Undefeated (© IRD / M. Dukhan)
Bednets seemed the perfect malaria intervention: cheap, needing no doctors or needles but saving the lives of perhaps five children for every thousand covered. But unfurling hundreds of millions of mosquito-killing nets across Africa has provoked a wave of insecticide resistance. Resistant mosquitos pass through and bite instead of dying. Also, children eventually come out from under bednets when they are older which might be worse than having had no protection to begin with in areas with intense malaria transmission. Remarkably, the most recent and comprehensive research on malaria mortality shows weak or no evidence that bednets save the lives of children in Africa.
In 2000, health officials set a goal to protect 60 percent of the population at greatest risk of dying from malaria, children under five and pregnant women. Compelling studies had shown that bednets dramatically reduced malaria and saved lives. In 2005, the World Health Assembly voted to hoist the target to 80 percent. Distribution of nets leapt to 47 million in 2006, up from 17 million the year before. In 2007, Melinda Gates called for the total global eradication of malaria. In 2008, the world spun up and delivered more than 60 million nets. Nets became a cause célèbre, with Ashton Kutcher leading the charge on Twitter in 2009. In 2010, more than 140 million nets were shipped to sub-Saharan Africa, where more than 750 million people are at risk for malaria.
Scale-up drives resistance
But living organisms try to stay that way. And the immense selective pressure of mosquitocidal nets drove a proportionate resistance pushback. More nets, deployed for more time, select for a more resistant mosquito population. For example, in a large trial in Asembo, Kenya, as bed net coverage ascended, a key mutation conferring insecticide resistance expanded through the mosquito population. When bednet coverage reached 100 percent, the resistance mutation also neared 100 percent frequency.
Treated nets all use pyrethroids, a class of insecticides originally derived from chrysanthemums. Pyrethroids are enormously toxic to mosquitos but comparatively safe for humans. Pyrethroids act on nerve cells by binding to a receptor site on a sodium channel, inhibiting its deactivation. In susceptible mosquitos, pyrethroids trigger rapid paralysis or “knockdown,” then death.
Not all die, however. Mosquitos have evolved a number of defenses. Some are metabolic — insects rapidly detoxifying or sequestering poisons. In addition, researchers looking at mosquito feet with an electron microscope have even detected “cuticular thickening” which slows or blocks insecticide absorption when mosquitos touch down on nets. Mosquitos might be evolving their behavior as well to avoid bednets. A recent study of two villages in Benin found that mosquitos shifted their peak feeding time from the middle of the night, when nets protect people sleeping under them, toward dawn when villagers are waking up and exposed.
The frequency of resistance genes within a population ebbs and flows, and pyrethroids can still do serious damage even where resistance is present. Although the large number of nets drives selection for resistance, the insecticide onslaught also kills huge numbers of mosquitos, reducing transmission. Against susceptible mosquitos, bednets radically reduce bloodfeeding, by 90 percent or more. By contrast, bloodfeeding of knockdown-resistant mosquitos is essentially unaffected by the pyrethroids on bednets. And, by itself, the physical barrier presented by nets provides only very partial protection.
The search for alternatives
Venerable pyrethroids are now roughly half a century old. There are efforts to find new insecticides, but none are in sight. The Innovative Vector Control Consortium (IVCC), set up and funded by the Bill and Melinda Gates Foundation, has been working on the problem since 2005. IVCC’s current portfolio shows no new chemicals entering into a development or registration for use phase. If there are any promising candidates further upstream, IVCC chief operating officer Tom McLean won’t talk about them. He fielded a question on status by saying: “At this early stage of the development process it is not appropriate to publish specific chemical structures of what is in the pipeline because it is essential to preserve the commercially competitive nature of these products.”
The Gates Foundation directly funded out-of-the-box projects like “click chemistry” in which two non-toxic chemicals bind together lethally inside mosquitos. But that clever idea did not pan out.
According to Helen Pate Jamet, senior scientist for bednet maker Vestergaard Frandsen, “ideally we need at least 2-3 new insecticides from completely different insecticide classes in order to have a real impact on resistance and have the ability to rotate/mix different classes.”
Meanwhile, Vestergaard Frandsen is testing nets impregnated with chlorfenapyr which comes from a new class of insecticides called pyrroles. Unfortunately, they aren’t as good as pyrethroids. Chlorfenapyr is less toxic to mosquitos and more harmful to humans than pyrethroids. Chlorfenapyr is a "prodrug" that has to be broken down before starting the chain of events that, in time, kills the mosquito. Consequently, chlorfenapyr-treated nets provide little to no personal protection from malaria. Mosquitos still bite, only dying later. “Any inhibition of blood feeding associated with the insecticide treatment was not statistically significant,” according to one study of chlorfenapyr–laced bednets. To work, much of the community must also sleep under a net, thereby reducing the mosquito population. The direct life-saving benefit of pyrethroid bednets is lost.
Geographic extent and implications
Pyrethroid resistance has been found all over the African continent. Mosquitos have developed resistance to other insecticides, but according to WHO's most recent report, "Resistance to pyrethroids seems to be the most widespread." And it's worsening. Previously there were pockets of resistance; now there are pockets of susceptibility.
Yet remarkably there is debate about whether insecticide resistance impacts malaria control. “[T]here is broad consensus that the degree of resistance that has developed and its likely trajectory are a cause for serious concern,” according to Scott Filler, senior advisor at the Global Fund for Aids, TB and Malaria. The trajectory, Filler says is toward “widespread control failure,” but “the pace of this process and the degree of reduction in malaria control effectiveness remains unknown.” The Global Fund purchases the majority of the world’s bednets, some 56 million in 2010.
Janet Hemingway, director of the Liverpool School of Tropical Medicine concurs that there is “undoubtedly a rapid increase” in pyrethroid resistance in Africa and that “at some point we will get failure.”
However, according to Christian Lengeler, it is “probably right” that “we have already now some detrimental effect...” because of pyrethroid resistance. Lengeler is director of the health interventions unit at the Swiss Tropical and Public Health Institute. Together with WHO, Lengeler advocated for bednets in the mid-1990s. He also authored the influential meta-analysis of bednet trials in 2004 showing 5.5 lives could be saved for each 1,000 children covered by nets.
Uncertainty principle
Getting a grip on the actual effects of bednets is difficult. The Global Fund’s Filler gave a mixed message on whether a decline in effectiveness can be measured. “No – no such [study] design exists,” said Filler. But he then added: “This can be accomplished in carefully designed trials but these are complex, expensive and need a high level of epidemiological expertise to conduct….”
A study of bednets in the village of Dielmo in Senegal published last year showed nets rapidly reduced malaria when first introduced, consistent with many previous studies. However, within two years, 48 percent of mosquitos had a mutation for pyrethroid knockdown resistance, up from 8 percent at the beginning of the study. Cases of malaria rebounded to just below pre-bednet levels.
Controversy ensued. “This paper is bad,” Lengeler said of the Dielmo study. The study, Lengeler continued, “has no credit whatsoever in the malaria community.” A commentary accompanying the Dielmo study applauded the rigor of the research but cautioned against extrapolating its conclusions to the rest of Africa.
However, the authors of the commentary themselves produced a study just a few months earlier which appeared to show bednet failure. In Luangwa, Zambia, bednet use rose dramatically in two years from about half the population to 86 percent. However, malaria infections went up. Although the paper seemed to demonstrate some kind of failure, one of the authors, Thomas Eisele, wrote in email: "That is not accurate.” Eisele, of the Tulane University School of Public Health and Tropical Medicine, did not reply to subsequent requests to elaborate. He pointed instead to research from the Institute for Health Metrics & Evaluation (IHME) showing more favorable results.
Claims on nets overstretching evidence
The IHME study found that bednets were associated with a statistically significant reduction of mortality from any cause of 23%. However, the study did not examine the effect of insecticide treated nets (ITNs) on death from malaria. As the study authors pointed out, "we were only able to examine the relationship between ITNs and all-cause mortality as the surveys we used do not include information on cause-specific mortality."
However, a press release from IHME about the study used less cautious language, claiming: "researchers found clear evidence that bed nets reduce the number of child deaths from malaria." That statement did not appear in the peer-reviewed paper and is not supported by evidence in the paper.
IHME recently published a more comprehensive, exhaustive malaria mortality study. It made global headlines, reporting a higher death toll for malaria than previous estimates. Part of the difference came from a much larger estimate of deaths among adults. However, the study found that bednets did not reduce adult deaths from malaria in Africa: "coverage of insecticide-treated bednets," read the report, "was not a statistically significant predictor of African adult malaria mortality." But on the even larger question of whether bednets save children in Africa, the study is silent.
Asked that question in email, however, one of the paper’s authors, IHME’s Stephen Lim, replied that “ITNs [bednets] were a statistically significant predictor of African child mortality.” In other words, bednets worked to save the lives of the largest and most vulnerable group, children in Africa.
But the basis for this unpublished claim isn’t clear. The IHME study incorporates data from many smaller studies of particular geographic areas and then extrapolates as needed to country and continent levels using sophisticated, computationally-intensive modeling techniques. IHME actually generated many hundreds of models which were then averaged together into an ensemble to most closely approximate reality. However, Lim said IHME did not calculate an average hazard ratio for the effects of bednets. (A hazard ratio is a number that indicates whether an intervention increases or decreases risk, in this case the risk of dying from malaria.) “Analytically,' said Lim in email, "we can calculate an ‘average’ hazard ratio but it is not something we have currently in place and would involve a considerable amount of work.”
Not having a hazard ratio raises the question of how the statistical significance of bednets was assessed. IHME spokesperson William Heisel wrote in email that 131 models found bednet coverage to be a significant predictor of malaria mortality for children under five in Africa. However, at one point in the analysis, there are a total of 214 models for children under five in Africa. IHME did not reply to an email asking if this meant 131 models were and 83 models were not significant for bednets.
A greater number of models does not necessarily mean the variable being tested is statistically significant because models are weighted differently. IHME had earlier cautioned against simply counting the models in their list: “This list by itself," wrote Lim, "is not easily interpretable as different individual models are given more weight in generating the ensemble model.”
Asked whether IHME had based their assessment of statistical significance on a count of models, Heisel replied that IHME would not answer any more questions in email, .
Possible mistake?
Although difficult to countenance, distributing bednets in high transmission areas—like much of sub-Saharan Africa—might have been a mistake.
Intensity of malaria transmission worldwide. Darkest color indicates very high (>40 percent) infection prevalence and high transmission. Gething, et al., “A new world malaria map: Plasmodium falciparum endemicity in 2010” DOI: 10.1186/1475-2875-10-378
Where malaria is intense, being bitten is a kind of deadly hazing ritual with survival conferring a degree of immunity. In very young, non-immune children, malaria infection leads to fever—and possibly death. The fatality rate of malaria infections is, perhaps contrary to expectations, very low. Only an estimated 0.3 percent of infections globally cause death. But infections are so numerous that hundreds of thousands of children die each year. Children who survive, however, generally can better control infections later in life and even show no symptoms while carrying perhaps millions of parasites.
Research in the late 1990s concluded that “a critical determinant of life-time disease risk is the ability to develop clinical immunity early in life…” Malaria, including cerebral malaria and severe malaria, declined as children got older. Risk for severe malaria was highest where transmission was less intense, likely because people don’t acquire immunity without exposure to considerable infective biting.
Such natural tolerance is a mystery. There is no definitive set of biomarkers for it. And it’s no free pass: immunity may wane without some amount of continued infective biting, making severe disease a possibility.
One of the authors of the study, Robert Snow, now head of the public health group at the Kenya Medical Research Institute/Wellcome Trust Program, said recently, “I remain convinced that a certain degree of parasite exposure is required to develop functional immune responses to reduce risks of death and severe disease from malaria.”
Nets were originally targeted at children under five because most deaths from malaria occurred in that age range. But where malaria is intense, infection is unavoidable, with bednets deferring it to a later age. The age range least likely to sleep under a net is age 5 to 19. The most protected become the least protected—with potentially more adverse health consequences. Studies have found a shift in disease burden to older age groups following introduction of bednets. Trape and colleagues found this in Dielmo, Senegal. Other researchers, in an earlier 2009 paper, showed that nets reduced malaria risk in younger but not older children, a finding “consistent with older children having used [bednets] when they were younger, and therefore having acquired less immunity.” Thus to the extent bednets have saved lives in high transmission settings, they may also have created a population with reduced natural immunity, possibly setting the stage for a rebound of malaria.
“The issue of rebound and building up a time-bomb of susceptibles is interesting and you will find people willing to argue either side,” said Simon Hay, of Oxford University where he heads the Malaria Atlas Project.
The Global Fund’s Scott Filler said rebound concerns were “one major progenitor to move from targeted distribution of [bednets] to children under five to the goal of achieving universal coverage…” WHO switched to recommending universal coverage in 2007. (As the graph above shows, however, bednets distributed actually declined in 2011.)
Gerry Killeen of the Ifakara Health Institute in Tanzania believes rebound “is highly implausible unless the interventions themselves fail (resistance) or are withdrawn.” His parenthetical mention of resistance, however, could mean trouble. Azra Ghani, of Imperial College London, and colleagues concluded that “If the effectiveness of the intervention gradually wanes, the impact on immunity is likely to be minimal and the incidence of clinical disease will return to pre-intervention settings,” perhaps ten years later.
Diaphanous nets and ghosts of the past
Bednets were hoped to be a precise, stealthy intervention beneath the notice of mosquitos. But protecting even a portion of the population appears to have engaged their evolutionary attention. The switch to universal coverage also shifted the strategy: the purpose of the nets has become to kill mosquitos. “In order for their full potential to be realized,” reads the WHO position statement, bednets “should be deployed as a vector control intervention.”
However, if the goal was to knock transmission into an unrecoverable tailspin, it hasn’t worked. Transmission in high intensity areas dropped, but the force of infection still “needs a bit more help get it over a hump of stability that will impact on disease burden in the longer term,” said Robert Snow. The question is how because, continued Snow, the “expectation that [bednets] alone were to be the panacea in high transmission areas was misplaced.”
Resistance to DDT caused the technical failure of the mid-20th century effort to eradicate malaria. Nonetheless, over time, bednet policy has taken on a worrying semblance to this unsuccessful strategy. The previous effort didn’t even attempt to take on the heartland of malaria in sub-Saharan Africa, with some researchers arguing success was impossible using DDT. Also, planners of the oft-maligned effort actually knew resistance would be a problem where DDT was used and consequently they set a blitzkrieg timeline for achieving victory in a few years before resistance rose up. In comparison, the bednet strategy seems ad hoc and improvisational.
The coverage targets, initally 60 percent and then 80 percent, “were moved because we weren't going to meet them on time,” said David Smith of the University of Florida. “Instead of admitting nothing was happening, the intervention coverage target was increased and the date moved back—seeming to have some thought behind it, but mostly just saving face.”
Pyrethroid resistance was not part of the agenda at the Gates World Malaria Forum last October nor was it mentioned in congressional hearings on malaria last December. That omission may come from a concern, expressed by Killeen, “that doom-and-gloom stories will kill public enthusiasm for things that have saved many lives,” which he estimates to be half a million in Tanzania over the last five years. Scaling up such interventions “took a long time to get in place.” He concludes, “I am the father of two under fives and I live in a part of rural Tanzania where over 80 percent of people use [bednets] so this is a very real issue for me.”
Malaria is horrific, nature unsentimental. Sustainably reducing or ridding the disease from the world is unquestionably desirable. “But as we are now seeing,” say other researchers, controlling mosquitos with chemical killing agents comes at a price, “and the price is resistance.”