Dieleman, JosephChakrabarti, Suman2022-07-142022-07-142022-07-142022Chakrabarti_washington_0250E_24225.pdfhttp://hdl.handle.net/1773/48681Thesis (Ph.D.)--University of Washington, 2022Suboptimal diets are important preventable risk factors for non-communicable diseases and mortality. Broadly, evidence from a plethora of systematic reviews suggests that food groups including whole grains, fibrous vegetables, fruits, nuts, legumes, red-meat, and processed meat, are linked to all-cause mortality. Despite the high quality evidence that has emerged from meta-analyses, high levels of heterogeneity reported across most systematic reviews presents challenges for rigorous estimation of average effects. It is also well established that suboptimal diets are important risk factors for high body mass index and elevated systolic blood pressure, which are leading metabolic risk factors for mortality. Moreover, evidence indicates that dietary patterns and mortality rates differ systematically across racial/ethnic groups in the US. Making progress towards reducing risks of mortality that stem from suboptimal diets requires a systematic and sequential approach that leverages learnings in one area as inputs to shed light on other emerging areas. Unpacking the interlinkages between dietary patterns, biological mediators and social factors will enable a holistic line of attack that will empower clinicians, researchers, and policy makers to take effective steps to address alarming emerging trends in morbidity and mortality in this century.To address these research priorities, this study sought to pool secondary individual-level data from seven US based cohorts including Atherosclerosis Risk in Communities Study, Cardiovascular Health Study, Framingham Heart Study-Offspring, Multi-Ethnic Study of Atherosclerosis, Women's Health Initiative Study, NHANES I Epidemiologic Follow-up Study, Reasons for Geographic and Racial Differences in Stroke. Pooling several cohorts supported a systematic and coherent analysis that consistently accounted for potential confounders and addressed differences in exposure measurement. Across the cohorts, interviews were conducted in-person at regular examinations, follow-up telephone calls to participants or next of kin, state records, linkage to the National Death Index, surveillance of medical records, death certificates, obituaries, Centers for Medicare and Medicaid Services. A total of 127,279 individuals were followed-up for 1,579,299 person-years in the seven cohorts combined. Principal Component Analysis of observed intake of food groups was used to estimate dietary patterns. Five dietary patterns were estimated. The ‘Western’ pattern was high in red and processed meats, fats/oils, starchy vegetables, and low in yogurt, fruits and vegetables. The ‘Prudent’ pattern was high in fibrous vegetables, fruits, legumes, starchy vegetables, whole grains; and low in eggs, fats and processed meat. The ‘Fatty-grainy’ pattern was high in nuts and seeds, cheese, whole grains, fats/oils and low in poultry. The ‘Lacto’ pattern was high in milk and moderate to low on all other food groups. The ‘Carnivore’ pattern was high in red and processed meats, poultry, fish, eggs and moderate on other food groups. Dietary patterns were subsequently employed in estimating associations of dietary patterns with all-cause mortality, unpacking direct and indirect relationships with metabolic mediators, and explaining differences in mortality between racial/ethnic groups. First, the risk to all-cause mortality associated with commonly occurring dietary patterns in the US was estimated using Cox proportional hazard models. Second, the direct and indirect effects of dietary patterns on all-cause mortality through body mass index and systolic blood pressure were disentangled using methods including the percentage of excess risk mediated and percent mediated measure. Third, the roles of dietary patterns and other behavioral risk factors to close racial/ethnic differences in all-cause mortality in the US were decomposed using the Oaxaca-Blinder decomposition technique and population attributable fractions. In seven large prospective cohorts with 8.5 to 36.7 median years of follow-up, greater adherence to Western and Carnivore patterns was consistently associated with higher risk of early mortality, while moderation eating patterns and those rich in plant foods lowered this risk. These findings support the recommendations of the Dietary Guidelines for Americans that multiple healthy eating patterns can be adapted to individual food traditions and preferences. Further, while interventions that reduce BMI and blood pressure, might address up to a third of the mortality risk associated attributed to suboptimal diets, maintenance of optimal diets is needed for complete health benefits. Finally, replacement of unhealthy diets with healthy ones could yield significant equitable improvements in mortality rates across all racial/ethnic groups in the US. Income disparities currently explain half the mortality differences between White and Black individuals in the US. Holistic improvements in income, physical activity, smoking and drinking are needed across all race/ethnic groups for further equitable reductions in mortality rates.application/pdfen-USCC BYall-cause mortalitydietary patternsdisparitynutritionobesityracePublic healthNutritionEpidemiologyGlobal HealthThesis