Fig. 3.

Entomological outcomes as predicted by bioassay survival. A Comparison of the level of pyrethroid resistance as measured using a discriminating dose bioassay and the percentage of mosquitoes which enter a hut with a pyrethroid-only ITN and survive 24 h after being collected. Solid lines show the fitted relationship using either the logistic (green) or log-logistic (yellow) models. Vertical and horizontal lines show 95% confidence interval estimates for each data point. B-C The relationship between EHT survival (24 h post-collection, unless the ITN incorporated the insecticide chlorfenapyr, in which case 72-h survival was used) and the probability of being caught inside a control hut relative to a hut with any type of ITN (where anywhere above the grey dashed-line indicates more mosquitoes were caught in the control hut) (B), mosquitoes successfully feeding and surviving (C), or exiting the hut without feeding (D). The point size in B-D is proportional to the total number of mosquitoes collected in the trial and coloured according to hut design: East (red) or West (blue). Solid lines in A-D show the best-fit relationship whilst the lighter shaded area indicates 95% credible intervals for the best-fit curves. E-F The models from A-D were combined to summarise how the average probability that blood-feeding mosquitoes will be killed, exit without feeding, deterred from entering or successfully blood-feed, varies with bioassay survival. The relationship between bioassay and EHT survival is highly uncertain, so both the logistic model (E) and the log-logistic model (F) are presented.