Institute for Physical Activity and Nutrition (IPAN), School of Exercise and Nutrition Sciences, Deakin University, Geelong, Victoria, Australia.
Epidemiology Program, University of Hawaii Cancer Center, Honolulu, HI, USA.
Int J Behav Nutr Phys Act. 2021 Mar 25;18(1):46. doi: 10.1186/s12966-021-01115-w.
The patterning of food intake at eating occasions is a poorly understood, albeit important, step towards achieving a healthy dietary pattern. However, to capture the many permutations of food combinations at eating occasions, novel analytic approaches are required. We applied a latent variable mixture modelling (LVMM) approach to understand how foods are consumed in relation to each other at breakfast.
Dietary intake at breakfast (n = 8145 occasions) was assessed via 24-h recall during the 2011-12 Australian National Nutrition and Physical Activity Survey (n = 3545 men and n = 4127 women, ⩾19 y). LVMM was used to determine breakfast food profiles based on 35 food group variables, reflecting compliance with Australian Dietary Guidelines. F and adjusted-chi2 tests assessed differences in timing of consumption and participant characteristics between the breakfast profiles. Regression models, adjusted for covariates, were used to examine associations between breakfast food profiles and objective adiposity measures (BMI and waist circumference).
Five distinct profiles were found. Three were similar for men and women. These were labelled: "Wholegrain cereals and milks" (men: 16%, women: 17%), "Protein-foods" (men and women: 11%) and "Mixed cereals and milks" (men: 33%, women: 37%). Two "Breads and spreads" profiles were also found that were differentiated by their accompanying beverages (men) or type of grain (women). Profiles were found to vary by timing of consumption, participant characteristics and adiposity indicators. For example, the "Protein-foods" profile occurred more frequently on weekends and after 9 am. Men with a "Bread and spreads (plus tea/coffee)" profile were older (P < 0.001) and had lower income and education levels (P < 0.05), when compared to the other profiles. Women with a "Protein-foods" profile were younger (P < 0.001) and less likely to be married (P < 0.01). Both men and women with a "Wholegrain cereals and milks" profile had the most favourable adiposity estimates (P < 0.05).
We identified five breakfast food profiles in adults that varied by timing of consumption, participant characteristics and adiposity indicators. LVMM was a useful approach for capturing the complexity of food combinations at breakfast. Future research could collect contextual information about eating occasions to understand the complex factors that influence food choices.
在进食时对食物摄入量进行模式化处理是实现健康饮食模式的一个尚未被充分理解但非常重要的步骤。然而，为了捕捉进食时食物组合的多种变化，需要采用新的分析方法。我们应用潜在变量混合模型（LVMM）方法来了解早餐时食物之间的相互关系。
通过 24 小时回忆法在 2011-12 年澳大利亚国家营养和身体活动调查中（男性 n=3545 人，女性 n=4127 人，年龄≥19 岁）评估早餐的饮食摄入情况。LVMM 用于确定基于 35 种食物组变量的早餐食物图谱，这些变量反映了对澳大利亚饮食指南的遵守情况。使用 F 检验和调整后的卡方检验评估不同早餐图谱之间的消费时间和参与者特征的差异。调整协变量后，回归模型用于研究早餐食物图谱与客观肥胖指标（BMI 和腰围）之间的关联。
发现了五个不同的图谱。其中三个图谱在男性和女性中相似。它们分别标记为：“全谷物谷物和牛奶”（男性：16%，女性：17%）、“蛋白质食品”（男性和女性：11%）和“混合谷物和牛奶”（男性：33%，女性：37%）。还发现了两种“面包和涂抹酱”图谱，它们通过所伴随的饮料（男性）或谷物类型（女性）来区分。图谱发现因消费时间、参与者特征和肥胖指标而异。例如，“蛋白质食品”图谱更多地出现在周末和 9 点以后。与其他图谱相比，“面包和涂抹酱（加茶/咖啡）”图谱的男性年龄更大（P<0.001），收入和教育水平更低（P<0.05）。年轻（P<0.001）和未婚（P<0.01）的女性具有“蛋白质食品”图谱。具有“全谷物谷物和牛奶”图谱的男性和女性的肥胖估计值最有利（P<0.05）。
我们在成年人中确定了五种早餐食物图谱，它们因消费时间、参与者特征和肥胖指标而异。LVMM 是捕捉早餐时食物组合复杂性的有用方法。未来的研究可以收集有关进食时的情境信息，以了解影响食物选择的复杂因素。
