| Summary: | 碩士 === 臺灣大學 === 預防醫學研究所 === 98 === Background and objectives: Hyperuricemia is the risk factor for gout and insulin resistance. In Taiwan the prevalence of adult hyperuricemia are 26% and 17% in men and women respectively. Previous studies showed single food or nutrient affected serum uric acid, however, single food or nutrient are not consumed in isolation, in fact, in numerous different combinations that generated complex synergistic effects. The combined effects of nutrients or foods were observed through dietary patterns and the results from dietary patterns analysis are more helpful in disseminating diet-related information rather than related to single food or nutrient. No data showed the relationship between dietary patterns and serum level of uric acid among Chinese. We conducted the data-driven methods to explore the association between dietary patterns and hyperuricemia and compared to different methods for derived dietary patterns.
Methods and materials: We recruited adults who age older 35 years. The participants without uric acid and confounding factors data or energy intake beyond normal range (women: 500 kcal < total energy < 2842 kcal; men: 800 kcal < total energy < 3245 kcal) or energy intake beyond 3 standard deviation we were excluded. All participants had signed the informed consents. Derived dietary pattern methods are principle component analysis, partial least square and factor analysis, and then the factor score for each pattern is derived and used in regression analysis to test relationship between patterns and uric acid. In our study, we use discriminant analysis to cumulate the sensitivity and specificity of dietary patterns. Three dietary patterns are chose from factor analysis, and p value of the model fitness test is great 0.05.
Results: The relationship between three patterns and food, the contents of uric acid-prone pattern is seafood, meat, viscus, beverage, egg, fried food, and staple; fish pattern is only fish; Soy products, fruit, dark vegetable and white vegetable are included in vegetable and fruit pattern. Test the trend for uric acid by quartile of factor score, p for trend are not significance. Discriminant analysis results shows to add uric acid-prone pattern into model, specificity is 80%, and sensitivity is 52%.
Conclusion: We derived three dietary patterns from factor analysis. The association between dietary patterns and uric acid are no significance after adjusted confounding factors. However, BMI and gender are significant effects for uric acid. In our study, total energy is adjusted by modeling to avoid reduction of food variations.
|
|---|
