Department of Statistics Palmetto Lecture, Part 1Wednesday, March 22, 2017 - 1:00pm
Location: LeConte College, Room 210
Speaker: Raymond Carroll, Texas A&M University, Department of Statistics
Topic: Building dietary index scores that are informed by disease data using data integration
Abstract: Nutritionists interested in diet and health (cancer, cardiovascular disease, etc.) have developed what I call Global Healthy Diet Indices. I will focus on the Healthy Eating Index – 2005 (HEI-2005), which is an index of 12 component scores for a healthy diet that was built to conform to the USDA 2005 Dietary Guidelines. This total score, with a minimum possible value of 0 (not really possible, even Americans cannot get a 0) to 100 (impossible as well, no one is perfect), is then used to related diet to many health outcomes: various cancers, morbidities and mortality, via logistic and Cox regression.
The various dietary indices are built on knowledge of the biology of diet and health, but are actually not informed by how well they predict disease in data. I show how to construct a new score that is informed by diseases, and can be used as a global index. The technical aspects of the problem are extremely interesting to statisticians only, so that part, while in evidence, will not be emphasized. The differences in the scores that are based on cancer data, and those based on a thought exercise, are interesting. I will also contrast global indices with indices built for individual disease (colorectal cancer, lung cancer), and show that, while interesting, such “if I am worried about disease A, what should I do, but what about disease B?” differ wildly across disease, and are of little public health interest.
Sponsored by the College of Arts and Sciences and the Department of Statistics