Invited commentary: dietary fiber, estradiol, and cholesterol.

The limitations of examining mediating factors by adjusting for them in standard regression models have been well-documented in the literature. Although alternative analytic models have been suggested, they are rarely used. In the accompanying article, Mumford et al. (Am J Epidemiol. 2010;173(2):145-156) use marginal structural linear mixed models to determine the association between dietary fiber intake and cholesterol through pathways that do not involve estradiol. Their findings suggest that overall high fiber intake decreases levels of total and low density lipoprotein (LDL) cholesterol and that there are multiple pathways through which fiber can act. The estradiol-mediated pathway seems to lead to increases in total and LDL cholesterol which are more than counterbalanced by pathways leading to decreases in total and LDL cholesterol. In addition to answering a scientifically interesting question, this work provides a concrete example of the use of marginal structural models for examination of direct effects and may serve as a guide to future researchers.