Abstract
Malaria control programs implementing Long-Lasting Insecticidal Nets (LLINs) are encouraged to conduct field monitoring of nets' survival, fabric integrity and insecticidal bio-efficacy. The reference method for testing the insecticide activity of LLINs needs 100 two-to-five-day-old female mosquitoes per net, which is highly resource-intensive. We aimed at identifying an alternative protocol, using fewer mosquitos, while ensuring a precision in the main indicator of ±5 percentage points (pp). We compared different laboratory methods against the probability of the LLIN to fail the test as determined by a hierarchical Bayesian model. When using 50 mosquitoes per LLIN and considering mortality only instead of mortality or knock-down as validity criteria, the average error in the measure of the proportion of nets considered as valid was 0.40 pp. The 95% confidence interval of this value never exceed 5 pp when the number of LLIN tested was ≥40. This method slightly outperforms the current recommendations. As a conclusion, testing the bio-efficacy of LLINs with half as many mosquitoes provides a valid evaluation of the proportion of valid LLINs. This approach could increase entomology labs' testing capacity and decrease costs, with no impact in the decision process for public health purposes.