Abstract
BACKGROUND: The 'overall insecticidal effect' is a key measure used to evaluate public health pesticides for indoor use in experimental hut trials. It depends on the proportion of mosquitoes that are killed out of those that enter the treated hut, intrinsic mortality in the control hut, and the ratio of mosquitoes entering the treatment hut to those entering the control hut. This paper critically examines the way the effect is defined, and discusses how it can be used to infer effectiveness of intervention programmes. FINDINGS: The overall insecticidal effect, as defined by the World Health Organization in 2006, can be negative when deterrence from entering the treated hut is high, even if all mosquitoes that enter are killed, wrongly suggesting that the insecticide enhances mosquito survival. Also in the absence of deterrence, even if the insecticide kills all mosquitoes in the treatment hut, the insecticidal effect is less than 100%, unless intrinsic mortality is nil. A proposed alternative definition for the measurement of the overall insecticidal effect has the desirable range of 0 to 1 (100%), provided mortality among non-repelled mosquitoes in the treated hut is less than the corresponding mortality in the control hut. This definition can be built upon to formulate the coverage-dependent insecticidal effectiveness of an intervention programme. Coverage-dependent population protection against feeding can be formulated similarly. CONCLUSIONS: This paper shows that the 2006 recommended quantity for measuring the overall insecticidal effect is problematic, and proposes an alternative quantity with more desirable properties.