Estimation of relative bioavailability of nutrients using SAS procedures.

The General Linear Models procedure (PROC GLM) in SAS/STAT software can be programmed to perform the standard statistical analyses used for relative bioavailability studies. The first steps are validity checks to test for statistical validity (linearity), fundamental validity (intersection of regression lines at 0 supplemental level), and equality of the basal diet mean to the point of intersection. The CLASS variable capabilities of PROC GLM can be exploited to expedite these tests. After the validity checks, the GLM procedure can be used to obtain parameter estimates for calculation of relative bioavailability. Optional output provides an inverse matrix to calculate standard errors of slopes and slope ratios. Logarithmic and other transformations of the dependent variable to reduce variance heterogeneity or achieve linearity for subsequent calculation of appropriate bioavailability values also can be accomplished within the SAS System. When nonlinear regression models are more appropriate than linear models, the NLIN procedure can be used.