UC Irvine

Diet is central to the management of type 2 diabetes (T2D), yet the diet of the diagnosed U.S. population is rarely characterized by its degree of processing or by the balance of dietary protein against ultra-processed foods (UPFs). This dissertation examines UPF intake, protein intake, and a protein-to-UPF ratio motivated by the protein leverage hypothesis among U.S. adults with diagnosed T2D in seven cycles of the National Health and Nutrition Examination Survey, 2005 to 2018, using design-based methods and the NOVA classification.Aim 1 (Chapter 2) describes trends and cross-sectional patterns in 4,382 adults. UPFs supplied a majority of energy in every cycle (53.8% in 2005-2006 and 56.4% in 2017-2018), the protein share of energy fell from 17.3% to 15.7% while absolute protein intake was unchanged, and the UPF increase corresponded to a decrease in minimally processed foods. Higher UPF intake was associated with lower diet quality, higher adiposity, and more current smoking, but not with HbA1c, lipids, blood pressure, or attainment of American Diabetes Association goals. Aim 2 (Chapter 3) describes 4,200 adults grouped by HbA1c level. Glycemic control did not change over the period; 52.7% had HbA1c below 7.0% and 13.6% had HbA1c of 9.0% or higher. Adults at 9.0% or higher were younger, more often uninsured and food insecure, and more often had albuminuria, and about half were untreated or taking a single glucose-lowering class. UPF intake and diet quality did not differ by level of control. Aim 3 (Chapter 4) links 4,283 adults to the National Death Index (971 deaths over 27,990 person-years). Neither the protein-to-UPF ratio nor protein intake was associated with all-cause mortality. For UPF intake, the hazard ratio for the fourth versus first quartile was 1.18 (95% CI 0.93, 1.51), concentrated in deaths early in follow-up and in adults younger than 65 years. The three aims show that the diet of adults with diagnosed T2D follows the same trend toward ultra-processed foods as the general population, that HbA1c, lipids, and blood pressure do not differ by dietary intake, and that the protein-to-UPF ratio adds no prognostic information beyond its components. Improving diet in this population and increasing treatment among adults with high HbA1c are separate objectives.