Methodological focus and application of exploratory dietary patterns in epidemiological research
Anita Carol Nash
MetadataShow full item record
Studying nutritional effects on health is an important piece of chronic disease epidemiology, but many challenges arise when attempting to adequately measure and account for complete dietary exposure. The exploratory approach to dietary patterns is a useful method that utilizes multivariate data reduction techniques to summarize overall dietary exposure into variables that can represent diet in observational studies. This dissertation presents two projects that focus on methodological considerations for the exploratory approach to dietary patterns. The first project compared different analytic methods used to derive dietary patterns, with and without data transformations. Findings from the project demonstrated that, with the use of appropriate data transformations, different analytic methods provide similar dietary patterns. The second project focused on the collapsing/grouping of dietary data prior to patterning to understand whether the number of food group variables alters the resulting patterns. Results demonstrated that dietary patterns were similar as food groups were collapsed to a smaller number of variables. A third project then utilized methods from the previous projects to explore diet’s association with A1C, a biomarker of glucose control, and examined the possible role of diet in the race/ethnic disparities in A1C. Findings demonstrated that non-Hispanic blacks and Mexican-Americans have higher A1C levels compared to non-Hispanic whites, and while a dietary pattern characterized by higher intakes of dark breads is associated with lower A1C levels, this dietary pattern and other lifestyle characteristics such as body mass index and physical activity did not appreciably explain race/ethnic differences in A1C levels.