Lessons from malaria control to help meet the rising challenge of dengue

Summary

Achievements in malaria control could inform efforts to control the increasing global burden of dengue. Better methods for quantifying dengue endemicity—equivalent to parasite prevalence surveys and endemicity mapping used for malaria—would help target resources, monitor progress, and advocate for investment in dengue prevention. Success in controlling malaria has been attributed to widespread implementation of interventions with proven efficacy. An improved evidence base is needed for large-scale delivery of existing and novel interventions for vector control, alongside continued investment in dengue drug and vaccine development. Control of dengue is unlikely to be achieved without coordinated international financial and technical support for national programmes, which has proven effective in reducing the global burden of malaria.

Malaria's fall and dengue's rise

Malaria has been one of the major challenges to global health during the past century. In 1900, 58% of the world's land area was estimated to have sustained stable malaria transmission.¹ More than a million deaths annually have been attributed to malaria throughout the latter half of the 20th century,^2,3 most in children younger than 5 years, with countries of sub-Saharan Africa bearing the largest toll.^4,5 However, mounting evidence suggests a decline in the global burden of malaria, a decrease that began in the mid 20th century in some regions but was most notable in parts of Africa over the past decade.^1,2,6–8 This reduction has led to renewed focus among the malaria community on a goal of malaria elimination in many countries.⁹ While malaria has been in decline, the geographical range and disease burden of another tropical infectious disease has been on the rise.

Dengue has emerged as an increasing public health problem over the past 50 years, particularly in southeast Asia and Central and South America,¹⁰ with an unknown but possibly substantial level of transmission in Africa.¹¹ Like malaria, dengue is a vector-borne disease of the tropics and is a major cause of morbidity in endemic areas, particularly in children and young adults;¹² however, the scale of dengue morbidity and mortality is uncertain and thought to be less than that of malaria. Dengue is caused by four distinct but related viruses (serotypes DENV 1–4) that are transmitted among people by aedes mosquitoes. The disease burden and geographical range of dengue have expanded, from about 15 000 cases reported annually from fewer than ten countries during the 1960s to almost one million cases a year across more than 60 countries in 2000–2005.¹³ As a result, dengue has been identified as an important threat to global public health.¹⁰ In view of this rising challenge, could lessons learned from global efforts to control malaria help inform strategies to prevent and perhaps reverse the spread of dengue?

In this Personal View, we compare and contrast malaria and dengue with respect to epidemiology, current and future interventions available for prevention and control, and their prioritisation as global health issues, in terms of funding, capacity, and international collaborations. We also argue that improved data on the range and endemicity of dengue are a vital component of global prevention and control efforts.

Effect of vectors on epidemiology

The geographical ranges of both malaria and dengue are limited by the spatial extent of the competent vectors—particular species of anopheles and aedes mosquitoes, respectively. The bionomics of the vectors shapes the epidemiology of each disease. With a few exceptions, anopheles mosquitoes favour rural environments, mainly because of their larval habitat requirements.^14–16 By contrast, the primary vector of dengue, Aedes aegypti, thrives in urban environments where abundant container breeding sites in and around human habitations allow immature vectors to develop and adults to feed and rest close to high densities of humans, their preferred host for blood meals.¹⁷

Although aedes mosquitoes have a restricted flight range of about 100 m in the field,¹⁸ passive transport of Ae aegypti by land and, in immature stages, by sea led to their re-establishment in countries of South and Central America, from which they had previously been eliminated in the mid 20th century,¹⁹ and dispersal throughout southeast Asia during and after World War 2.²⁰ The geographical range of a secondary dengue vector, Aedes albopictus, has also expanded substantially over the past 30 years, but it is a less efficient vector and is not currently seen as a major contributor to or risk factor for increased dengue transmission.²¹

The growing mobility of viraemic people, both within endemic settings and into new regions by increased domestic and international travel and migration, has been key in driving the global expansion of dengue in recent decades.²² This movement has created conditions in which multiple virus serotypes cocirculate, leading to an increase in the risk of sequential infections and severe disease. By contrast, the ecological requirements of anopheles mosquitoes have not facilitated their dispersal,²³ and the unprecedented urbanisation that has characterised the past century is associated with reduced risk of malaria transmission, at least in the African setting.⁴

Quantifying disease burden and distribution

The global burden of a disease is a function of both its geographical range and the intensity of transmission in affected areas. By both these measures, the global burden of malaria has unequivocally decreased over the past century,¹ although this decline has not been consistent across all malaria-endemic countries.⁸ Serious efforts to define the geographical limits and intensity of malaria transmission go back to the mid 20th century,^24–26 when global control and eradication efforts were gathering momentum. A renewed effort to quantify the magnitude and distribution of the burden of malaria has seen new epidemiological and cartographic techniques applied to multiple collated data sources to model the spatial extent of malaria transmission and so to estimate populations at risk of exposure.^27–29 These calculations place 2·4 billion people living in 87 countries at risk of Plasmodium falciparum infection in 2007,²⁷ resulting in around 450 million clinical cases of P falciparum malaria annually.²⁹ The inclusion of uncertainty intervals around estimates has been a major step forward with these cartographic methods.³⁰ For dengue, assessments of the spatial extent of transmission have been based largely on empirical data of reported dengue cases from endemic and epidemic settings, with models then fitted to correlate the observed distribution with environmental and climatic characteristics.^31,32 WHO estimates that 50 million dengue virus infections occur every year across about 100 countries, representing a population at risk of 2·5 billion people,¹⁰ although this number could be an underestimate of the true burden.³³ The most recent assessment of the global distribution of dengue identifies 128 countries with good evidence of transmission and puts almost four billion people at risk.¹¹

The intensity of transmission of both malaria and dengue is spatially and temporally heterogeneous.^34–37 The most commonly used measure of malaria endemicity is the parasite prevalence rate, which represents the proportion of a population with malaria parasites detectable in their blood.³⁸ This measure has been used widely in malaria surveys throughout the past century and has been used to generate the first evidence-based global map of malaria endemicity in 2007, recently updated for 2010 (figure).³⁹ Another key metric of malaria transmission risk is the entomological inoculation rate, which represents the rate at which people are bitten by infectious mosquitoes.⁴⁰ The relationship between the entomological inoculation rate and the parasite prevalence rate is non-linear.^41,42 Empirical measurements for the entomological inoculation rate have been gathered less routinely and consistently than for the parasite prevalence rate, making the former a less useful measure for global endemicity mapping.⁴³ Consequently, the entomological inoculation rate and other important metrics for malaria have been inferred with modelled relationships⁴⁴ between them and extensive maps of parasite rates.³⁹

Figure.

Global risk of malaria and dengue

Annual mean prevalence of Plasmodium falciparum malaria for 2010 (A). Reprinted from reference 39, with permission of BioMed Central. Global dengue risk (B), based on data from WHO, Centers for Disease Control and Prevention, GIDEON online, ProMED, DengueMap, Eurosurveillance, and other published work. Reprinted from reference 55, with permission of the Massachusetts Medical Society.

Epidemiological data on the global burden of dengue rely almost entirely on reports of clinically apparent disease, derived from national surveillance systems^10,45–52 and, in a few cases, from prospective longitudinal studies.^53,54 The figure shows a map of dengue risk that combines disease notification and outbreak data from international organisations, case reports on returning travellers, published scientific literature on dengue occurrence, and a biological model of environmental suitability.⁵⁵ Serological data from longitudinal studies^56–62 permit estimation of infection incidence in a population, including the ratio of symptomatic to inapparent infections. This type of study depends, however, on follow-up of cohorts, which needs far greater investment of time and money compared with cross-sectional surveys that are used to obtain estimates of the malaria parasite prevalence rate.

Traditional indicators of the abundance of aedes mosquitoes, based on immature vector stages (house index, container index, Breteau index), are collected routinely in many dengue-endemic countries, but their correlation with human infection and disease is poor.^63,64 Counts of Ae aegypti pupae per person might correlate more closely with adult vector density and, therefore, potential for dengue transmission.^36,65 Direct measurement of the density of adult Ae aegypti—with PCR to ascertain the proportion infected with dengue virus—would be most informative, but this approach is logistically and financially demanding to do on a sufficiently large scale in view of the difficulty in sampling adult vectors and the expected large variance in both adult numbers and prevalence of infection.^36,66,67

For both malaria and dengue, the relationship between the risk of infection and the risk of disease is non-linear and depends on host immune status and age at infection.^68–72 The most appropriate metric will be determined by its purpose. Clinical case numbers are relevant to the prediction of demand for diagnostic tests, health-care services, and treatments. WHO also defines laboratory-confirmed clinical dengue cases of any severity as the most appropriate endpoint for dengue vaccine trials.⁷³ However, for describing transmission extent and intensity, especially when making comparisons between countries and over time, reliance on case-burden data is fraught with issues of inconsistent reporting patterns (both spatially and temporally), differences in clinical case definitions, over-reporting when laboratory testing is not routine, and under-reporting of patients who do not present to health services or who are managed as outpatients only.^74,75 A measure of the incidence of infection, rather than disease, might also be an appropriate endpoint for trials of dengue vector-control interventions in the community; not only is active surveillance of clinical outcomes more resource-intensive than cross-sectional blood sampling but also a large (and variable) proportion of prevented infections are likely to be asymptomatic.⁷⁶ Therefore, a smaller sample size will be needed to show an effect on infection rates, compared with a clinical effect of the same size, because of the higher overall event rate for infections versus clinical cases.

What alternative metric could be used to measure dengue endemicity, equivalent to the parasite rate for malaria? Virological markers of dengue—such as viraemia and presence of the NS1 antigen in blood—are short-lived compared with untreated malaria parasitaemia, disappearing about 1 week after onset of clinical symptoms.⁷⁷ Furthermore, the magnitude and duration of viraemia varies with severity of disease, virus serotype, and host immune status.^78–80 Cross-sectional age-stratified serological surveys of dengue-specific IgG can indicate the prevalence of past exposure to dengue virus but are confounded by antibodies directed against other flaviviruses, where these cocirculate. Cross-sectional seroprevalence surveys of dengue-specific IgM might indicate recent infection with dengue virus or other flaviviruses. Aside from potential low specificity, interpretation is complicated by the variable kinetics of the IgM response, most importantly, the difference between first and subsequent infections,⁸¹ making comparison of population-based IgM surveys between epidemiological settings difficult. Population-based surveys of dengue neutralising antibody, measured by the plaque reduction neutralisation test, would provide the most sensitive and specific information on virus transmission patterns, including serotype-specific data and multiple heterotypic exposures. However the plaque reduction neutralisation test is substantially more resource-intensive than standard IgM and IgG immunoassays. Finding the appropriate metric to measure the endemic level of dengue is a clear research priority.

Interventions for prevention and control

Success in controlling malaria over the past century has been attributed predominantly to widespread implementation of insecticide-treated bednets, household spraying of residual insecticides, and effective drugs to reduce mortality and interrupt transmission.⁶ The countries in which little progress has been made with malaria control are commonly those where political instability, war, or economic underdevelopment have hindered widespread implementation of these interventions.⁸²

The situation with dengue is different; vector control is the only currently available approach for prevention and control and is pursued mainly through reduction of larval development sites, via environmental clean-up campaigns to dispose of discarded or unnecessary water containers, and prevention of mosquito access to breeding sites. Other methods include treatment of water-storage vessels with larvicide⁸³ or predacious copepods⁸⁴ to kill larval stages. The effectiveness of these interventions has been demonstrated at a local community level^84–86 but rarely on a large scale or across diverse epidemiological settings (not since the Ae aegypti eradication campaign of the 1950s), with Singapore and Cuba perhaps the only exceptions.^87–89 Success of such efforts depends on sustained community support and participation.⁹⁰ However, even when mosquito populations have been reduced drastically, as in Singapore, cases of dengue still occur,⁹¹ with evidence of increasing risk of clinical disease associated with older age at first infection.^72,92 Killing adult mosquitoes has a theoretically greater effect on transmission than does targeting larvae. Space-spraying of insecticide to kill adult vectors in and around households is popular because it represents a highly visible response to localised outbreaks of dengue, but a sustained effect on virus transmission has not been demonstrated.^10,93

Indoor residual spraying of insecticide has a long history of use in malaria control, and its importance as a key intervention for interruption of malaria transmission has been reaffirmed by WHO.⁹⁴ Many behavioural characteristics of anopheles vectors that make indoor residual spraying an effective malaria intervention, such as their anthropophagic biting preferences and tendency to rest and feed indoors,^14–16 are common also to aedes dengue vectors. There is some evidence that high household coverage of indoor residual spraying in an outbreak setting could reduce dengue transmission.⁹⁵ Use of this method to control yellow fever in the Americas in the mid 20th century had a concomitant and striking effect on dengue transmission, but there are very few reports of the application of indoor residual spraying specifically to control dengue.^95–98 The preference of aedes vectors for daytime activity and feeding means that insecticide-treated bednets are ineffective for dengue control. Findings of several small trials^99–101 of other insecticide-treated materials, such as curtains and water-jar covers, indicate a reduction in indices of household vectors, and larger trials are warranted to investigate effectiveness in a range of epidemiological settings.

Early diagnosis and treatment with effective drugs reduces morbidity and mortality from malaria. International guidelines recommend parasitological confirmation, when possible, of all suspected cases of malaria and prompt initiation of treatment to prevent progression to severe disease.¹⁰² Timeliness is also very important for effective clinical management of dengue; progression from an acute febrile phase to non-complicated recovery, or through a critical phase characterised by thrombocytopenia and capillary permeability with potential for haemorrhage and shock, takes place over 3–7 days.¹⁰

Unlike malaria, no specific treatment for dengue is available, and clinical management entails close haematological monitoring, fluid-replacement therapy as required, and recognition of warning signs of severe disease. Although serological, molecular, and rapid diagnostic tests for dengue are widely available, the expense, waiting time, and large case numbers mean that clinical management and case reporting in most endemic settings is based on clinical diagnosis alone. Increasing availability of rapid diagnostic tests could theoretically improve timeliness and accuracy of dengue diagnoses. However, studies of the effect of test results on clinical management and outcome of dengue cases, including cost-effectiveness studies, are needed to inform recommendations for widespread use. Demand for routine diagnostic testing for dengue could increase substantially if an antiviral drug were available.

Research efforts towards vaccines against malaria and dengue are similarly complicated by (among other challenges) the antigenic or serotypic variability of the organisms.^103–105 A longlasting highly effective dengue vaccine should be much easier to develop than an equivalent malaria vaccine because of the relative antigenic complexity of the two pathogens and the longevity of immune responses to viral infections, compared with those to malaria parasites.¹⁰⁶ However, developers of a dengue vaccine must contend with the theoretical risk of severe disease associated with sequential infection with a heterologous serotype and, thus, aim to develop a tetravalent vaccine.¹⁰⁶ Candidate vaccines for malaria and dengue are in phase 3 field trials,^107,108 but despite publication of promising clinical trial data for the leading malaria vaccine candidate,¹⁰⁹ a substantial vaccine-mediated reduction in the global burden of either disease is not imminent. Thus, vector control, effective diagnosis, and clinical management remain the cornerstones of control for both diseases, for the foreseeable future.

The challenge now in malaria control is equitable and effective implementation of interventions that have proven efficacy. However, to tackle the increasing burden of dengue, well designed and controlled field trials are needed of both existing and novel vector-control interventions, linked to detailed epidemiological data, to improve the evidence base and inform local and national dengue-control strategies. Further challenges for evaluation of dengue interventions might include the effect of human movement on patterns of transmission, and the pronounced temporal and spatial heterogeneity in transmission, which will necessitate very large cluster-randomised study designs. These issues are also likely to be challenges for malaria control though the elimination or eradication phases.

Prioritisation and investment of funding and resources

Malaria control throughout the past century has been a combined effort of national, regional, and international programmes. The global malaria eradication programme launched in 1955 by WHO was the largest coordinated international public health campaign in history.¹¹⁰ With an intensive strategy of vector control using residual insecticides, combined with detection and treatment of cases, 22 countries in the Americas and 27 in Europe achieved malaria elimination between 1950 and 1978.¹¹¹ Despite these successes, the goal of elimination was not met universally and was never proposed for sub-Saharan Africa; in 1969, WHO's strategy was revised to one of control.¹¹²

Efforts to control dengue also benefited from an elimination campaign in the mid 20th century; in 1947, the Pan-American Health Organization adopted a proposal by Brazil for a so-called hemispheric (pan-American) strategy to remove the Ae aegypti vector.¹¹³ Although the aim of this campaign was eradication of urban yellow fever, which shares the same vector as dengue, the successful elimination by 1967 of Ae aegypti from all countries of the Americas (except for the USA, Venezuela, and the Caribbean region)¹¹⁴ saw a substantial reduction in dengue morbidity across this region.¹¹⁵ Unfortunately, this campaign had the same outcome as the global malaria eradication programme, with a reversion to a strategy of control because of a combination of reduced political will, insufficient financing to sustain intensive control efforts, and increasing decentralisation of national public health authorities, among other factors.^115,116 The Ae aegypti vector re-established itself in areas from which it had been eliminated, with a resultant rise in dengue epidemics in the Americas throughout the 1970s and 1980s. In southeast Asia, an elimination goal for dengue or its vector has never been proposed formally.¹¹⁷

Efforts to control malaria during the past 15 years have intensified after the development of international initiatives to coordinate and finance the scale-up of interventions, beginning with the Roll Back Malaria Partnership, launched in 1998, and the Global Fund for AIDS, Tuberculosis, and Malaria, founded in 2002. More recently, the Bill and Melinda Gates Foundation has allocated substantial funds to malaria control and eradication efforts. These initiatives recognise the need for external funding and support to malaria-endemic countries to achieve coverage of interventions at a level that will affect transmission and morbidity. An estimated US$9·9 billion was committed by international donor agencies for malaria control in endemic countries between 2002 and 2010.¹¹⁸ By contrast, vector-control interventions for dengue remain the financial and logistical responsibility of national control programmes in endemic countries, which are funded from national budgets with no substantial or sustained external sources of financing. Dengue is a high public health priority in endemic countries,¹¹⁹ but the main target of spending is responsive vector-control activities around reported cases, combined with passive case surveillance and some routine virus and vector surveillance. Budgets are usually insufficient to implement these actions fully, let alone to sustain breeding source reduction activities, larval control, and environmental management, which might be more effective^120,121 but are highly resource-intensive.

It is difficult to see how the continuing geographical spread and increasing intensity of dengue transmission can begin to be reversed without support in coordination and financing from outside endemic countries. Support should include applied research to improve the evidence base for existing vector-control techniques, for novel interventions such as transfection of Wolbachia spp into Ae aegypti to suppress dengue transmission,¹²² and for strategic planning to implement and finance a future vaccine. Even when a dengue vaccine becomes a reality, external assistance for financing and implementation will be needed by some endemic countries, as will continued concerted efforts in vector control. Unlike malaria, which receded from southern Europe in the mid 20th century, aedes mosquitoes, and possibly dengue, could continue to expand into warmer areas of high-income countries, including Australia, the USA, and southern Europe.^31,32 This possibility should provide any additional impetus needed for dengue to be viewed as more than a neglected tropical disease. The burden of morbidity, mortality, and economic loss attributable to dengue is not comparable with that caused by malaria. However, the coordinated initiatives for funding of regional and global collaborative research and control activities, which have proven effective to address the global burden of malaria,¹²³ could also drive similar gains in dengue control.

Moving forward

Based on lessons learned from malaria control, we propose that development of better methods to quantify dengue endemicity and disease burden, permitting comparisons across countries and regions, is an essential step towards halting the current rise in disease range and intensity. We must be able to quantify these increases accurately so we can establish baselines against which future trends can be compared. Analysis of the clinical and demographic profile of acute cases can tell us much about local dengue transmission dynamics, but improved indices of transmission—including means of accounting for asymptomatic infections—that can be measured at a population level and within specific subgroups would provide a much more complete picture of local transmission patterns. This information can guide effective surveillance and implementation of interventions, including future vaccines. Use of serological markers of dengue infection in epidemiological studies has some limitations, but age-stratified serosurveys of neutralising antibodies against the virus probably represent the best equivalent to the malaria parasite prevalence survey for population-based estimates of the incidence of infection. Improved entomological measures of risk for dengue transmission, based on density of the adult vector and infection prevalence, could complement these estimates but might also be inferred, similar to malaria, from measurement of incidence in human populations.

Improved estimates of the dengue disease burden would inform economic analyses of vector-control activities and future vaccination strategies. Effective implementation of these interventions could be achievable within the national budgets of a few dengue endemic countries, but many national dengue control programmes would benefit from coordinated international funding to achieve adequate coverage, as has proven effective in malaria control. This fact reinforces the importance of developing improved indicators of local, regional, and global dengue endemicity and disease burden, to advocate for funding directed to areas of greatest need, to identify locations where interventions are most likely to succeed, and to monitor future progress of disease prevention efforts, including vaccines.