Empirical and theoretical investigation into the potential impacts of insecticide resistance on the effectiveness of insecticide-treated bed nets

Abstract

In spite of widespread insecticide resistance in vector mosquitoes throughout Africa, there is limited evidence that long-lasting insecticidal bed nets (LLINs) are failing to protect against malaria. Here, we showed that LLIN contact in the course of host-seeking resulted in higher mortality of resistant Anopheles spp. mosquitoes than predicted from standard laboratory exposures with the same net. We also found that sublethal contact with an LLIN caused a reduction in blood feeding and subsequent host-seeking success in multiple lines of resistant mosquitoes from the laboratory and the field. Using a transmission model, we showed that when these LLIN-related lethal and sublethal effects were accrued over mosquito lifetimes, they greatly reduced the impact of resistance on malaria transmission potential under conditions of high net coverage. If coverage falls, the epidemiological impact is far more pronounced. Similarly, if the intensity of resistance intensifies, the loss of malaria control increases nonlinearly. Our findings help explain why insecticide resistance has not yet led to wide-scale failure of LLINs, but reinforce the call for alternative control tools and informed resistance management strategies.

Keywords: Anopheles; insecticide resistance; malaria; transmission.

Figure 1

Mortality of mosquito strains exposed to an long‐lasting insecticidal bed net (LLIN) in cone assays or in a free‐ranging laboratory trials. When five groups of five females were exposed to an Olyset LLIN for 3 min in WHO cone assays, few were killed. When fifty females were released into a 1.5 × 1.5 × 2 m enclosure and allowed to interact freely with the LLIN placed over a human host, we observed significantly greater mortality (p < .05). This was true regardless of Anopheles species or strain: (a) and (b) An. arabiensis, (c) An. funestus, (d) An. gambiae. Bars show mean values ± SEM. Gray bars show mortality from exposures to an untreated net, while the blue bars show mortality for an LLIN

Figure 2

Blood feeding success and mortality following long‐lasting insecticidal bed net (LLIN) exposure. Two groups of ten mosquitoes were exposed to untreated netting or PermaNet 2.0 for 1 min (SENN‐DDT), 5 min (FUMOZ‐BASE), or 10 min (FUMOZ‐R). (a) Immediately after LLIN exposure, mosquitoes were offered access to a host for 5 min. LLIN‐exposed mosquitoes were significantly less likely to take a blood meal. (b) Survival was assessed 24 hr later. Bars show mean values ± SEM; all comparisons significantly different, p < .05. Gray bars show results for an untreated net, and the colored bars (red for feeding and blue for mortality) show results for the LLIN

Figure 3

Host‐seeking in resistant An. arabiensis (SENN‐DDT) exposed to untreated netting or an long‐lasting insecticidal bed net (LLIN). At about one, seven, or 24 hr after exposure to untreated netting or PermaNet 2.0, individual females were tested for their ability to locate a host over a short distance (~0.15 m) within two minutes. Bars in (a) show mean values ± SEM. Shortly after LLIN exposure, SENN‐DDT females were (a) less likely to find a host and those that found the host (b) did so more slowly. By 24 hr after exposure, exposed females were still less likely to find a host, but those able to locate a host were just as fast as unexposed females. Gray bars/lines show results for an untreated net, and the blue bars/lines show results for the LLIN

Figure 4

Host‐seeking in resistant An. funestus (FUMOZ‐R) exposed to untreated netting or an long‐lasting insecticidal bed net (LLIN). At about one, seven, or 24 hr after exposure to untreated netting or PermaNet 2.0, individual females were tested for their ability to locate a host over a short distance (~0.15 m) within two minutes. Bars in (a) show mean values ± SEM. Immediately after LLIN exposure, FUMOZ‐R females were (a) less likely to find a host and those that found the host (b) did so more slowly. Exposed females recovered their short‐range host location ability with seven hours of exposure, though they remained slower than unexposed females for up to 24 hr. Gray bars/lines show results for an untreated net, and the blue bars/lines show results for the LLIN

Figure 5

Host‐seeking in field‐caught, resistant Anopheles spp. from Mozambique exposed to untreated netting or an long‐lasting insecticidal bed net (LLIN). At about one or six hours after exposure to untreated netting or PermaNet 2.0, individual females were tested for their ability to locate a host over a short distance (~0.15 m) within two minutes. Bars in (a) show mean values ± SEM. Immediately after LLIN exposure, field‐caught females were (a) less likely to find a host and those that found the host (b) seemed to do so more slowly, though only one LLIN‐exposed female responded to host cues. Exposed females recovered their short‐range host location ability with seven hours of exposure, although they remained slower than unexposed females. Gray bars/lines show results for an untreated net, and the blue bars/lines show results for the LLIN

Figure 6

Experimental hut trial in an area with resistant An. gambiae outside of Bouake, Côte d'Ivoire. Experimental huts were outfitted with an artificially damaged untreated net or long‐lasting insecticidal bed net (LLIN) (PermaNet 2.0). On five consecutive mornings, mosquitoes were collected and their location and blood feeding status recorded. Mortality was recorded 24 hr later. Bars represent mean values ± SEM for the proportion of mosquitoes collected attempting to exit the hut, dead within the hut, and/or with a blood meal. Proportions do not total to one, as the categories are not mutually exclusive. Females that entered a hut with an LLIN were more likely to be found in the veranda of the hut (i.e., exiting the hut) than inside the hut. They were also more likely to be killed, but less likely to take a blood meal (p < .05). Gray bars show results for an untreated net, and the blue bars show results for the LLIN

Figure 7

Changes in relative transmission potential for combinations of assumed probabilities of long‐lasting insecticidal bed net (LLIN)‐generated mortality and feeding impairment per feeding attempt. Panels show the relative transmission potential (RTP) of a mosquito population exposed to various levels of LLIN coverage, compared to the RTP if no bed nets were present. Axes represent the probability, during each feeding attempt on an LLIN‐protected host, of prebite mortality (x‐axis) or feeding impairment (y‐axis) caused by the LLIN. Proportion killed plus proportion impaired cannot exceed 100%, so plots are only generated in range x + y ≤ 1