Major patterns of dietary intake in adolescents: identification, stability over time, socio-demographic and socio-environmental correlates, and association with obesity.

University of Minnesota Ph.D. dissertation. October 2010 Major: Epidemiology. Advisors: Andrew Flood, Ph.D., Dianne Neumark-Sztainer, Ph.D., M.P.H., R.D. 1 computer file (PDF); ix, 267 pages, appendices A-G.Purpose: Few studies have used data-driven dietary pattern analysis in adolescents, but it can be a useful method to summarize dietary intake. This dissertation had three aims: 1) describe the major patterns of dietary intake in a cohort of ethnically and socio-economically diverse adolescents, examine the stability of these patterns over a five-year period, and study the dietary profile of the identified patterns, 2) examine the socio-demographic and socio-environmental correlates of adherence to the indentified dietary patterns with specific regard to the dietary profile of these patterns (healthier vs. unhealthy), and 3) study the relationship between adherence to these healthier and unhealthy patterns and weight status. Methods: Data from the longitudinal Project EAT (Eating Among Teens) study were used in all analyses. Project EAT-I (Time 1), collected data on 4746 middle school (younger cohort) and high school (older cohort) students during the 1998-1999 academic year. Project EAT-II (Time 2) resurveyed 53% (n=2516) of the original cohort in 2003-2004. Dietary intake was assessed using the Youth/Adolescent Food Frequency Questionnaire at both time points. Principal components factor analysis was used to identify dietary patterns at Time 1 and Time 2. Multivariable linear regression was used to examine the relationship between socio-demographic and socio-environmental characteristics and factor scores for the identified dietary patterns. Multivariable logistic regression was used to examine the relationships between scores for each dietary pattern and risk for overweight/obese weight status. All models were run separately by age cohort and gender. Results: Four dietary patterns were identified at Time 1: vegetable, fruit, starchy food (e.g., mashed potatoes, pancakes), and sweet & salty snack food. Similar patterns were identified at Time 2, with the exception of a new `fast food' pattern. Multiple socio-demographic and socio-environmental characteristics were found to be significantly associated with adherence to healthier and unhealthy patterns of dietary intake. Socio-economic status, family meal frequency, healthy home food availability, and parental and peer support for healthy eating were positively associated with adherence to the healthier patterns, and inversely associated with the unhealthy patterns. The opposite relationships were seen for availability of unhealthy food in the home. Inverse associations were seen between the healthier vegetable and fruit patterns and overweight/obese weight status in girls, while inverse associations between the unhealthy `sweet & salty snack food' pattern and overweight/obese status were found in boys. Conclusion: Dietary patterns were identified in this adolescent population that were analogous across gender and age cohorts, and were relatively similar over time, with the exception a new "fast food" pattern identified at Time 2. The nutrient profile showed that the identified dietary patterns reflect intake of important nutrients and food groups, underscoring the value of this method to summarize dietary data in adolescent populations. Multiple correlates of dietary patterns were identified in adolescents in both cross-sectional and prospective analyses, including modifiable characteristics that may be possible targets for dietary interventions. Consistent or intuitive associations were not found between dietary patterns and weight status. Identified patterns may not capture the elements of diet that are truly important in determining adolescent weight, or diet may not be the primary driver in determining weight status at this age