Gottschald M, Knüppel S, Boeing H, Buijsse B
Department of Epidemiology, German Institute of Human Nutrition Potsdam-Rehbruecke, Nuthetal, Germany.
Eur J Clin Nutr. 2016 Nov;70(11):1318-1324. doi: 10.1038/ejcn.2016.131. Epub 2016 Jul 27.
BACKGROUND/OBJECTIVES: Previous cohort studies elucidated unexpected inverse relations of cake and cookie (CC) consumption with chronic disease risk. We assessed CC intake in relation to cardiometabolic disease risk factors in a well-phenotyped population with emphasis on misreporting as the potential driving force behind inverse relations.
SUBJECTS/METHODS: In a cross-sectional EPIC-Potsdam sub-study individual usual CC intake was modeled by combining 24 h recall and food frequency questionnaire data. Cardiometabolic risk factors were anthropometry, blood lipids, blood pressure (BP), physical activity and fitness. Analysis of covariance models adjusted for (i) age/education/lifestyle and (ii) additionally for energy misreporting (ratio of energy intake over energy expenditure) were used to compute mean values of risk factors for quartiles of CC intake.
Adjustment for misreporting had considerable impact on relations of CC intake. Initial inverse links with anthropometry were reversed to direct associations. Misreporting adjustment also nullified inverse relations with triglycerides and with total cholesterol in women. Negligible associations with high density lipoprotein cholesterol turned inverse (men: cross-quartile difference (Δ)=-1.7 mg/dl; women: Δ=-3.6 mg/dl), so did fitness (men: Δ=-1.2 ml/kg/min; women: Δ=-0.9 ml/kg/min). Direct relations with total/low density lipoprotein cholesterol in men were not changed by misreporting (Δ max. 7.5 or 11.3 mg/dl). Reduced BP was observed in females with increased CC intake; only systolic BP remained relevant after misreporting adjustment (Δ=-4.6 mmHg).
The strong impact of energy misreporting on relations of CC intake with risk factors emphasizes a careful analysis and interpretation of nutritional data. We showed that apparent favorable relations of CC intake changed with a different model specification, highlighting proper modeling considerations when analyzing diet-disease relations.
背景/目的：既往队列研究阐明了蛋糕和饼干（CC）摄入量与慢性病风险之间出人意料的负相关关系。我们在一个表型特征良好的人群中评估了CC摄入量与心血管代谢疾病风险因素的关系，重点关注误报情况，因为它可能是负相关关系背后的潜在驱动因素。
受试者/方法：在一项横断面的欧洲癌症与营养前瞻性调查（EPIC）-波茨坦子研究中，通过结合24小时回忆法和食物频率问卷数据对个体的CC通常摄入量进行建模。心血管代谢风险因素包括人体测量指标、血脂、血压（BP）、身体活动和体能。采用协方差分析模型，分别对（i）年龄/教育程度/生活方式以及（ii）另外对能量误报情况（能量摄入量与能量消耗量之比）进行校正，以计算CC摄入量四分位数对应的风险因素平均值。
对误报情况进行校正对CC摄入量与各因素之间的关系有相当大的影响。最初与人体测量指标的负相关关系转变为正相关。对误报情况进行校正后，与甘油三酯以及女性总胆固醇之间的负相关关系也消失了。与高密度脂蛋白胆固醇的微弱关联变为负相关（男性：四分位数间差值（Δ）=-1.7mg/dl；女性：Δ=-3.6mg/dl），体能方面也是如此（男性：Δ=-1.2ml/kg/min；女性：Δ=-0.9ml/kg/min）。男性中与总胆固醇/低密度脂蛋白胆固醇的正相关关系在对误报情况进行校正后未改变（Δ最大值为7.5或11.3mg/dl）。在CC摄入量增加的女性中观察到血压降低；在对误报情况进行校正后，仅收缩压仍具有相关性（Δ=-4.6mmHg）。
能量误报对CC摄入量与风险因素之间关系的强烈影响强调了对营养数据进行仔细分析和解读的重要性。我们表明，CC摄入量明显的有利关系会因不同的模型设定而改变，这突出了在分析饮食与疾病关系时进行恰当建模考量的必要性。
