Fig. 8From: How radical is radical cure? Site-specific biases in clinical trials underestimate the effect of radical cure on Plasmodium vivax hypnozoitesEffect of efficacy metric and infection endpoint on efficacy estimates. Efficacy estimates obtained from simulated clinical trials at different EIRs is shown when calculated using the Cox proportional hazards model, incidence rates or the proportion at risk. The infection endpoint was clinical (maroon), LM-detectable (red), or PCR-detectable (orange) recurrent infections identified during follow-up. Each point is the median of 200 simulations, and each bar is the interquartile range. The dotted line is the clearance probability simulated in each trialBack to article page