**10. Discussion**

#### *10.1. The Evidence for Divergent Resistance within the Pyrethroid Class*

Molecular analysis of metabolic resistance, together with analysis of phenotypic resistance in mosquito populations (including analyses of intensity data, of diagnostic dose bioassay data from populations that have been tested with multiple pyrethroids, and of spatiotemporal trends), provide evidence that there is strong cross-resistance among pyrethroids, particularly between permethrin and deltamethrin. P450s SAR (structure– activity relationship) findings concluded that the more commonly used pyrethroids examined were the most vulnerable to metabolic attack (by cytochrome P450s), while bifenthrin, <sup>λ</sup>-cyhalothrin, and α-cypermethrin were less vulnerable to metabolic attack. Bioassay data from *Aedes aegypti* and *An. sinensis* suggested that bifenthrin may demonstrate relatively low cross-resistance with other more commonly used pyrethroids. Bifenthrin has not been widely used in malaria control in Africa and no discriminating dose has been defined, but its potential use in malaria vector control warrants further investigation. There is also evidence that resistance to etofenprox could diverge from resistance to the more commonly used pyrethroids; however, further investigation regarding vulnerability to metabolic attack by P450s is required.

In field populations, variability in discriminating-dose and dose-response assay mortality was high. This variability was predominantly at a fine geographical scale (i.e., assays done within 50 km of each other were highly variable), indicating that if there were a difference between Type I and II pyrethroids, it would be very local and beneath the size of the regions to which insecticidal nets are currently allocated. There was good evidence that the mortalities from exposure to deltamethrin, permethrin, α-cypermethrin, and <sup>λ</sup>-cyhalothrin were strongly correlated across *An. gambiae* s.l. populations. These correlations were also seen for deltamethrin, permethrin, and <sup>λ</sup>-cyhalothrin ( α-cypermethrin was not tested) in the *An. funestus* subgroup, and in all three of the main malaria vectors within the *An. gambiae* complex.

#### *10.2. The Suitability of Current Testing Methods to Monitor Insecticide Resistance and Make Vector-Control Decisions*

Deployment decisions for ITNs are being guided by information arising from the discriminating-dose and resistance-intensity bioassays, but it is not clear how well differential mortality in WHO tube or CDC bottle bioassays predict how well an ITN treated with one or another pyrethroid will perform in a specific site. Bioavailability may play an important role in the relative efficacy of different ITNs, and testing for the relative performance of different nets against field populations would provide more directly relevant information for deployment decisions, alongside or in place of conventional bioassays. Given the limited products available for vector control, as well as narrow collection of available chemistries, programmes must make ITN deployment decisions based on the data that can realistically be collected. The current monitoring system for insecticide resistance is imperfect and should be adapted to make better use of the available resources, while being mindful that limited mosquito collections preclude the testing of all insecticides/products.

In general, following exposure of characterised lab strains in WHO tube bioassays under controlled conditions, intrastrain mortality to permethrin, deltamethrin, and αcypermethrin were similar. However, in intermediately resistant strains, some divergence in mortality rates was observed. However, importantly, the level of variability in observations of mortality between tubes (measured using standard deviations) was also greater in these intermediately resistant strains, which reduced certainty in apparent contrasts between insecticides. Discriminating-dose assays are poor tools for quantitative analysis of resistance levels where resistance is established, producing the most variability in results in laboratory colonies and field populations where resistance was moderate and mortality was intermediate, which is likely to be the case for all or most pyrethroids in most populations of African malaria vectors. For comparisons across insecticides, intensity assays (e.g., 1×, 5×, 10×) suffered from the same problem as the discriminating doses on which they depended—an apparent lack of parity across pyrethroid insecticides. Quantitative dose-response assays, which do not depend on a discriminating doses, are recommended for robust comparisons between insecticides.

All bioassays are vulnerable to ambient conditions, including humidity and temperature, in addition to other environmental effects more easily standardized by the user. It is crucial that deviations from optimal conditions are reported, along with the improved provision of sampling details, to understand extrinsic factors that could influence bioassay results.
