Inferences about the mean area under the curve in preclinical sparse sampling designs
Sparse sampling designs are commonly used in preclinical pharmacokinetic and toxicology studies. Although the individual area under the curve (AUC) is not acces- sible under the sparse sampling design, statistical inference regarding the mean AUC is still desirable. Traditionally, parametric methods are applied to perform inferences on the mean AUC based on assumed normality. Although such methods perform well with larger sample sizes, the preclinical studies of interest are often associated with small sample size and oftentimes highly skewed data. We investigated several methods to make inference on the mean AUC, which can be expressed in the form of the linear com- bination of log-normal distributed random variables. Inferential properties of our ap- proach are compared with those from standard methods of analysis using Monte Carlo simulation studies. The robustness of the methods are also evaluated when distribution assumptions are violated. The proposed methods are further examined in testing the equality and assessing the equivalence of two independent and correlated AUCs.