SCIENTIFIC STATEMENT FROM THE AMERICAN SOCIETY FOR NUTRITION. Susan S Cho, 5 Lu Qi, 6 George C Fahey Jr, 7 and David M Klurfeld 8*

FROM THE AMERICAN SOCIETY FOR NUTRITION SCIENTIFIC STATEMENT Consumption of cereal fiber, mixtures of whole grains and bran, and whole grains and risk reduction in type 2 diabetes, obesity, and cardiovascular disease 1 4 Susan S Cho, 5 Lu Qi, 6 George C Fahey Jr, 7 and David M Klurfeld 8* 5 NutraSource, Clarksville, MD; 6 Department of Nutrition, Harvard School of Public Health, Boston, MA; 7 Department of Animal Sciences, University of Illinois, Urbana, IL; and 8 USDA/Agricultural Research Service, Beltsville, MD. ABSTRACT Background: Studies of whole grain and chronic disease have often included bran-enriched foods and other ingredients that do not meet the current definition of whole grains. Therefore, we assessed the literature to test whether whole grains alone had benefits on these diseases. Objective: The objective was to assess the contribution of bran or cereal fiber on the impact of whole grains on the risk of type 2 diabetes (T2D), obesity and body weight measures, and cardiovascular disease (CVD) in human studies as the basis for establishing an American Society for Nutrition (ASN) position on this subject. Design: We performed a comprehensive PubMed search of human studies published from 1965 to December 2010. Results: Most whole-grain studies included mixtures of whole grains and foods with $25% bran. Prospective studies consistently showed a reduced risk of T2D with high intakes of cereal fiber or mixtures of whole grains and bran. For body weight, a limited number of prospective studies on cereal fiber and whole grains reported small but significant reductions in weight gain. For CVD, studies found reduced risk with high intakes of cereal fiber or mixtures of whole grains and bran. Conclusions: The ASN position, based on the current state of the science, is that consumption of foods rich in cereal fiber or mixtures of whole grains and bran is modestly associated with a reduced risk of obesity, T2D, and CVD. The data for whole grains alone are limited primarily because of varying definitions among epidemiologic studies of what, and how much, was included in that food category. Am J Clin Nutr 2013;98:594 619. INTRODUCTION The prevalence of type 2 diabetes (T2D) 9 and obesity has been increasing in the United States (1 3). Today, two-thirds of US adults (1) and nearly 1 in 3 children are overweight or obese (4). Intake of calories over physiologic needs has been linked to increasing T2D, obesity, cardiovascular disease (CVD), and other chronic conditions (5). The 2005 and 2010 Dietary Guidelines for Americans (DGA) recommend that whole grains account for at least half of 6 to 11 daily grain servings (6, 7) to reduce the risk of chronic diseases, including T2D, obesity, and heart disease. The NHANES 1999 2004 found that the daily consumption of whole-grain servings was less than one-third of recommended intakes: only 1.5% of children aged 1 3 y, 4.3% of adolescents, 4.8% of adults aged 19 50 y, and 6.4% of adults.50 y meet the recommended 3 5 servings per day (8, 9). The National Health Interview Survey 2000 showed that the average American consumes only one-half of the recommended intake of dietary fiber;.84% of American adults do not meet recommended intakes established by the Institute of Medicine (10, 11). Increased consumption of whole grains and cereal fiber rich foods such as bran can contribute to 1 This statement was peer reviewed and approved by the American Society for Nutrition s (ASN s) Reviews, Papers, and Guidelines Committee and approved by ASN s Board of Directors. The statement did not undergo editorial peer review by the editors of The American Journal of Clinical Nutrition. 2 This project was funded by an unrestricted educational grant from CJ CheilJedang, Garuda International, and the Kellogg Company. 3 The opinions expressed in this article are those of the authors and not necessarily those of the USDA, the Agricultural Research Service, or any of the authors affiliations. 4 Author disclosures: SS Cho, research support from Garuda International, Innophos, JRS, Kellogg, Roquette; consultancy for CJ CheilJedang, Corn Products International, Cyvex Nutrition, Optimum Nutrition, Pacific Rainbow, RFI Ingredient, and Shangdong Longlive. L Qi, no disclosures to report. GC Fahey, research support from Roquette America Inc, Ingredion, Abbott Nutrition, Del Monte, and Hartz Mountain; consultancy for Ingredion, Novus lnternational, Procter & Gamble, Perfect Companion Group, Pronaca, Nova Foods, NuPec, Dae Han Feed Co Ltd, Milk Specialties Co, and Watt Publishing Co. DM Klurfeld, was a member of Unilever North America Scientific Advisory Board at the time this work was performed. 9 Abbreviations used: CVD, cardiovascular disease; DGA, Dietary Guidelines for Americans; FDA, US Food and Drug Administration; NHS, Nurses Health Study; RCT, randomized controlled trial; T2D, type 2 diabetes; WC, waist circumference. *Address correspondence to DM Klurfeld, USDA/Agricultural Research Service, Human Nutrition Program, 5601 Sunnyside Avenue, Beltsville, MD 20705-5138. E-mail: david.klurfeld@ars.usda.gov. First published online June 26, 2013; doi: 10.3945/ajcn.113.067629. 594 Am J Clin Nutr 2013;98:594 619. Printed in USA. Ó 2013 American Society for Nutrition

FROM THE AMERICAN SOCIETY FOR NUTRITION WHOLE-GRAIN, BRAN, AND FIBER INTAKE AND RISK OF DISEASE 595 improved fiber intake by Americans. The 2010 DGA recommend whole grains as a source of dietary fiber but do not mention other sources of dietary fiber. Furthermore, the DGA do not recommend dietary fiber to reduce the risk of chronic diseases despite the conclusion statements in the USDA Nutrition Evidence Library that whole grains and cereal fiber intake are associated with reduced risk of obesity and CVD (7, 12). A change in the working definition of whole grains has contributed to the confusion on the health effects of whole grains and their components. Research conducted by nutritional epidemiologists from the University of Minnesota and Harvard University provided much of the early data in this area; however, these investigators used different definitions of a whole grain in their studies (13 15). For example, some studies included brown bread and wheat germ in the whole-grain category and others did not (16, 17). In addition, there were no USDA databases available at that time to calculate how much whole grain was in a serving of the various products, and the studies provided no information that this was accounted for. But the most problematic inclusion in the whole-grain category was bran cereals. In the 1990s, foods with $25% whole grains and bran, as well as high-fiber bran cereals, were included in the whole-grain working definition (13 15). Today, high-fiber bran cereals are excluded (18, 19). Definitions of whole grain vary depending on the context and purpose. Research from the 1990s (13 15) supported the benefits of bran as an integral part of the health benefits of whole grains. According to this definition established in the 1990s, foods with $25% whole-grain or bran content by weight were classified as whole grain. Of note, however, the definition commonly used in the scientific community is not the same as that on food labels, which exclude foods with $25% bran. To establish a whole-grain health claim under the Food and Drug Administration Modernization Act, the definition of a whole-grain food is one that contains $51% whole-grain ingredients by weight per reference amount customarily consumed (18, 19). In 2006, the US Food and Drug Administration (FDA) provided additional guidance that describes specific sources of whole grains (20). Accordingly, examples of whole-grain foods and ingredients in the Expert Panel report of the USDA/Health and Human Services DGA include brown rice, oatmeal, whole oats, bulgur (cracked wheat), popcorn, whole rye, graham flour, and whole wheat (21). This list includes some low-fiber foods but does not include bran-rich foods that are excellent sources of fiber. Compared with refined-grain foods, foods made with whole grains are purported to contain fewer starches and calories and more micronutrients and phytochemicals that may offer significant health advantages (22). Whole grains are composed of bran, germ, and starchy endosperm. Bran is a concentrated source of fiber, vitamins, minerals, and phytonutrients, which together are thought to provide many of the health benefits of whole grains (22). The use of bran-concentrated or -enriched foods likely can provide consumers with more choices for healthy foods, promote healthful dietary practices, and help individuals meet public health recommendations for the intake of whole grains and dietary fiber without consumption of excess calories. FIGURE 1. Flow diagram of systematic review on type 2 diabetes.

596 CHO ET AL As the basis of an American Society for Nutrition position statement, this systematic review seeks to determine whether bran, cereal fiber, and whole-grain intakes have an effect on the development of T2D, obesity, and heart disease. METHODS Approach and methodology A comprehensive PubMed (http://www.ncbi.nlm.nih.gov/ pubmed) search of the scientific literature for articles published from January 1965 to December 2010 was performed. For T2D, the following search strategy was used: (whole grain OR whole grains OR whole-grain OR cereal fiber OR cereal fiber OR grain fiber OR grain fiber OR bran OR brans) AND (diabetes). This strategy identified 614 human studies. Twenty additional records were identified through other sources. After removing duplicates and nonrelevant articles, 66 abstracts were selected for more detailed review. Finally, we selected 15 prospective studies and 13 cross-sectional studies that met predetermined inclusion and exclusion criteria described below and shown in Figure 1 (23). For obesity and measures of body weight, the PubMed search of the following terms whole grain OR whole grains OR wholegrain OR cereal fiber OR cereal fiber OR grain fiber OR grain fiber OR bran OR brans AND body weight OR body weight gain OR body mass index OR BMI z score OR obesity OR overweight OR adiposity OR waist circumference OR waist-to-hip ratio (WHR) identified 538 articles. Seven additional records were identified from other sources, such as review articles. After removing duplicates and nonrelevant articles, 43 studies were reviewed in detail. From these, 19 studies (5 prospective and 14 cross-sectional) were selected on the basis of our inclusion and exclusion criteria described below and as shown in Figure 2 (23). For CVD, search terms (whole grain OR whole grains OR whole-grain OR cereal fiber OR cereal fiber OR grain fiber OR grain fiber OR bran OR brans) AND (heart disease OR cardiovascular disease OR myocardial infarction OR hypertension OR arterial disease OR stroke OR blood pressure) identified 894 human studies. Fifteen additional papers were identified through other sources. After removing duplicates and nonrelevant articles, 85 abstracts were reviewed, from which we selected 22 prospective and 12 cross-sectional studies by using the inclusion and exclusion criteria described below and as shown in Figure 3 (23). When assessing the contribution of whole grains alone, our analysis included only those studies that used the current definition of a whole grain as established by the FDA (ie, foods that contain $51% whole-grain ingredients by weight per reference amount customarily consumed). Inclusion criteria We included controlled feeding trials (intervention period of $1 y) and observational studies reporting risk and risk factors of FIGURE 2. Flow diagram of systematic review on obesity. RCT, randomized controlled trial.

FROM THE AMERICAN SOCIETY FOR NUTRITION WHOLE-GRAIN, BRAN, AND FIBER INTAKE AND RISK OF DISEASE 597 FIGURE 3. Flow diagram of systematic review on cardiovascular disease (CVD). T2D or CVD. For obesity outcomes, we used risk of obesity or weight-related endpoints, such as BMI, weight gain, waist circumference (WC), and waist-to-hip ratio. Exclusion criteria Editorials, meta-analyses, reviews, and studies published in languages other than English were excluded. Articles related to total, soluble, and insoluble fibers with no data on cereal fiber were excluded. Short-term feeding studies (,1 y) were also excluded because they do not reflect the long-term health impact of these grain components as consumed by the population. Rating scientific evidence We used the evidence grading system shown in Table 1. This system is similar to that used by prominent organizations such as the American Diabetes Association (24) or recommended by other experts (25). The assigned grade reflects the totality of the evidence on a particular subject and was established by consensus of the writing group. RESULTS A total of 15, 5, and 22 prospective cohort studies were available for summarization for T2D, obesity/body weight measures, and CVD, respectively. Cross-sectional studies (13 reports on T2D, 14 reports on obesity, and 12 studies on CVD) are also presented as supportive data in Appendixes A K. We found no long-term ($1 y duration), randomized controlled trials (RCTs) that assessed the impact of cereal fiber, mixtures of whole grains or brain, or whole grains that represent intakes of these grain components as consumed by the population. Thus, all of the studies identified were observational, either prospective or cross-sectional. The majority of whole-grain studies have investigated mixtures of whole grains and $25% bran rather than whole grains alone. Many studies reported that the inverse associations between intake of mixtures of whole grains and bran or whole grains alone and risks of T2D, CVD, or reduced weight gain disappeared or were attenuated after adjustment for cereal fiber or bran, suggesting that cereal fiber and bran account for much of the whole-grain effects. T2D A number of well-designed, large, prospective cohort studies showed a consistent inverse association between consumption of cereal fiber or mixtures of whole grains and bran and the risk of T2D. Prospective studies consistently showed a reduced risk of T2D with high intakes of cereal fiber (10 of 11 reports showed an 18 40% risk reduction) or mixtures of whole grains and bran (5 of 6 reports showed a risk reduction of 21 40%); one prospective study on whole grains (meeting the FDA definition excluding $25% bran) reported a risk reduction of 32 37% in females.

598 CHO ET AL TABLE 1 Evidence grading system Level of evidence (refers to the body of evidence) Description A: strong Clear evidence from at least one large, well-conducted, generalizable, RCT 1 that is adequately powered with a large effect size and is free of bias or other concerns or Clear evidence from multiple RCTs or many controlled trials that may have few limitations related to bias, measurement imprecision, inconsistent results, or other concerns B: moderate Evidence obtained from multiple, well-designed, conducted, and controlled prospective cohort studies that have used adequate and relevant measurements and that gave similar results from different populations or Evidence obtained from a well-conducted meta-analysis of prospective cohort studies from different populations C: limited Evidence obtained from multiple prospective cohort studies from diverse populations that have limitations related to bias, measurement imprecision, or inconsistent results or have other concerns or Evidence from only one well-designed prospective study with few limitations or Evidence from multiple well-designed and conducted cross-sectional or case-controlled studies that have very few limitations that could invalidate the results from diverse populations or Evidence from a meta-analysis that has design limitations D: inadequate Evidence from studies that have one or more major methodologic flaws or many minor methodologic flaws that result in low confidence in the effect estimate or Insufficient data to support a hypothesis or Evidence derived from clinical experience, historical studies (before-after), or uncontrolled descriptive studies or case reports 1 RCT, randomized controlled trial. Studies on cereal fiber The results of 11 reports are listed in Table 2: 10 reports from 8 independent prospective cohorts, each of which reported quintile or quartile analysis of cereal fiber consumption (16, 17, 26 33), and one additional study based on regression analysis (34). All of the studies using quintile or quartile analysis of cereal fiber consumption showed consistent associations between intake of cereal fiber and risk reduction of T2D (16, 17, 26 33). The Melbourne Collaborative Cohort Study was the only epidemiologic report that failed to show an association between cereal fiber and risk of T2D (34). The Melbourne Collaborative Cohort s primary purpose was to study the relation between glycemic index and risk of T2D. Fiber and cereal fiber concentrations were reported for each quartile of glycemic index. Relative risk from cereal fiber was assessed on the basis of a 10-g/d intake, not fiber intake quartiles. Most of the studies followed groups of mixed ethnicity, but Krishnan et al (33) reported that cereal fiber intake was inversely associated with risk of T2D, with an incidence rate ratio of 0.82 (95% CI: 0.70, 0.96; P-trend = 0.01) in a prospective cohort study in 59,000 US black women. A stronger association was seen among black women with a BMI (in kg/m 2 ),25: the incidence rate ratio for the highest compared with lowest quintile was 0.41 (95% CI: 0.24, 0.72; P-trend = 0.003). Nine-year follow-up in the Atherosclerosis Risk in Communities Study (32) found that cereal fiber intake was inversely associated with the risk of T2D in whites and African Americans, but the relation was significant only in whites (HR: 0.956; 95% CI: 0.925, 0.987 for 1 g cereal fiber/d). The weaker association in African Americans may be due to fewer individuals and lower statistical power or a smaller difference in cereal fiber intake amounts between the highest and the lowest consuming groups. Data from cross-sectional studies are consistent with the findings from prospective studies (Appendix A; 35 39). Evidence level B was assigned for the association between cereal fiber intake and risk reduction of T2D. Studies on mixtures of whole grains and bran Most of the studies (16, 26, 35, 36, 40 50) on whole grains included foods with $25% bran in the whole-grain category. Six prospective cohort studies on whole-grain intake and reduction in risk of T2D are summarized in Table 3 (16, 26, 40 43). All of these investigations (6 of 6 studies) followed different populations and included added bran in the whole-grain food category; thus, whole grains in these studies were mixtures of whole grains and bran (16, 26, 40 43). One study (16) also included wheat germ in the whole-grain definition (Table 3). The study by Kochar et al (41) was limited to whole-grain breakfast cereals. These studies showed a significant inverse association between intake of mixture of whole grains and bran and the incidence of T2D.

TABLE 2 Prospective cohort studies of cereal fiber: risk of T2D 1 FROM THE AMERICAN SOCIETY FOR NUTRITION WHOLE-GRAIN, BRAN, AND FIBER INTAKE AND RISK OF DISEASE 599 (reference) Study name participants Age Follow-up cases Highest Cereal fiber intake Lowest RR 95% CI P-trend de Munter et al, 2007 (17) Meyer et al, 2000 (26) Montonen et al, 2003 (16) Salmerón et al, 1997 (27) Hu et al, 2001 (28) Schulze et al, 2004 (29) Salmerón et al, 1997 (30) Schulze et al, 2007 (31) Stevens et al, 2002 (32) Krishnan et al, 2007 (33) Hodge et al, 2004 (34) y y g/d NHS I and II 161,737 F 36 65 12 18 6486 12.0 1.1 0.64 0.54, 0.76,0.001 IWHS 35, 988 F 55 69 6 1141 9.43 2.66 0.64 0.53, 0.79 0.0001 0.71 2 0.56, 0.89 0.002 FMCHES 4316 M + F 40 69 10 156 29 6 11 12 6 3.4 0.39 0.20, 0.77 0.01 NHS 66,173 F 40 65 6 523 7.5 2.0 0.72 0.58, 0.90 0.001 NHS 84,941 F 30 55 16 3300 NA NA 0.60 0.53, 0.67,0.001 NHS II 91,249 F 24 44 8 741 8.8 3.1 0.64 0.48, 0.86 0.004 HPFS 42,759 M 40 75 6 915 10.2 2.5 0.70 0.51, 0.96 0.007 EPIC-Potsdam study 25,067 M + F 35 65 7 844 16.6 6.6 0.72 0.56, 0.93 0.02 ARIC study 12,251 M + F (W and AA) A few studies were able to analyze the effects of cereal fiber, whole grains, and/or bran in the same population. The associations of mixtures of whole grains and bran were attenuated or disappeared after adjustments for other dietary factors such as dietary fiber or cereal fiber/magnesium (16, 26, 40). These data suggest that dietary fiber and magnesium account for much of the risk reduction associated with intake of mixtures of whole grains and bran. In the study by Meyer et al (26), women in the highest intake quintile of mixtures of whole grains and bran had an adjusted RR of 0.79 (95% CI: 0.65, 0.96; P-trend = 0.0089) for T2D relative to those in the lowest quintile. Cereal fiber and dietary magnesium, 2 components of whole grains, were strongly related to T2D diabetes: RRs in the highest quintiles were 0.64 (95% CI: 0.53, 0.79; P-trend = 0.0001) for cereal fiber and 0.67 (95% CI: 0.55, 0.82; P-trend = 0.0003) for dietary magnesium. The association of mixtures of whole grains and bran was not significant after the models were adjusted for cereal fiber and magnesium (RR: 0.93; 95% CI: 0.75, 1.16; P-trend = NS). Cereal fiber and dietary magnesium had significant inverse relations with T2D, even after simultaneous adjustment for grains, and cereal grains: RRs were 0.71 (95% CI: 0.56, 0.89; P-trend = 0.0017) in the highest quintiles of cereal fiber and 0.76 (95% CI: 0.62, 0.95; P-trend = 0.048) in dietary magnesium intakes (data not shown). 45 64 9 1447 W: 5.1 6 3.6 W: 2.7 6 1.7 0.75 0.60, 0.92,0.05 AA: 4.0 6 2.6 AA: 2.8 6 1.6 0.86 0.65, 1.15 NS All = W + AA 0.79,0.05 BWHS 59,000 F, AA 21 69 8 1938 8.3 6 2.5 1.5 6 0.7 0.82 0.70, 0.96 0.01 MCCS 36,787 M + F 40 69 4 365 Every 10 g 0.97 0.79, 1.20 NS 1 RRs are for the comparisons between highest quintile or quartile compared with that of the lowest. AA, African Americans; ARIC, Atherosclerosis Risk in Communities; BWHS, Black Women s Health Study; EPIC-Potsdam, European Prospective Investigation into Cancer and Nutrition Potsdam; FMCHES, Finnish Mobile Clinic Health Examination Survey; HPFS, Health Professionals Follow-Up Study; IWHS, Iowa Women s Health Study; MCCS, Melbourne Collaborative Cohort Study; NA, not available; NHS, Nurses Health Study; P-trend = P value for trend test across all levels of exposure; T2D, type 2 diabetes; W, whites. 2 Multivariable model plus adjustment for cereal grains and magnesium. Similarly, Fung et al (40) showed that the benefit seen with mixtures of whole grains and bran disappeared when the data were adjusted for cereal fiber, magnesium, and glycemic load. In the Finnish Mobile Clinic Health Examination Survey, Montonen et al (16) found that consumption of whole grains and cereal fiber was associated with a reduced risk of T2D. The RRs between the highest and lowest quartiles were 0.65 (95% CI: 0.36, 1.18; P-trend = 0.02) for whole grains and 0.39 (95% CI: 0.20, 0.77; P-trend = 0.01) for cereal fiber. However, this inverse relation between high intake of whole grains and T2D was similar but not significant (RR in the highest quartile: 0.67; 95% CI: 0.32, 1.38; NS) after adjustment for intake of saturated fat, antioxidant vitamins (vitamins E and C, b-carotene), vitamin B-6, folic acid, flavonoids, and magnesium. After further adjustment for cereal fiber, the RR in the highest quartile of whole grain intake was 1.14 (NS). The association of cereal fiber remained significant after additional adjustment for saturated fat, antioxidant vitamins, vitamin B-6, folic acid, flavonoids, and magnesium. The authors suggested an independent association between cereal fiber intake and T2D. It is noteworthy that the whole-grain intake amounts in Finnish cohorts is significantly higher than those in the US cohorts; the highest and the lowest intake group consumed 236 and 79 g of mixtures of whole grains and bran, respectively. Despite the

600 CHO ET AL TABLE 3 Prospective cohort studies of mixtures of whole grains and bran: risk of T2D 1 (reference) Study name No of participants Age at baseline Follow-up cases Highest Intake of mixtures of whole grains and bran Lowest RR 95% CI P-trend Fung et al, 2002 (40) Kochar et al, 2007 3 (41) Liu et al, 2000 (42) Meyer et al, 2000 (26) van Dam et al, 2006 (43) Montonen et al, 2003 (16) y y HPFS 42,898 M 40 75 #12 1197 3.2 servings/d 0.4 servings/d 0.70 0.57, 0.85 0.0006 0.98 2 0.76, 1.26 NS PHS 21,152 M 39.7 85.9 19.1 1958 $7 servings/wk 0 servings/wk 0.60 0.50, 0.71,0.0001 NHS 75,521 F 38 63 10 1879 2.70 servings/d 0.13 servings/d 0.73 0.63, 0.85,0.0001 IWHS 35,988 F 55 69 6 1141.17.5 servings/wk,3.0 servings/wk 0.79 0.65, 0.96 0.0089 0.93 4 0.75, 1.16 NS BWHS 41,186 F 21 69 8 1964 1.29 servings/d 0.03 servings/d 0.69 0.60, 0.79,0.0001 FMCHES 5 4316 M + F 40 69 10 156 302 g/d 79 g/d 0.64 0.36, 1.15 0.02 0.67 6 0.32, 1.38 NS 1.14 7 0.66, 2.49 NS 1 BWHS, Black Women s Health Study; FMCHES, Finnish Mobile Clinic Health Examination Survey; HPFS, Health Professionals Follow-Up Study; IWHS, Iowa Women s Health Study; NHS, Nurses Health Study; PHS, Physicians Health Study; T2D, type 2 diabetes. 2 Multivariable model after further adjustment for cereal fiber, magnesium, and glycemic load. 3 Limited to breakfast cereals with $25% of whole grains and bran. 4 Multivariable model plus adjustment for cereal fiber intake. 5 Included germs in the whole-grain definition. 6 Multivariable model after adjustments for intakes of other dietary factors including saturated fat, antioxidant vitamins (vitamins E and C, b-carotene), vitamin B-6, folic acid, flavonoids, and magnesium. 7 Multivariable model after further adjustment for cereal fiber differences in major sources of whole grains (rye in Finland and wheat in the United States), trends for inverse associations were similar between the 2 countries. Data from cross-sectional studies are consistent with the findings from prospective studies (Appendixes B and C; 35, 36, 44 50). Evidence level B was fulfilled for the association between intake of mixtures of whole grains and bran and risk reduction of T2D. Studies on whole grains Two studies (17, 51) met the whole-grain definition criteria defined by the FDA. Evidence for the association between wholegrain intake and T2D risk reduction was shown in one large, welldesigned prospective cohort of females only [Nurses Health Study (NHS)], (Table 4; 17). No prospective study in males was found. Although the NHS was a carefully designed and wellconducted study, the absence of any confirmatory prospective study limits the confidence that whole grains reduce the incidence of T2D. Nonetheless, the data from the NHS allow one to make a comparison between the benefit of whole grain compared with cereal fiber in reducing the risk of diabetes. Daily intakes of bran, cereal fiber, and whole grains were 9.6 12 g, 12 g, and 36.9 45.6 g in the highest quintiles and 0.6 1.1 g, 1 g, and 3.2 5.5 g in the lowest quintiles, respectively. These data suggest that daily intakes of 8 11 g of cereal fiber and 34 40 g of whole grains provide comparable RR values. One crosssectional study showed no association with risk factors related to T2D (Appendix D; 51). Evidence for the association between whole-grain intake and development of T2D was considered as level C. Obesity and body weight measures Studies on cereal fiber The 2 prospective studies (52, 53) relating cereal fiber to various body weight measures are listed in Table 5. These 2 TABLE 4 Prospective cohort studies on whole grains: risk of T2D 1 (reference) Study name participants Age Follow-up cases Whole-grain intake 2 Highest Lowest RR 95% CI P-trend de Munter et al, 2007 (17) y y g/d NHS I 73,327 F 37 65 12 18 4747 36.9 3.2 0.63 0.57, 0.69,0.001 NHS II 88,410 F 26 46 12 18 1739 45.6 5.5 0.68 0.57, 0.81,0.001 1 RRs are for the comparison between the extreme quintiles. NHS, Nurses Health Study; T2D, type 2 diabetes. 2 The definition of whole grain met the US Food and Drug Administration criteria.

TABLE 5 Prospective cohort studies on cereal fiber; body weight measures 1 FROM THE AMERICAN SOCIETY FOR NUTRITION WHOLE-GRAIN, BRAN, AND FIBER INTAKE AND RISK OF DISEASE 601 (reference) Study name participants, age Study design, follow-up Cereal fiber intake Highest Lowest Endpoints Body weight measures (highest vs lowest) P-trend Koh-Banerjee et al, 2004 (52) HPFS 27,082 M, aged 40 75 y Du et al, 2010 (53) Diogenes 89,432 M + F, Project aged 20 78 y prospective cohort studies in males and females showed significant inverse associations between cereal fiber intake and body weight measures. However, the absolute reduction in weight gain from cereal fiber consumption was 0.39 kg between the highest intake groups. The study by Du et al (53) also reported a small change in body weight, 77 g/y per daily intake of 10 g cereal fiber. Although these differences were significant, it is difficult to evaluate their biological significance due to minimal differences in absolute body weights. A total of 4 of 4 cross-sectional reports (36, 39, 54, 55) summarized in Appendix E reported inverse associations between cereal fiber intake and body weight measures. The differences in BMI, body weights, or WC between the highest and the lowest intakes were large enough to have biological significance. We assigned evidence level B/C for an inverse association between cereal fiber intake and various body weight measures. Studies on mixtures of whole grains and bran Most studies on whole grains included $25% bran in the definition of whole-grain foods (36, 44, 45, 47, 48, 52, 54, 56 62). One study reported both definitions of whole grains (52), and another study (57) confined the evaluation to breakfast cereals with $25% whole grains and bran. The prospective reports comparing whole grains and bran with measures of body weight are listed in Table 6 (52, 56, 57). All of the 3 large prospective cohort studies in both men and women (52, 56, 57) and a small prospective cohort study in children (58) reported consistent inverse correlations between intakes of mixtures of whole grain and bran and BMI, weight gain, body weight, or risk of obesity. Despite consistent inverse associations, these prospective studies showed minimal differences in absolute body weight or body weight gain between the highest and the lowest intake groups: The studies by Koh-Banerjee et al (52), Liu et al (56), and Bazzanzo et al (57) reported body weight gain differences of 0.52, 0.39, and 0.35 0.42 kg, respectively, during 8- to 13-y follow-up periods. It is difficult to assess the health impact of such minimal differences. However, Liu et al (56) reported that, over a 12-y period, the reductions in risk of obesity and weight gain of $25 kg were 19% and 23%, respectively. Also, Bazzano et al (57) reported g/d P, 8 y Change in Change in Body weight +0.91 vs +1.30 0.0004 intake: 5.1 intake: 22.2 change (kg/8 y) (0.39-kg difference) P, 6.5 y 12.6 6 4.6 9.3 6 3.5 Body weight change 277 (2127, 26) 0.01 (g/y) per 10 g cereal fiber/d WC change (cm/y) 20.10 (20.18, 20.02),0.001 per 10 g cereal fiber/d 1 Values in parentheses are 95% CIs. HPFS, Health Professionals Follow-Up Study; P, prospective; WC, waist circumference. that, compared with the lowest consumers, the highest consumers of breakfast cereals with $25% whole grains and bran had a 23% lower risk of weight gain of $10 kg (RR: 0.78; 95% CI: 0.64, 0.96; P = 0.01). Steffen et al (58) reported a reduction in BMI of 7.2% during the 2-y follow up period in the highest consumers of mixture of whole grains and bran. All of the cross-sectional studies with.1500 subjects each (36, 47, 48, 59, 60) were consistent with the inverse association between intake of whole grains and bran and body weight measures (Appendixes F and G). The differences in BMI, body weight, or WC between the highest and the lowest intake were large enough and may have biological significance. The studies reporting no associations (61, 62) or mixed results (44) had low numbers of subjects and may not have had sufficient statistical power (Appendixes F and G). Evidence level B/C was the grade for the association between intake of mixtures of whole grains and bran and measures of body weight. Studies on whole grains One study defined whole grains as foods containing $10% whole grains or bran (63). There was one prospective study in a male cohort with a lengthy follow-up (Health Professionals Follow-Up Study) (52; Table 7), and the remainder were crosssectional studies (9, 55, 64; Appendix H). In the Health Professionals Follow-Up Study, the absolute reduction in weight gain (0.29 kg) in the highest intake group was minimal: weight gains over the 8-y follow-up period were 0.69 and 0.96 kg for the highest and the lowest quintile category, respectively (Ptrend = 0.002). Despite statistical significance, the differences are likely not biologically meaningful. The relation between whole grains and reduction in weight gain was weakened after adjustment for added bran and cereal fiber intakes but still persisted. A cross-sectional study by O Neil et al (Appendix H; 9) showed that inverse associations between whole-grain intake and body weight measures (BMI and WC) disappeared after adjustment for cereal fiber. Because only one prospective study in men reported a minimal difference in body weight gain between the highest and the lowest intake groups, evidence for the inverse association of whole grains is considered level C/D.

602 CHO ET AL TABLE 6 Prospective cohort studies on mixtures of whole grains and bran: risk of obesity and body weight measures 1 (reference) Study name participants, age Follow-up Intake of mixtures of whole grains and bran Highest Lowest Endpoints RR (95% CI) or body weight measure (highest vs lowest) P-trend Liu et al, 2003 (56) Koh-Banerjee et al, 2004 (52) Bazzano et al, 2005 (57) NHS 74,091 F, aged 38 63 y HPFS 26,082 M, aged 40 75 y Steffen et al, 2003 (58) MPSS 285 M + F, mean age of 13 y y 12 Servings/1000 kcal per day: at baseline, 1.62; change in intake in 12 y, 0.90 Servings/1000 kcal per day: at baseline, 0.07; change in intake in 12 y, 20.59 OR for weight gain, $25 kg in 12 y OR for BMI (in kg/m 2 ) $30 in 12 y 0.77 (0.59, 1.01) 0.03 0.81 (0.73, 0.91) 0.0002 Weight gain (kg) 4.12 vs 4.51,0.0001 2 4 Average changes in BMI 0.46 vs 0.56,0.0001 (kg/m 2 ) 1.23 vs 1.52,0.0001 Average changes in weight (kg) 8 +27.0 g/d 211.0 g/d Body weight change (kg/8 y) +0.73 vs +1.25 (difference: 0.52) PHS 2 17,881 M 8 $1 serving/d Rarely Weight gain (kg) 1.13 vs 1.55 0.003 RR for BMI $25 0.83 (0.71, 0.98) 0.06 13 Weight gain (kg) 1.83 vs 2.18 0.08 RR for BMI $25 0.91 (0.79, 1.05) 0.13 RR for weight gain $10 kg,0.0001 0.78 (0.64, 0.96) 0.01 2.1.5 servings/d,0.5 servings/d BMI (kg/m 2 ) 21.9 vs 23.6 0.05 1 HPFS, Health Professionals Follow-Up Study; MPSS, Minneapolis Public School Students; NHS, Nurses Health Study; PHS, Physicians Health Study. 2 A report from the PHS was limited to breakfast cereals with $25% whole grains and bran.

TABLE 7 Prospective cohort studies of whole grain: body weight measures FROM THE AMERICAN SOCIETY FOR NUTRITION WHOLE-GRAIN, BRAN, AND FIBER INTAKE AND RISK OF DISEASE 603 (reference) Study name participants, age Follow-up Highest Whole-grain intake Lowest Endpoints Highest vs lowest P-trend Koh-Banerjee et al, 2004 (52) HPFS 1 27,082 M, aged 40 75 y 8 y Change in intake: +15.6 g/d Change in intake: 217.8 g/d Body weight change (kg/8 y) +0.69 vs +0.96 (difference: 0.27) 0.002 1 HPFS, Health Professionals Follow-Up Study. CVD Prospective studies consistently showed a reduced risk of CVD with high intakes of cereal fiber (a risk reduction of 14 26% for CVD mortality and 22 43% for stroke) or mixtures of whole grains and bran (a risk reduction of 7 52% for CVD mortality, CVD events, and heart failure). Only 1 of 2 prospective studies showed an inverse association between whole-grain intake and reduction in risk of CVD. Subtypes of stroke were not evaluated in this review. Also, one study (63), which used the cutoff of 10% whole-grain content to define whole-grain foods, is not included in the review. Studies on cereal fiber A summary of the reports showing the relation between cereal fiber consumption and CVD or hypertension are provided in Table 8. Six of the 10 publications (65 74) on CVD were independent studies, of which one study addressed hypertension. For the reports related to the incidence of CVD, a variety of outcome measures were reported. In general, a consistent but modest risk reduction was seen in CVD mortality (65, 67) and stroke (69 71): risk reductions were in the range of 14 26% for CVD mortality and 22 43% for stroke. However, a nonsignificant risk reduction (RR: 0.70; 95% CI: 0.46, 1.06; NS) in CVD mortality was observed in subjects with T2D (66). The risk reductions for CVD events (68 70) and myocardial infarction (69, 73) were inconsistent. No studies showed adverse effects of cereal fiber. Two cross-sectional studies reported mixed results (Appendix I; 39, 75). The evidence level is B for the association between cereal fiber intake and reduction in risk of CVD. In the study by Alonso et al (74), fiber from cereals was inversely associated with a lower risk of hypertension (HR comparing the fifth and first quintile: 0.60; 95% CI: 0.3, 1.0; P-trend = 0.05). The relation with hypertension was stronger among individuals over the age of 40 y than in younger people and stronger among males than in females. A cross-sectional study supported the findings from a prospective study (75; Appendix I). Evidence is considered inadequate (level D) due to the limited number of large, well-designed prospective studies. Studies on mixtures of whole grains and bran A summary of 10 reports from 6 independent prospective studies showing a relation between consumption of mixtures of whole grains and bran and reduction in risk of CVD or hypertension is provided in Table 9 (13 15, 45, 76 81). Most of the reports (w80%) related to heart disease considered whole-grain foods as products containing whole grains and $25% bran. Reports from the Physicians Health Study (76, 79) limited the investigation to breakfast cereals containing whole grains and $25% bran. All 9 reports from 4 large and 1 small prospective cohort studies showed significant inverse associations between intakes of mixtures of whole grains and bran and risks of CVD mortality, CVD events, and heart failure in both males and females (Table 9; 13 15, 45, 76 79). However, the risk reduction for ischemic stroke was not significant (77, 80). Cross-sectional studies on risk factors for CVD in relation to consumption of whole grains and bran reported mixed results (Appendix J; 36, 44, 47, 48, 50, 61, 82). Overall, moderate evidence (grade B) exists for the association between intake of mixtures of whole grains and bran and reduction in risk of CVD because all large, prospective cohort studies showed relatively consistent associations and the number of studies was adequate. Only one study (81; Table 9) reported an inverse association between reduction in risk of hypertension and intakes of mixtures of whole grains and bran in women only, and cross-sectional studies generally found no association (36, 44, 45, 47, 48; Appendix J). Thus, there is inadequate evidence to suggest that consumption of whole grains with added bran will affect the incidence of hypertension (evidence level D). Studies on whole grains A summary of the reports showing a relation between wholegrain consumption and risk of CVD or hypertension is provided in Table 10 (66, 83, 84). The inverse association between whole-grain intake and CVD risk was attenuated and became nonsignificant after adjustment for dietary fiber, magnesium, and other dietary factors (83), indicating that dietary fiber and magnesium account for whole-grain actions. The study by He et al (66), which reported no association, was confined to diabetic women. One cross-sectional study (Appendix K; 51) reported inconsistent associations of whole-grain or bran intake with risk factors for CVD. Thus, there is limited evidence for the association between intake of whole grains and reduction in risk of CVD (level C). Flint et al (84) reported an inverse association between wholegrain intake and hypertension, with an RR of 0.81 (95% CI: 0.75, 0.87; P, 0.0001) in the highest quintile compared with the lowest (Table 10). The inverse association was attenuated or disappeared after adjustment for bran (RR: 0.88; 95% CI: 0.77, 1.00; P = 0.04) or cereal fiber (RR: 0.93; 95% CI: 0.84, 1.05; NS). A cross-sectional study (51) reported no association with diastolic blood pressure. Evidence for the association of wholegrain consumption with hypertension is considered inadequate to draw any conclusions (level D).

604 CHO ET AL TABLE 8 Prospective cohort studies on cereal fiber: risk of CVD and hypertension 1 (reference) Study name No of participants Age Follow-up Endpoints cases Cereal fiber intake Highest Lowest RR 95% CI P-trend Eshak et al, 2010 (65) He et al, 2010 (66) Pietinen et al, 1996 (67) Pietinen et al, 1996 (67) Wolk et al, 1999 (68) Liu et al, 2002 (69) Mozaffarian et al, 2003 (70) Mozaffarian et al, 2003 (70) Oh et al, 2005 (71) Ascherio et al, 1996 (72) Liu et al, 2002 (69) Rimm et al, 1996 (73) Alonso et al, 2006 (74) JCCSECR 58,730 M + F, general population y y g/d 40 79 14.3 CVD mortality in men 2080.2.1,1.4 0.86 2 0.64, 0.99 0.042 0.89 3 0.65, 1.01 0.060 CVD mortality in women.1.7,1.1 0.77 2 0.59, 0.98 0.031 0.76 3 0.59, 0.97 0.044 NHS 7822 F, diabetic 30 55 26 CVD mortality 295 32.6 g 4.8 g 0.70 0.46, 1.06 NS ATBC study 21,930 M, general population 50 69 6.1 CHD death 581 26.3 8.8 0.74 0.57, 0.96 0.01 ATBC study 21,930 M 50 69 6.1 Coronary event 1399 26.3 8.8 0.91 0.77, 1.09 NS NHS 68,782 F 37 64 10 CHD events 591 7.7 2.2 0.63 0.49, 0.81 0.002 NHS 39,876 F $45 5.8 CVD events 570 6.5 3.0 1.11 0.84, 1.46 NS CHS 3588 M + F $65 8.6 CVD events 811.6.3,1.7 0.79 0.62, 0.99 0.02 CHS 3588 M + F $65 8.6 Total stroke 392.6.3,1.7 0.78 0.64, 0.95,0.05 NHS 78,799 F 30 55 18 Total stroke 1020 5.7 1.4 0.66 0.52, 0.83 0.001 HPFS 43,738 M 40 75 8 Total stroke 328 CF: NA (total fiber: 28.9) CF: NA (total fiber: 12.4) 0.57 0.36, 0.92 0.03 NHS 39,876 F $45 5.8 MI 177 6.5 3.0 0.91 0.56, 1.47 NS HPFS 43,757 M 40 75 6 Fatal and nonfatal MI 734 9.7 2.2 0.71 0.54, 0.92 0.01 SUN follow- up study 5880 M + F.20.2 Hypertension 180 NA NA 0.60 0.3, 1.0 0.05 1 ATBC, Alpha-Tocopherol, Beta-Carotene Cancer Prevention; CF, cereal fiber; CHD, coronary heart disease; CHS, Cardiovascular Health Study; CVD, cardiovascular disease; HPFS, Health Professionals Follow-Up Study; JCCSECR, Japanese Collaborative Cohort Study for Evaluation of Cancer Risks; MI, myocardial infarction; NA, not available; NHS, Nurses Health Study; SUN, The Seguimiento Universidad de Navarra. 2 Multivariable model. 3 Multivariable model plus further adjustment for fiber.

TABLE 9 Prospective cohort studies on mixtures of whole grain and bran: risk of CVD and hypertension 1 FROM THE AMERICAN SOCIETY FOR NUTRITION WHOLE-GRAIN, BRAN, AND FIBER INTAKE AND RISK OF DISEASE 605 (reference) Study name participants Age Follow-up Endpoints cases Highest Whole-grain intake Lowest RR 95% CI P-trend y y servings Jacobs et al, IWHS 31,284 F 55 69 9 Mortality from IHD 438 3.2 0.2/d 0.70 0.50, 0.98 0.02 1998 (13) Jacobs et al, IWHS 31,284 F 55 69 9 CVD mortality 1097 3.2 0.2/d 0.82 0.63, 1.06 0.03 1999 (14) Liu et al, PHS 2 86,190 M 40 84 5.5 CVD Mortality 3114 $1.0 Rarely 0.80 0.66, 0.97 0.008 2003 (76) Sayhoun et al, NA 535 M + F 60 98 12 15 CVD Mortality 89.1.94 #0.56/d 0.48 0.25, 0.96 0.04 2006 (45) Liu et al, NHS 75,521 F 38 63 10 CHD, fatal and 761 2.7 0.13/d 0.75 3 0.59, 0.95 0.01 1999 (15) nonfatal MI 0.79 4 0.62, 1.01 0.07 Steffen et al, 2003 (77) ARIC study 11,940 M + F 45 64 11 Incident CAD 535 3.0 0.1/d 0.72 0.53, 0.97 0.05 Steffen et al, 2003 (77) ARIC study 11,940 M + F 45 64 11 Heart failure 1140.1/d 0.93 0.87, 0.99,0.05 Nettleton et al, 2008 (78) ARIC study 14,153 M + F 45 64 13 Heart failure, hospitalized 1140 Per difference of 1 serving/d 0.93 0.87, 0.99,0.05 Djoussé et al, 2007 (79) PHS 2 21,376 M 40 86 19.6 Heart failure 1018 $1 0/wk 0.72 0.59, 0.88,0.001 Liu et al, NHS 75,521 F 38 63 10 Ischemic stroke 352 2.7 0.13/d 0.64 5 0.47, 0.89 0.04 2000 (80) 0.69 6 0.50, 0.98 0.08 Steffen et al, 2003 (77) ARIC study 11,940 M + F 45 64 11 Ischemic stroke 214 3.0 0.1/d 0.75 0.46, 1.22 NS Wang et al, 2007 (81) WHS 28,926 F $45 10 Hypertension 8722 5.0 0.28/d 0.89 0.82, 0.97 0.007 1 Age, age at baseline; ARIC, Atherosclerosis Risk in Communities; CAD, fatal and nonfatal myocardial infarction, coronary artery disease death, and stroke; CHD, coronary heart disease; CVD, cardiovascular disease; IHD, ischemic heart disease; IWHS, Iowa Women s Health Study; MI, myocardial infarction; NA, not applicable; NHS, Nurses Health Study; PHS, Physicians Health Study; WHS, Women s Health Study. 2 Reports from PHS were limited to breakfast cereals with $25% whole grains and bran. 3 Multivariable model adjusted for age, BMI, cigarette smoking, alcohol intake, parental or family history of myocardial infarction before age 60 y, selfreported hypertension or hypercholesterolemia, menopausal status, hormone replacement usage, protein intake, aspirin use, multiple vitamin or vitamin E use, vigorous activity, total energy intake, and dietary fatty acid classes. 4 Multivariable model after further adjustment for dietary fiber, folate, and vitamins E and B-6. 5 After adjustments for age and smoking. 6 Multivariable model, after further adjustments for other known CVD risk factors. DISCUSSION This review focused on the observational studies linking consumption of whole grains, mixtures of whole grains and bran, or cereal fiber intake with risk reduction for T2D, obesity, or CVD. The strength of evidence from observational studies is not as strong as that from intervention trials (23, 85); however, we could not identify any long-term (.1 y) RCTs that used cereal fiber, mixtures of whole grains and bran, or whole grains. Although there are many controlled trials that used a specific fiber, whole grain, or bran on putative intermediate biomarkers, none measured disease endpoints. All such studies were of short-term duration, recruited small numbers of subjects, or participants were given controlled portions of the foods that do not mimic ordinary daily consumption. Unfortunately, it is difficult to conduct long-term, adequately powered randomized trials of consumption on cereal fiber, bran, and whole grains because of the difficulty in controlling food intake over long-enough periods to show a difference in outcomes between groups, even though they are needed to make sound recommendations. Clinical trials have compared whole with refined grains, but none studied whole grains compared with bran or cereal fiber. In addition, a study on a single grain or a mixture of 1 2 grains does not necessarily assess the impact of the whole-grain food category as consumed by the population. Thus, observational studies have to be used for decision making concerning potential associations of grain components with health. A well-designed observational study may be more persuasive than poorly controlled and performed or otherwise very limited randomized trials (86). However, observational studies have several limitations as follows: 1) not all confounders can be controlled, 2) protocols (including food-frequency questionnaires) of each observational study are different, and 3) food composition tables may not accurately reflect individual foods consumed by participants of the study. We considered that the data from large prospective studies are superior to and more reliable than those from cross-sectional studies. Thus, we have rated the strength of scientific evidence on the basis of the quality, quantity, and consistency of results from large prospective studies and have simply presented cross-sectional summaries as supporting data in Appendixes A K. We found that most RCTs and meta-analyses of RCTs did not capture the impact of major whole grains, such as wheat and corn,

606 CHO ET AL TABLE 10 Prospective cohort studies on whole grains: risk of CVD and hypertension 1 (reference) Study name participants Age Follow-up Endpoints cases Whole-grain intake Highest Lowest RR 95% CI P-trend Jensen et al, 2004 (83) He et al, 2010 (66) Flint et al, 2009 (84) y y g/d HPFS 42,850 M 40 75 14 CHD 1818 42.4 3.5 0.82 2 0.70, 0.96 0.01 0.84 3 0.71, 0.98 0.02 0.85 4 0.71, 1.02 0.06 NHS 7822 F with T2D 30 55 26 CVD specific mortality 295 32.6 4.8 0.70 0.46, 1.06 NS HPFS 31,648 M 40 75 18 Hypertension 9227 46.0 3.3 0.81 0.75, 0.87,0.0001 0.88 5 0.77, 1.00 0.04 0.94 6 0.84, 1.05 NS 1 RRs are for the comparison between the extreme quintiles. CHD, coronary heart disease; CVD, cardiovascular disease; HPFS, Health Professionals Follow-Up Study; NHS, Nurses Health Study; T2D, type 2 diabetes. 2 Multivariable model including adjustment for added bran, added germ, age, energy intake, smoking, alcohol, physical activity, family history of myocardial infarction, vitamin E supplement use, and intakes of fats, fruit, vegetables, and fish. 3 Multivariable model plus BMI. 4 Multivariable model plus BMI plus dietary fiber, magnesium, and other nutrients. 5 Multivariable model including adjustment for bran. 6 Multivariable model including adjustment for cereal fiber. consumed in the United States. Health Canada rejected health claims for whole grains and heart disease risk reduction, because the studies on minor whole grains such as oat and barley showed risk reductions for heart disease but there were not enough data on major grains such as wheat and corn (87). Also, no long-term RCTs are available. Thus, observational studies may more accurately reflect the impact of whole grains as consumed by the population. There is reasonable evidence for an inverse association of intake of cereal fiber or mixtures of whole grains and bran and risk reductions for T2D and CVD (Table 11). However, the relation is less convincing for whole grains by themselves when using the currently accepted definition. The evidence (evidence level B/C) for the association between intakes of cereal fiber or mixtures of whole grains and bran and body weight measures is not as strong as those for T2D and CVD. This is partly due to the fact that the absolute amounts of body weight changes were relatively small despite significant differences between the highest and the lowest intake groups, and it is difficult to assess clinical benefits related to such small differences. Despite minimal differences in absolute weight gain, the studies reporting risks of obesity (OR: 0.81; 95% CI: 0.73, 0.91; P-trend = 0.0002; 56), weight gain of $25 kg (OR: 0.77; 95% CI: 0.59, 1.01; P = 0.03; 56), or weight gain of $10 kg (RR: 0.78; 95% CI: 0.64, 0.96; P = 0.01; 57), or BMI (21.9 compared with 23.6; P, 0.05; 58) presented stronger evidence for mixtures of whole grains and bran. With regard to risk reduction for CVD, evidence for the association with cereal fiber intake is considered moderate (evidence level B) due to consistent inverse associations noted for CVD mortality and stroke. Evidence for whole grains and bran is considered moderate (evidence level B) due to consistent inverse associations found across different populations. However, evidence for whole grains, per se, is considered limited (66, 84). Overall, the evidence for whole grains alone is limited or very limited for reduction in risk of T2D, CVD, or obesity/body weight measures. The inverse associations for the consumption of whole grains or mixtures of whole grains and bran and the risk of T2D, body weight measures, or CVD were attenuated or disappeared after the models were adjusted for cereal fiber, magnesium, bran, and/or other dietary components in whole grains (9, 26, 40, 52, 83, 84). The data indicated that the inverse relation between whole-grain intake and chronic conditions may be partly due to cereal fiber and bran in whole grains, and that cereal fiber and/or bran may account for much of the risk reduction associated with whole grains. Our review supports the summary from the Life Sciences Research Office (88), which stated that the associations between whole-grain intake and risk reduction for T2D and CVD are inconclusive when the definition of whole grain does not include added bran. Similar conclusions were reached by the European Food Safety Authority (89), which rejected health claims related to whole grains (blood cholesterol concentration, carbohydrate metabolism and insulin sensitivity, low glycemic index, weight control, and weight management, among others). These conclusions may be due, in part, to the following factors: 1)veryfew studies had a long follow-up period; 2) different types of whole grains may have different physiologic roles, thus it is difficult to reach a conclusion from studies investigating the effects of one type of whole grain; and 3) the limited number of studies investigated whole-grain effects alone (ie, most whole-grain studies are confounded by the inclusion of bran cereals). It is TABLE 11 Summary of evidence level 1 T2D Obesity CVD Hypertension Cereal fiber B B/C B D Mixtures of whole B B/C B D grains and bran Whole grains C C/D C D 1 CVD, cardiovascular disease; T2D, type 2 diabetes.