Integrating data from across insecticide resistance bioassay types using a mechanistic model incorporating mosquito genetic variation and behaviour

Abstract

AbstractMalaria kills roughly 500,000 people each year, and our most effective intervention against the disease is insecticide treated nets (ITNs) which kill mosquitoes. Mosquito resistance to the most important class of insecticidal compound, pyrethroids, has unfortunately grown dramatically over the past few decades, but gauging the impact of resistance on public health remains difficult. Past work has shown how mortality data from cheap-to-conduct discriminating-dose bioassays can be used to predict mosquito mortality in more expensive and more true-to-life experimental hut trials. Here, we develop a new predictive approach to modelling these data which incorporates data from intensity-dose bioassays, which is a WHO-recommended test increasingly used in malarial regions that outputs a more nuanced measure of resistance. This new mechanistic model explicitly estimates epidemiologically important quantities that describe the interaction of mosquitoes with the insecticide. The model accounts for variation in the lethal dose in the mosquito population, for example, due to genetic variability in field populations. The utility of the model is illustrated by fitting it to data from a systematic review of experimental hut trials evaluating the efficacy of pyrethroid-only ITNs. This work represents a first step towards using intensity-dose bioassay data to determine the effects of mosquito insecticide resistance on public health.Author summaryBednets laden with insecticides that kill mosquitoes have been responsible for substantial reductions in the burden of malaria over the last few decades. Mosquito resistance to insecticides, however, threatens to halt further progress and potentially erode these gains. It is crucial to be able to gauge changes in insecticide resistance over time and how these changes affect the effectiveness of bednets. Important tools for quantifying these changes include intensity-dose bioassays, which expose mosquitoes to a range of insecticide doses and measure their mortality, and experimental hut trials, which are more expensive and aim to mimic how mosquitoes interact with insecticides on bednets in the field. Here, we develop a mathematical model which includes mechanistic detail about how mosquitoes interact with insecticides in each of these types of experiment. We show how our models allow us to make accurate predictions of mosquito mortality in hut trials using data only from intensity-dose bioassays. Our models provide a more granular understanding of this important class of experiment and could be embedded into larger transmission dynamics models of malaria to predict the public health impact of measured changes to insecticide resistance.