Background
Insecticide-treated nets (ITNs) are an important tool to protect individuals against the morbidity and mortality caused by malaria [
1‐
3]. The distribution of long-lasting insecticidal nets (LLINs) (factory-treated ITNs designed to retain insecticidal activity for up to three years [
4]) by governments, non-governmental organizations (NGOs) and donors, has resulted in a dramatic increase in their ownership and contributed to the decline in malaria burden since 2000 [
5]. Ownership and use of ITNs within households, as measured by the number of children under five years of age reported to have used an ITN the previous night, increased by three to tenfold between 2000 and 2008 in many African countries [
5‐
7]. According to the latest World Malaria Report, 41% of children and 33% of all persons residing in malaria-endemic regions of sub-Saharan Africa reported sleeping under an ITN [
5]. The distribution of LLINs in malaria endemic areas of Kenya has increased household ownership of any net to 70% and household ownership of an ITN to 60% [
8]. The increase in ownership and use of ITNs has contributed to a significant decline in the burden of malaria throughout sub-Saharan Africa [
5], including Kenya [
9,
10].
All mosquito nets act as a physical barrier, preventing access to vector mosquitoes and thus providing personal protection against malaria to the individual(s) using the nets [
11]. However, untreated nets with even a few holes provide little protection [
12]. The addition of pyrethroid insecticides serves to enhance the effectiveness of intact mosquito nets and to extend the effectiveness of nets with holes. The pyrethroids used to treat the ITNs have an exito-repellent effect, thus adding a chemical barrier to the physical one and enhancing personal protection by nets [
11,
13]. The insecticides incorporated in the ITNs kill malaria vectors that come into contact with them and when used by a majority of the target population, may provide protection for the entire community, including those who do not themselves sleep under an ITN [
14‐
16]. A meta-analysis of data from trials of treated and untreated nets suggested that approximately half of the protection was derived from the physical barrier of the net and half from the chemical barrier [
17].
In western Kenya, malaria transmission in the lowland areas around Lake Victoria has historically been very high, with entomological inoculation rates estimated to be as high as 300+ infectious bites per person per year [
18‐
20]. However, the scaling up of ITNs and other malaria control tools has reduced transmission and malaria-associated morbidity and mortality. Between 2003 and 2007, demographic surveillance indicated a 42% reduction in all-cause mortality among children less than five years of age coinciding with a scale up of ITNs as well as improvement in diagnostics and introduction of ACT [
21].
Anopheles gambiae s.l. and
Anopheles funestus densities declined markedly in a randomized trial of permethrin-treated bed nets in a comparison of treatment
versus control villages in western Kenya [
22,
23]. Subsequent monitoring demonstrated a decline in the proportion of
An. gambiae s.s.
, the principal vector of malaria, relative to
Anopheles arabiensis[
24].
However, a resurgence in malaria vectors, parasite prevalence and malarial disease burden has been observed in several sites in western Kenya despite high ownership of ITNs [
25,
26]. This resurgence could be due to one or a combination of the following factors: reduced efficacy of ITNs, insecticide resistance in mosquitoes, improper use of ITNs, stock-outs of anti-malarial drugs or even a poor dosing regimen of policy recommended drugs by private outlets [
25,
27]. This study investigated the impact of high levels of pyrethroid resistance in
An. gambiae on their tendency to enter and rest inside nets. The results suggest that pyrethroid resistance in mosquitoes may undermine the effectiveness of nets with even a few holes. These findings have serious implications for malaria control programmes in sub-Saharan Africa where resistance to pyrethroid insecticides is increasing rapidly [
28].
Discussion
In Bungoma, an area with high levels of pyrethroid resistance [
30,
36], live
Anopheles mosquitoes were routinely observed resting inside nets. This may have been due to declining bioefficacy of the nets, reduced susceptibility of the mosquitoes to pyrethroid insecticides or both. Susceptibility testing using females reared from field-collected larvae or from the f1 generation derived from mosquitoes collected inside nets confirmed high resistance to pyrethroids. Average knockdown and mortality of a susceptible strain of mosquitoes exposed to field-collected nets was greater than 90% regardless of whether the mosquitoes had been observed resting inside, indicating that the nets in the field generally had adequate levels of insecticide: the average mortality rates were well above the 80% threshold recommended by WHOPES as a criteria for a functional ITN/LLIN [
31]. In contrast, the mortality of f1 adult mosquitoes raised from adult mosquitoes collected inside nets in Bungoma and exposed to unused, unwashed Olyset and PermaNet 2.0 nets was 37.5 and 22%, respectively. Several mosquitoes collected inside the nets were positive for
P. falciparum sporozoite infection.
In contrast to Bungoma, no mosquitoes were observed inside 213 nets in Gem. Although no formal analysis was done, there were several differences between the two sites. The main difference was that the
Anopheles population in Gem is comprised largely of
An. arabiensis. This species has lower resistance to pyrethroid insecticides [
30,
36], which may have reduced the likelihood that mosquitoes would enter an ITN and survive exposure. This species is also more likely to feed on alternative hosts compared to
An. gambiae s.s., which may allow it to persist in presence of high coverage of ITNs [
24].
Anopheles gambiae s.s. historically was very common in Gem and surrounding areas and it is still present at low levels. It has been shown to be resistant to pyrethroid insecticides (unpublished data) and it is not clear why its numbers have not rebounded or why none were observed resting inside nets. It is possible that the resistance mechanism or intensity of resistance in
An. gambiae in Gem is different from that of Bungoma, where mass LLIN campaigns were first conducted in 2011. Alternatively, there may be additional ecological constraints, which in combination with the widespread use of ITNs, result in the continued suppression of the population of
An. gambiae s.s. in Gem.
Pyrethroid resistance has been spreading rapidly in sub-Saharan Africa and has been documented in 23 countries [
28]. This may partly be in response to agricultural application and run-off of insecticides into mosquito breeding sites [
38‐
40], but increasingly in response to selection pressure resulting from the scale up of insecticide-treated nets and indoor residual spraying as malaria prevention tools [
4,
36,
41‐
45]. Regardless of the source of insecticide pressure, insecticide resistance in malaria vectors has been predicted to eventually undermine control programmes that are solely reliant on insecticides such as indoor residual spraying (IRS) and ITN programmes [
28]. While pyrethroid resistance has been documented in malaria vectors throughout sub-Saharan Africa, there is surprisingly little information on the impact of resistance on the effectiveness of vector control efforts. An experimental, hut trial in two sites in Benin, one with susceptible mosquitoes and the other with resistance to pyrethroids, showed blood-feeding was reduced by 96% at the site with susceptible vector population, but was largely unaffected at the site with high levels of pyrethroid resistance, while the mortality of mosquitoes entering huts at the susceptible site was nearly three times as high as that at the site with high levels of pyrethroid resistance [
46]. Household trials in other parts of Benin also showed that sleeping under an ITN in an area with resistant mosquitoes was no more protective than sleeping under an untreated net, regardless of its physical condition [
47]. During a longitudinal study of inhabitants of Dielmo village, Senegal, a rise in the incidence of malaria following the distribution of LLINs was attributed to increasing pyrethroid resistance in the local vector populations [
48]. In contrast, a study in Ivory Coast found no reduction in the protective efficacy of ITNs in an area with high levels of pyrethroid resistance [
41], while in Malawi, increasing pyrethroid resistance in
An. funestus was not associated with an increase in malaria transmission in areas with LLINs although in areas with IRS, no additional impact was observed [
49]. In Benin, mosquitoes were collected from inside nets with 12 holes that were 4 cm × 4 cm. Insecticide treated nets reduced the number of mosquitoes entering compared to an untreated net but an average of 5 mosquitoes were collected each night under LLINs [
50]. A modeling study to measure the effect of pyrethroid resistance on the cost effectiveness of LLINs showed strong, positive correlations between insecticide susceptibility status and predicted population level insecticidal effectiveness of and protection against blood feeding by LLIN intervention programmes [
51]. With the most resistant mosquito population, LLIN mass distributions would avert up to 40% fewer episodes of malaria compared to areas with a fully susceptible population [
51]. An ongoing study in western Kenya shows prospects of generating evidence within the next year or two on the impact of insecticide resistance on the efficiency of malaria control interventions (Mbogo, pers. comm).
Several factors associated with the number of mosquitoes inside nets were explored. As described above, the location was a strong determinant of the presence of mosquitoes inside nets, presumably due to the composition of the local vector population, and further analyses included only Bungoma. In that site, neither net brand nor the age of the nets was associated with the number of mosquitoes inside nets. Although the nets were not stratified by age, the study demonstrated high mortalities of susceptible mosquitoes exposed to nets collected from the field indicating that most nets had adequate levels of insecticide. An increase in the number of mosquitoes inside nets with increasing levels of physical damage was however, observed. Nets with estimated hole areas of >50 cm2 had more mosquitoes than nets with no holes. Although the sample sizes were limited, the data suggested that a threshold is reached beyond which increasing damage does not lead to increasing numbers of mosquitoes. This may indicate that beyond a certain amount of damage, nets are equally likely to be penetrated by mosquitoes. However, the possibility that increasing damage also allows for more mosquitoes to escape from nets, which may also account for the apparent threshold effect, cannot be ruled out. Interestingly, nets with no holes had an average of just over one mosquito per net. Presumably, this was due to improper usage and residents should be instructed on how to tuck their nets in to prevent mosquitoes from entering them.
It has been suggested that the physical integrity of the LLINs may be compromised before the insecticidal activity falls below established thresholds indicating the need for replacement [
49,
50] and multiple reports have documented physical damage to nets under conditions of routine use. Rehman
et al. noted that 39, 24 and 63% of all the nets in use in Bioko Island, continental Equatorial Guinea and Malawi, respectively, were holed within two years of distribution [
52]. During a long-term assessment of a polyester-based LLIN in Uganda, more than 70% of nets had holes after only one year and more than 85% after two years [
53]. Wills
et al., reported 54.5% of nets having holes after just six months of distribution in Ethiopia [
54]. In Kenya, in an ongoing net durability study in western Kenya, it was observed that up to 40% of some net types had holes within six months of deployment in Siaya County (Bayoh, pers. comm.) while some recent surveys reported that up to 74% of the bed nets in use in Kwale County had holes [
55,
56]. The WHOPES guidelines on monitoring the durability of LLINs outline methods to estimate the hole sizes on the net fabric [
31]. However, the guidelines do not provide criteria for physical damage that is indicative of net failure and requiring the replacement of the nets. Mutuku
et al. proposed a pHI of 88 corresponding to approximately 500 cm
2 of damage [
55] while Gnanguenon
et al. observed mosquitoes entering nets with 12 holes 4 cm × 4 cm corresponding to a proportionate hole indexes (pHI) of 276 [
50,
55]. Several authors have suggested criteria based upon pHI and the probability that owners will discontinue use due to the owners’ perception that the nets are no longer effective [
57,
58]. The cut-off for an unacceptable net ranged from a pHI of 300 in Chad corresponding to a hole area of approximately 1,000 cm
2 to a pHI of 764 in Ethiopia which corresponded to a hole area of approximately 1,200 cm
2. In studies in Bioko Island and Malawi, the risk of malaria increased with deteriorating condition of nets with untreated nets with at least one hole providing the least protection [
52] although specific thresholds for net replacement were not presented. The data suggest that in areas with high levels of pyrethroid resistance, the threshold for a net requiring replacement may be at the lower end of the spectrum. While there is complex relationship between hole area and insecticidal activity of the nets and insecticide resistance and behaviour of the vector population, specific criteria for the physical integrity of nets should be developed to assist national malaria control programmes in determining the appropriate replacement strategies for LLINs.
The spread of pyrethroid resistance combined with increasing evidence that it may compromise malaria vector control programmes highlights the need for new insecticides and new tools for malaria prevention. Currently, LLINs are treated with pyrethroid insecticides only and their loss as an effective tool would seriously undermine malaria control programmes throughout sub-Saharan Africa. IRS with non-pyrethroid insecticides is an option that is immediately available. However, IRS is expensive relative to LLINs, particularly when spraying is done with non-pyrethroids, and is unlikely to be widely implemented without a significant increase in the amount of funding available for malaria control programmes. Two new LLIN products are currently available that incorporate a synergist to mitigate the effects of pyrethroid resistance. The Permanet 3.0 is treated with deltamethrin on the sides and deltamethrin plus piperonyl butoxide (PBO) on top. PBO is a synergist and increases the potency of the pyrethroid insecticides by inhibiting oxidase enzymes that have been implicated as one mechanism of resistance [
59]. Evidence that the PermaNet 3.0 is more effective than the PermaNet 2.0 which is treated with deltamethrin alone, however, is limited and occasionally mixed, presumably due to the presence of other resistance mechanisms that are unaffected by PBO and the WHO Pesticide Evaluation Scheme did not recommend this product for use as a resistance management tool [
60]. The Olyset Plus is another bi-treated net with permethrin plus PBO throughout the net [
33]. However, there is limited data on the efficacy of this net against wild populations of mosquitoes that are resistant to pyrethroid insecticides. Larviciding is an alternative vector control tool with a wide range of activity that are recommended for use against malaria vectors. However, as with IRS, larviciding can be expensive and is currently only recommended for specific settings. Other insecticides such as chlorfenapyr, indoxacarb [
46], and diafenthiuron [
61] are being investigated as options for IRS but it is likely to be several years before commercially available formulations will be available. Spatial repellents [
62] and toxic sugar baits [
63] have also been proposed for malaria prevention but these too require several years of evaluation and refinement before they can be considered viable tools for malaria control programmes.
While this study has demonstrated that pyrethroid resistant mosquitoes are entering and surviving exposure to LLINs, the results should not be interpreted to indicate that LLINs are no longer useful in malaria control programmes. First, while pyrethroid resistance is widespread, the intensity of resistance in many areas is likely low and in these areas LLINs may still be effective. The lack of mosquitoes collected inside nets in Gem demonstrates that LLINs are not compromised everywhere and differences in the intensity of resistance, as well as the effectiveness of LLINs, may vary over relatively short distances. Second, this study was a cross-sectional survey and differences in the age of mosquitoes may affect their susceptibility to pyrethroid insecticides as older mosquitoes have been shown to be more susceptible. Additionally, mosquitoes may repeatedly encounter insecticides over their life and, although this has not been demonstrated, the cumulative exposure may eventually result in the death of the mosquitoes. However, the finding of P. falciparum-infected mosquitoes inside nets suggests that, at least in Bungoma, older mosquitoes are able to survive exposure to treated nets. Lastly, intact untreated nets still provide some protection and there may be community-wide effects where malaria transmission is reduced when most people in the population regularly sleep under nets. Without baseline data on the effectiveness of LLINs before the rise of pyrethroid resistance, the impact of resistance on the effectiveness of LLINs cannot be reliably measured. However, the data strongly suggest that the efficacy of pyrethroid-treated nets may be compromised in areas with high levels of pyrethroid resistance.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
EO, JEG, NMB, MO, BA, CO, AK, JV, YG and EDW designed the study, developed the study and took part in the manuscript preparation. EO, JEG and NMB contributed to development of the protocol and data analysis. BA and EO performed the laboratory analysis of the samples. All authors read and approved the final manuscript.