Carbohydrate intake and obesity: an association that needs "refining".
Carbohydrate intake and obesity: an association that needs "refining".
复制标题
碳水化合物摄入量和肥胖：需要“细化”的关联。
DOI：
10.1016/j.jada.2009.04.016
复制
发表时间：
2009
影响因子：
--
通讯作者：
Liu,Simin
中科院分区：
文献类型：
--
作者：
Roberts,ChristianK;Liu,Simin
即使在过去的半个世纪里进行了大量的研究，关于饮食与长期健康结果的关系的各个方面仍然充满争议。早期有人提出，脂肪摄入(1)，特别是饱和脂肪(2)，在冠状动脉疾病的发生中起着重要作用，因此，用碳水化合物代替脂肪是预防冠状动脉疾病的首选。然而，在过去的几十年里，美国的饮食变化主要是由于精制碳水化合物摄入量的增加（3,4）。即使是已故的低脂饮食的著名支持者安塞尔·凯斯（Ancel Keys），也赞成“更自然的营养混合物”，而不是“高度精制的碳水化合物食物——糖和白面粉”。直到20世纪70年代，Burkitt和Trowell才在他们的经典著作(6)中系统地将西方世界许多慢性病模式的差异归因于植物细胞壁（即膳食纤维）的丧失。这些营养概念随着由多伦多大学的Jenkins及其同事首创的升糖指数的发展而进一步完善(7)，他们将不同碳水化合物的体内升糖效应纳入单一的营养质量指数。不同碳水化合物在其化学结构之外的鉴定为膳食碳水化合物的不同代谢效应和长期健康结果的应用和研究奠定了基础（3,8,9）。然而，在确定长期健康的最佳膳食成分方面仍存在许多困惑。在这期杂志上，Merchant和他的同事通过分析加拿大社区健康调查收集的数据，调查了碳水化合物摄入量与体重和肥胖的关系。这项横断面饮食调查的二次分析主要基于单一的24小时回忆。大约30年前，加拿大营养学领域的另一位杰出人物乔治·比顿教授（George Beaton）博士证明，每天的饮食摄入量变化很大，单次24小时的回忆根本无法提供任何可靠的估计个人通常的摄入量（11,12）。意识到研究设计中固有的这一主要限制，Merchant及其同事在选择参与者进行分析时付出了很大的努力。例如，碳水化合物摄入量最高的人往往是女性，或者消耗较少的能量，这两个变量通常是高度相关的。排除患有糖尿病或合并症的参与者，以及后来的分析额外排除了那些食用“高蛋白饮食”或“高碳水化合物饮食”的人，以解释由于这些适应症引起的混淆，这是可以理解的，但在横断面设置中可能适得其反，因为这些饮食实际上可能仍然是具有这些条件的参与者的常规饮食。作者在数据分析中强加的高度选择性的纳入和排除标准（即，从20,197名符合条件的调查参与者中选出4,451人，仅占22%）使得他们对加拿大成年人口的研究结果进行条件推断也不合理。另一种方法可以是分析那些饮食、身高和体重数据可用的人的整个数据集（至少9801人），这样读者就可以根据这些情况评估饮食碳水化合物和体重之间的关系是否确实存在真正的横截面差异。此外，这种情况本可以通过询问参与者在患糖尿病或试图减肥后是否改变饮食来解决。这些事情……
Even with a tremendous body of research conducted during the past half century, controversy still abounds regarding various aspects of diet in relation to long-term health outcomes. Early on it was proposed that fat intake (1), specifically saturated fat (2), plays a significant role in the development of coronary artery disease and, as such, replacing fats with carbohydrates is preferable for coronary artery disease prevention. However, much of the dietary change in the United States during the past few decades has been due to an increase in refined-carbohydrate intake (3, 4). Even the late Ancel Keys, a prominent proponent of low-fat diets, was in favor of “more natural mixtures of nutrients” rather than “highly refined carbohydrate foods—sugar and white flour”(5). It was not until the 1970s that Burkitt and Trowell, in their classic book (6), systematically attributed differences in many chronic disease patterns to the loss of plant-cell walls (ie, dietary fibers) in the Western world. These nutrition concepts were further refined with the development of the Glycemic Index pioneered by Jenkins and colleagues from the University of Toronto (7), who incorporated the in vivo glycemic effects of different carbohydrates in a single nutritional quality index. The qualification of different carbohydrates beyond their chemical structure set the stage for the application and investigation of different metabolic effects of dietary carbohydrates and long-term health outcomes (3, 8, 9). Nevertheless, much confusion remains in determining the optimal dietary composition for long-term health.In this issue of the Journal, Merchant and colleagues (10) investigate carbohydrate intake in relation to body weight and obesity by analyzing data collected in the Canadian Community Health Survey. This secondary analysis of a cross-sectional dietary survey is based predominately on a single 24-hour recall. Approximately 30 years ago it was another luminary of the nutrition field from Canada, Professor George Beaton, PhD, who demonstrated that dietary intake is extremely variable day to day and a single 24-hour recall simply cannot provide any reliable estimate for an individual’s usual intake (11, 12). Realizing this major limitation inherent in the study design, Merchant and colleagues took great pains in selecting participants to be analyzed. For example, those with the highest carbohydrate intake tended to be either women or consume less energy, two variables that are generally highly correlated with one another. The exclusion of participants who had diabetes or comorbidities, and a latter analysis additionally excluding those who consumed “high protein diets” or “high carbohydrate diets” to account for confounding due to these indications, is understandable but may be counterproductive in a crosssectional setting, because the possibility remains that these diets might in fact be the usual diets of participants with these conditions. The highly selective inclusion and exclusion criteria imposed by the authors for data analysis (ie, 4,451 from 20,197 eligible survey participants, a mere 22%) makes even conditional inference of their findings to the Canadian adult population unjustifiable. An alternate approach could have been to analyze the entire dataset of those whose diet, height, and weight data were available (at least 9,801), so readers could evaluate whether there were indeed real cross-sectional differences in the relationships between dietary carbohydrates and body weight according to these conditions. Also, this situation could have been addressed by asking participants whether they were changing their diets after they developed diabetes or attempted to lose weight. These matters …