fact that RDTs can be performed by the PHF worker did not enhance adherence to the test result, and in particular children under five years with a negative RDT were more likely to be prescribed an anti-malarial than those with a negative slide [
40,
41]. Mathematical modelling using data across a broad range of malaria transmission intensities concludes that RDTs can be cost-effective compared to presumptive treatment, but these calculations assume that prescribers treat according to test results [
42], and there are conflicting data as to what extent this could be the case in real life. In Burkina Faso, a similar amount of overtreatment was found in patients managed clinically or with RDT, indicating that test results were simply ignored [
43]. In a Kenyan trial, prescribing of ACT largely followed the results however despite a negative test, 75% of children under five years and 61% of those over five years still received an anti-malarial [
44]. In an area of Zanzibar with parasite prevalence ranging from 10% to 50%, the use of RDTs improved treatment and health outcomes, without increasing the costs for patient management [
45].
It is important to note that revised WHO guidelines are not explicit whether or not to withhold anti-malarials from patients where there is a clinical suspicion of malaria but a negative test result. Here, the health care worker's decision depends on his/her trust into the quality of the test and his/her fear of the implications of a missed diagnosis. These situations often result in the half-hearted prescription of an anti-malarial other than the first-line drug [
44].
Thus, whether RDTs have the potential to become a cost-effective means to improve the management of febrile cases will depend on successful strategies to promote adherence to the result, and has to overcome the labelling and treatment of all febrile illnesses as "malaria". PHF workers need to be toned to make efforts to consider other, often more difficult to establish, diagnoses and to convince caregivers where a fever has no serious cause.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JUO conceived and designed the project, lead the field work, analysed the data and drafted the manuscript; KB and SD helped with the field work and were involved in the review of the manuscript; BW helped with the statistical analysis of the data; DS and DJC provided input into the study design and drafting of the manuscript; MW conceived and designed the project together with JUO, helped with the analysis and drafted the manuscript. All authors read and approved the final manuscript.