Models for measuring anthelmintic drug efficacy for parasitologists

• Modern statistical methods offer robust ways to estimate anthelmintic drug efficacy.
• Marginal models permit robust inference on population covariates of drug efficacy.
• Mixed models allow inference on individuals and quantify inter-individual variation.
• Models may be used to detect changing drug efficacy during mass drug administration.

Anthelmintic drug efficacy (ADE) is generally estimated as a population average effect, despite drug responses varying among individuals according to a variety of measurable and non-measurable factors. Model-based and/or individual-level analyses are scarce and often methodologically frail. We propose that wider application of marginal and mixed models would offer benefits to the evaluation of ADE. We demonstrate, with a worked example, how model-based analyses: (i) capture the effects of correlation among hierarchically structured longitudinal data on estimates of ADE; (ii) permit robust inference on the association of measurable factors with ADE; and (iii) enable estimation of variation among individual-level estimates of ADE. The application of modelling approaches is discussed in the context of mass drug administration-based control of human helminthiases.