Alves Mariane A, Lotufo Paulo A, Benseñor Isabela, Marchioni Dirce Maria L
Department of Nutrition, School of Public Health, University of São Paulo, São Paulo, Brazil.
Center for Clinical and Epidemiological Research, University Hospital, University of São Paulo, São Paulo, Brazil.
Eur J Nutr. 2024 Jun;63(4):1203-1211. doi: 10.1007/s00394-024-03350-w. Epub 2024 Feb 16.
Combining different statistical methods to identify dietary patterns (DP) may provide new insights on how diet is associated with adiposity. This study investigated the association of DP derived from three data-driven methods and adiposity indicators over time.
This study used data from the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). DP were identified at baseline applying three statistical methods: Factor Analysis (FA), Treelet Transform (TT), and Reduced Rank Regression (RRR). The association between DP and adiposity indicators (weight, body mass index, waist circumference, body fat percentage and fat mass index) over the period of 8.2 years of follow-up was assessed by linear mixed-models.
Convenience DP, marked by unhealthy food groups, was associated with higher adiposity over the follow-up period, regardless of the method applied. The DP identified by TT and marked by high consumption of rice and beans was associated with lower adiposity, whereas the similar DP identified by FA, but additionally characterised by consumption of poultry and red meat was associated with higher adiposity. Prudent DP, marked by plant-based food groups and fish, identified by FA was associated with lower adiposity across the median follow-up time.
Applying different methods to identify DP showed that a convenience DP was associated with higher adiposity independent of the method applied. We also identified the nuances within adherence to a Brazilian traditional dietary pattern characterised by the consumption of rice and beans, that only when combined with reduced consumption of animal protein and unhealthy foods was associated with lower adiposity over time.
结合不同统计方法来识别饮食模式（DP）可能会为饮食与肥胖之间的关联提供新的见解。本研究调查了通过三种数据驱动方法得出的饮食模式与肥胖指标随时间的关联。
本研究使用了巴西成人健康纵向研究（ELSA - Brasil）的数据。在基线时应用三种统计方法来识别饮食模式：因子分析（FA）、小波变换（TT）和降秩回归（RRR）。通过线性混合模型评估在8.2年随访期间饮食模式与肥胖指标（体重、体重指数、腰围、体脂百分比和脂肪量指数）之间的关联。
以不健康食物组为特征的便利饮食模式，在随访期间与更高的肥胖程度相关，无论采用何种方法。通过小波变换识别且以大米和豆类高摄入量为特征的饮食模式与较低的肥胖程度相关，而通过因子分析识别的类似饮食模式，但还以家禽和红肉的摄入量为特征，则与更高的肥胖程度相关。通过因子分析识别的以植物性食物组和鱼类为特征的谨慎饮食模式在中位随访时间内与较低的肥胖程度相关。
应用不同方法识别饮食模式表明，便利饮食模式与更高的肥胖程度相关，且与所采用的方法无关。我们还发现了在遵循以大米和豆类消费为特征的巴西传统饮食模式中的细微差别，即只有当与减少动物蛋白和不健康食物的消费相结合时，随着时间的推移才与较低的肥胖程度相关。
