A refined estimate of the malaria burden in Niger

Niger is a landlocked Sahelian country covering 1,267,000 km². As in many other African countries, malaria remains a leading cause of mortality and morbidity in vulnerable groups, such as children under the age of five years and pregnant women [1]. In 1984, Niger took a timely decision to face up to this situation, by implementing a voluntary national anti-malarial policy. In 2000, its health authorities signed the Abuja Declaration, pledging themselves to halving mortality due to endemic malaria by 2010 and decreasing the malaria burden by 75% by 2015 [2]. With support from the Global Fund and the Roll Back Malaria Partnership (RBMP), Niger has implemented several control strategies to reduce malaria incidence and morbidity in priority groups. These strategies are in line with the Millennium Development Goals, which aim to eliminate malaria by 2030 [3, 4].

The interventions integrated into public health policy in Niger include: (i) the adoption in 2005 of the use of artemether-lumefantrine (AL) combinations as the first-line treatment of uncomplicated malaria, replacing conventional anti-malarial drugs; (ii) the adoption in 2005 of intermittent preventive treatment (IPT) with sulphadoxine-pyrimethamine for pregnant women; (iii) the introduction in August 2007 of free health care for children under the age of five years; (iv) the distribution, between 2006 and 2009, of 4,634,706 insecticide-treated nets/long-lasting insecticidal nets. The distribution of these nets has continued, with the aim of achieving universal coverage by 2012; (iv) the management of malaria cases directly at healthcare points. These facilities constituted a network of 1,800 units in 2007 and the number of units has increased steadily ever since, reaching 2,500 by 2010. Similarly, the number of health centres has doubled and there are now 820 centres, each run by a nurse; (v) the introduction in 2008 of the use of artesunate-amodiaquine combinations to combat potential deficiencies in the paediatric form of the AL combination (Co-artesian^®) and to expand the supply of artemisinin-based combination therapies (ACT); (vi) authorization of the market release of dihydroartemisinin-piperaquine in 2010, and an increase in the number of health workers and physicians assigned to health facilities in the capacity of these centres; and (vii) greater availability of recently introduced malaria treatments and improvements in patient management, with the launch, in 2010, of the Affordable Medicines Facility - malaria, helping Niger to increase the supply of affordable ACT via the public and private sectors, while excluding the use of counterfeit or ineffective anti-malarial drugs [5–8].

However, the beneficial effects of these prevention and control measures remain barely perceptible in Niger, whereas considerable decreases in malaria transmission and incidence have been reported elsewhere in Africa following the introduction of similar control strategies [9, 10]. Indeed, over this same period, there has been an unprecedented increase in malaria incidence and the risk of the disease in Niger. The number of malaria cases recorded by the National Health Information System increased by a factor of 5.3 in the space of a decade, from 592,334 cases in 2000 to 3,138,696 cases in 2010, whereas the total population of Niger increased only slightly, by a factor of 1.3, over the same period. Between 2000 and 2010, malaria incidence quadrupled [11, 12]. However, this worrying trend was difficult to analyse, because the key performance indicators typically used by the Global Fund and RBMP to evaluate the efficacy and impact of control and prevention programmes, such as the number of people with access to treatment, the number of individuals sleeping under anti-mosquito bed nets, the number of pregnant women receiving IPT and malaria-related mortality were satisfactory [13].

There are, therefore, discrepancies between the considerable financial and human resources that have been mobilized by the health authorities to fight malaria, the satisfactory indicators and the apparent failure of the National Control Programme to reduce the malaria burden in Niger. However, by contrast to the observed increase in malaria morbidity, mortality has continually decreased, from 0.21% to 0.06%, over the same time period. Assuming that the system for recording births and deaths is sufficiently reliable and reproducible from year to year, this implies that the control interventions implemented have at least some effect. It may simply be difficult to determine their impact due to the low accuracy of the basic indicators for malaria assessment. Accurate measurements of malaria burden are essential when planning and advocating malaria control strategies at the national scale [14, 15]. In Niger, it is difficult to evaluate the contribution of malaria to disease, because incidence estimates are still largely based on clinical diagnosis, which overestimates the number of malaria cases [16]. The use of biological diagnosis methods for the confirmation of cases of suspected malaria has thus become a key issue in the planning of control activities and disease management in Niger.

A large cross-sectional survey was carried out in 2009 and 2010, in 13 health districts from seven regions of Niger, to address this issue and estimate the discrepancy between the numbers of suspected and confirmed cases of malaria in febrile patients consulting at healthcare points and treated with an anti-malarial drug. The results highlight inadequacies in the description of the malaria situation and disease risk in Niger, due to the over-diagnosis of malaria in patients with simple febrile illness. They highlight the absolute necessity of confirming all cases of suspected malaria by biological diagnosis methods, as recommended by the WHO. Niger extends over a number of latitudes, resulting in considerable geographic heterogeneity in terms of malaria endemicity and populations at risk of the disease. The systematic confirmation of malaria cases by biological diagnosis methods would have major implications for morbidity and future diagnostic in the context of the malaria control programme in Niger and WHO guidelines.

All partners involved in fighting malaria in Niger have been committed for a number of years to improving the coverage of populations at risk as concerns basic interventions, in line with the objectives of the RBMP and international recommendations. However, the progress made has not yet resulted in a decrease in malaria levels and morbidity as expected. To tackle this question in the best possible conditions and exploiting recent objective data, a large nationwide survey was carried out aiming at estimating the confirmation rates for the cases declared, most of which were identified on the basis of nonspecific clinical signs of malaria. In a context in which malaria levels are generally declining worldwide, it is essential to have an accurate assessment of the progress made by Niger towards controlling this endemic disease and to identify the trends observed [23, 24].

This survey showed that the number of cases of presumed malaria reported in health centres in Niger is largely overestimated, with almost half the patients treated for malaria during the rainy season not carrying parasites. These findings have important implications. Firstly, as concerns individuals, many patients are not receiving the most appropriate treatment. Secondly, at community level, the resulting over-consumption of drugs is very expensive and leads to unnecessary drug pressure favouring the emergence of resistance. According to the WHO, presumptive diagnosis is a widespread practice throughout Africa. Almost 80% of African countries confirm no more than 10% of infections, whereas more than half the cases, if not more, reported in Asia are confirmed. This difference between geographic zones may be accounted for, at least in part, by the higher rates of transmission in Africa, necessitating the mobilization of greater human and financial resources [12, 23].

These observations also raise questions about the precision of the malaria indicators used in Niger. The number of cases, divided by population size, provides information about the intensity of endemicity, making it possible to follow the impact of interventions. However, it is clear that the use of presumptive diagnosis statistics in Niger does not provide a sufficiently precise estimate of malaria incidence, particularly as free care and improvements in health cover have led to an increase in the number of consultations. In the past, many people with malaria would not have consulted the health services. The excess cases of malaria recorded in recent years may therefore also reflect a distinct improvement in health cover. Indeed, the incidence data reported in 2009 (154 per thousand) by the National Health Information System in Niger are similar to those reported for the same period by neighbouring Sahelian countries, notably Mali (191 per thousand) and Burkina Faso (195 per thousand), both of which also confirm very few cases. Senegal is the only country in this region in a better position, with an incidence of malaria only one 10th that in Niger (14.6 per thousand in 2009) [25]. Note that Senegal recently implemented a strategy involving the systematic confirmation of malaria cases before notification and treatment [26]. A more realistic estimate of the malaria situation in Niger may be provided by using a correction factor indexed on the mean rate of confirmation calculated here for the country. Weighting the raw data for 2009 by a confirmation index of 0.45 (for microscopy) or 0.29 (for RDTs), gives more realistic estimates of malaria incidence in Niger of 59 and 42 per thousand in 2009. However, this weighted incidence is still high, clear evidence of the progress required for Niger to join the group of African countries that have managed to decrease malaria levels significantly through the implementation of appropriate control measures (Botswana, Cap Verde, Eritrea, Madagascar, Namibia, Rwanda, Sao Tome, Senegal, South Africa, Swaziland, Zambia and Zanzibar, United Republic of Tanzania) [23]. However, these countries are not easily comparable in terms of health, geographic and socio-economic conditions. Niger has a low human development indicator and a high birth rate. The zero to four years age group, which is the most susceptible to malaria, accounts for 20% of the population in Niger, and the fecundity rate of adolescents aged 15 to 19 years is 199 per thousand. For the purposes of comparison, that in Senegal is only 100 per thousand [27].

Niger lies between the latitudes of 11°37' and 23°33' north, and between 16° east and 0°10' west and it covers a total area of 1,267,000 km². Due to its geographic location and its large surface area, climatic and rainfall conditions in Niger are highly diverse. This has resulted in a heterogeneous distribution of An. gambiae and An. funestus, the principal vectors of malaria, throughout the country. Malaria occurs in an endemic manner, with a seasonal increase in the number of cases during the rainy season. Transmission rates display major temporal and spatial variations, with a gradient of decreasing transmission rates from the south to the north linked to temperatures, rainfall and the availability of surface water. Malaria transmission is generally described by dividing the country into three zones. The extreme south constitutes a zone of stable malaria transmission over more than four months. The north constitutes a zone of instable malaria, with low levels of transmission and a risk of epidemics in certain rainfall conditions. Finally, the zone lying between the north and the south is intermediate in nature, with seasonal transmission limited between August and October [17]. The rate of case confirmation was found to decrease with decreasing endemicity. Thus, in Agadez and Arlit, where malaria is hypoendemic, a mean of 23.4% of slides tested positive in the rainy season, and 9.8% tested positive during the dry season, when transmission rates are low. This observation appears trivial, but the possibility of using slide-positivity rates as a surrogate measurement of incidence provides information about variations in transmission along a south/north line relative to the incidence at the Gaya site (the site furthest south, at 12° north), which is the most heavily exposed [28]. Figure 3 shows variations in the relative incidence of malaria (rΔinc) as a function of latitude in Niger. It was observed that rΔinc calculated from the estimated confirmation rates, based on microscopy, decreased along a latitudinal gradient during the season in which transmission rates were high. This indicator fell from 1 at Gaya to 0.01 at Arlit, corresponding to an incidence two orders of magnitude lower in the most northern part of Niger than in the southern Sudanese zone. The rΔinc values between these two zones were variable. The latitude of 13°5' N, which passes through the towns of Zinder and Niamey, approximately separates zones with different patterns of transmission. This spatial heterogeneity, in addition to the seasonal variations associated with the rainy season, must be taken into account when selecting the most appropriate diagnostic method for the epidemiological situation for malaria in this country. The large difference observed between the numbers of suspected and confirmed cases of malaria introduces a major, non-negligible bias into incidence measurements in Niger. Improvements in the way in which this issue is handled are required. This applies particularly to studies aiming to establish a link between environmental or climatic factors and the risk of malaria and in which estimates are based on historic data concerning the number of suspect cases of fever, indeed it is now known that more than half these cases are of uncertain, but probably viral or bacterial, origin [29]. There is a well-established relationship between exposure to malaria, transmission intensity, age and prevalence. The several studies performed so far in Africa have demonstrated that in areas with intense transmission the bulk of the morbidity caused by malaria mostly concentrated in the younger age groups whereas few cases were reported after the age of 10 years as a result of exposure-dependent acquired immunity. In Niger, the transmission pattern is strongly dependent on climate factors responsible for irregular level of malaria incidence form a year to the next. Thus, climate changes in addition to seasonal variations and fighting interventions would explain periods with low transmission resulting in a loss of immunity in populations, especially those in Northern Niger. Consistent with this hypothesis, we found that significant rates of disease similarly occurred into all age groups, including during into adolescence and adulthood, suggesting that malaria develops at a moderate transmission level in Niger [30, 31].

Despite its long history of use for the presumptive treatment of cases of malaria, fever is not a discriminating characteristic [32]. This generalized practice is based on the notion that the disadvantages of unnecessary anti-malarial treatment in patients that do not have malaria are outweighed by the risk of leaving malaria untreated, potentially leading to complications that could rapidly cause the death of the patient. This approach is justifiable in situations of strong endemicity, in which the available anti-malarial drugs are effective, well tolerated and cheap. However, the need to confirm malaria cases by biological diagnosis before notification and treatment, to increase indicator precision and improve the management of patients, is now very clear [33, 34]. In Niger, the many cases of presumed but unconfirmed malaria should be considered to be cases of infection of unknown etiology. In the patients concerned, the administration of an anti-malarial drug is of no benefit and, indeed, may actually delay the introduction of an appropriate treatment. Fever cannot be considered an adequate criterion for distinguishing malaria from other infections. In a previous study, Rougemont et al. showed that associating the use of fever with that of other criteria, such as the supposed cause and duration of the fever, greatly improved the performance of presumptive diagnosis [35]. However, the use of such a decisional algorithm nationally is not realistic, due to the excessively heavy workload it would create in healthcare structures from the start of the season of high-level transmission, and, in some cases, due to inappropriate clinical practices. The WHO currently recommends the confirmation of all cases of malaria by a biological test before treatment [36]. The malaria transmission in Niger is strongly influenced by epidemiological conditions, due to the large number of latitudes covered by the country. Sensitivity analysis have showed that cost-effectiveness of the diagnosis method depends on several factor among which the malaria prevalence among febrile outpatients is determining. RDTs were shown to be cost-effective at a > 95% confidence level compared with microscopy at malaria prevalence above 20%, which is likely not to be reached in the Northern regions of Niger. From this figure, and provided the question of the performances of RDTs upon exposure at elevated temperature is solved, RDTs are certainly the most appropriate option in the Saharan and Sahelian/Saharan regions, whereas microscopy seems to be more appropriate in the south, below a latitude of 13°5' N. Similarly, RDTs have the potential to be cost-effective in most parts of Niger during the dry season relatively to microscopy [37].

This survey showed that the number of cases of presumed malaria reported in health centres in Niger is largely overestimated. The results highlight inadequacies in the description of the malaria situation and disease risk in Niger, due to the over-diagnosis of malaria in patients with simple febrile illness. Presented data demonstrate the absolute necessity of confirming cases of malaria recorded in Niger by a biological diagnosis method providing a direct (microscopy) or indirect (RDT) demonstration of the presence of the parasite in the blood before treatment. However, the availability of RDTs and other diagnostic methods at health centres is not sufficient. Only the systematic and effective use of these means will increase the precision of the indicators used in patient care. Pilot studies carried out in 2011, at about a dozen centres in the capital, show that many of these structures had microscopes and RDTs but did not make use of biological detection methods for infections, preferring to rely instead on more rapid clinical diagnosis techniques. Case confirmation will thus also require a change in behaviour, through the training of healthcare staff, the introduction of quality control, greater supervision of the integrated health centres, the implementation of good clinical practice and a general optimization of the use of available diagnostic methods.