点击显示 收起
1 From the Division of Preventive Medicine (SL, HDS, JAEM, and JEB) and Channing Laboratory (JAEM and WCW), Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston; the Department of Ambulatory Care and Prevention, Harvard Medical School, Boston (JEB); and the Departments of Epidemiology (SL, HDS, JAEM, WCW, and JEB) and Nutrition (WCW), Harvard School of Public Health, Boston.
2 Supported by grants HL-42441 and DK02767 from the National Institutes of Health. 3 Reprints not available. Address correspondence to S Liu, Division of Preventive Medicine, Brigham and Women’s Hospital, 900 Commonwealth Avenue East, Boston, MA 02215. E-mail: simin.liu{at}channing.harvard.edu.
ABSTRACT
Background: Prospective studies suggested that substituting whole-grain products for refined-grain products lowers the risks of type 2 diabetes and cardiovascular disease (CVD) in women. Although breakfast cereals are a major source of whole and refined grains, little is known about their direct association with the risk of premature mortality.
Objective: We prospectively evaluated the association between whole- and refined-grain breakfast cereal intakes and total and CVD-specific mortality in a cohort of US men.
Design: We examined 86 190 US male physicians aged 40–84 y in 1982 who were free of known CVD and cancer at baseline.
Results: During 5.5 y, we documented 3114 deaths from all causes, including 1381 due to CVD (488 myocardial infarctions and 146 strokes). Whole-grain breakfast cereal intake was inversely associated with total and CVD-specific mortality, independent of age; body mass index; smoking; alcohol intake; physical activity; history of diabetes, hypertension, or high cholesterol; and use of multivitamins. Compared with men who rarely or never consumed whole-grain cereal, men in the highest category of whole-grain cereal intake ( 1 serving/d) had multivariate-estimated relative risks of total and CVD-specific mortality of 0.83 (95% CI: 0.73, 0.94; P for trend < 0.001) and 0.80 (0.66, 0.97; P for trend < 0.001), respectively. In contrast, total and refined-grain breakfast cereal intakes were not significantly associated with total and CVD-specific mortality. These findings persisted in analyses stratified by history of type 2 diabetes, hypertension, and high cholesterol.
Conclusions: Both total mortality and CVD-specific mortality were inversely associated with whole-grain but not refined-grain breakfast cereal intake. These prospective data highlight the importance of distinguishing whole-grain from refined-grain cereals in the prevention of chronic diseases.
Key Words: Whole-grain cereals • refined-grain cereals • prospective study • cardiovascular diseases • mortality • men • Physicians’ Health Study
INTRODUCTION
Dietary guidelines have long recommended consumption of grain products to maintain health and prevent chronic diseases (1, 2). However, most grain products consumed in the US are highly refined (3), which often leads to the loss of many potentially beneficial micronutrients, antioxidants, minerals, phytochemicals, and fiber (4–6). Large prospective studies found that increased intakes of whole-grain products are associated with reduced risks of type 2 diabetes (7, 8), hypertension (6, 9), and cardiovascular disease (CVD) (10–12). Two other prospective studies indicated increased risks of type 2 diabetes and total mortality associated with the intake of refined-grain products (8, 13, 14). Taken together, the results of these studies raise important questions regarding the net effect of increasing total grain intake on chronic disease risk. Although breakfast cereals are a major source of both whole and refined grains, their total effect on total and CVD-specific mortality remains to be clarified (6), especially in men. To provide further data on the relative importance of types or amounts of grain, we examined associations between the intakes of whole-grain and refined-grain breakfast cereals and the risks of total and CVD-specific mortality in a large prospective cohort of US male physicians.
SUBJECTS AND METHODS
Study design
The Physicians’ Health Study was a randomized, double-blind, placebo-controlled trial testing the efficacy of aspirin and ß-carotene in the primary prevention of CVD and cancer. The study’s methodology was described previously (12, 13). In brief, potentially eligible participants were male physicians who resided in the United States in 1982. Letters of invitation, informed consent forms, and baseline questionnaires were mailed to the 261 248 men listed on an American Medical Association mailing tape. By 31 December 1983, 104 353 physicians had responded to the initial enrollment questionnaire. Men with a history of cancer and CVD were excluded, which left a total of 92 785 participants. After further excluding those who did not provide information on breakfast cereal intake (7.1%), the final population for analyses consisted of 86 190 men.
Data collection
On the baseline questionnaire, physicians self-reported CVD risk factors, including age; cigarette smoking status [never, past, current (including number of cigarettes smoked daily)]; alcohol intake (rarely or never, monthly, weekly, daily); use of multivitamin supplements; history of hypertension, high cholesterol, and diabetes; and frequency of vigorous exercise (rarely, 1 time/wk, 2–4 times/wk, 5 times/wk). Body mass index (in kg/m2) was calculated by using reported height and weight.
Study participants also completed an abbreviated, simple semiquantitative food-frequency questionnaire (SFFQ). On these dietary questionnaires, men reported their average intake of breakfast cereals of a specific portion size during the past year, including the amounts, brands, and types of cereals consumed and the frequency of their consumption. These items were selected from a validated SFFQ used in the Nurses’ Health Study in 1980 to discriminate and rank dietary intake among participants (15). A full description of the SFFQs and data on reproducibility and validity in the Nurses’ Health Study was previously reported (15). The ability of the Nurses’ Health Study SFFQ to assess the intake of individual grain products was documented to be good (16). For example, in a sample of the participants, correlation coefficients between the SFFQ and detailed diet records were 0.75 for cold breakfast cereal. Overall, these data indicate that the SFFQ provides reasonably valid measures of long-term average dietary intakes. For each food item on the questionnaire, 7 responses regarding frequency of intake were possible, ranging from never to 2 servings/d. We used a procedure developed by Jacobs and colleagues to classify breakfast cereals into whole and refined grains (7). Specifically, the breakfast cereals listed in the SFFQ were evaluated for whole-grain and bran content; breakfast cereals that contained 25% whole grain or bran by weight were classified as whole grain, which is the classification used by Jacobs et al and others (10, 11). To maintain a high specificity in the definition of whole-grain cereals, we included responses in which brand names were missing in the category of refined grains, because refined-grain cereals were more readily available in the market in the 1980s than were whole-grain cereals (3). Sensitivity analyses in which this assumption was varied did not materially change our findings (data not shown).
Ascertainment of death endpoints
Deaths were identified through systematic searches of the National Death Index for the entire enrollment cohort, and death certificates were obtained from state agencies for all deaths that occurred before 1 February 1988. The deaths were classified by trained nosologists according to the International Classification of Diseases, ninth revision. The "Automated Classification of Medical Entities Decision Tables" was used to select the underlying cause of deaths that occurred during a mean follow-up period of 5.5 y. We chose as endpoints all deaths and deaths caused by CVD. The reliability of the National Death Index for epidemiologic purposes has been validated (17).
Statistical analysis
We considered intakes of whole-grain, refined-grain, and total breakfast cereals as both continuous (servings/d) and categorical variables. In initial descriptive analyses, we first examined the distributions of total, whole-grain, and refined-grain breakfast-cereal intakes. Because the distributions of these variables were not symmetrical (skewed to the high end) and were somewhat truncated, we did not use quintiles to categorize intake. Rather, we categorized cereal intake as rarely or never, 1 serving/wk, 2–6 servings/wk, and 1 serving/d to maintain a gradient of exposure and to include adequate person-years in each category. We then computed means or proportions of baseline risk factors according to categories of whole-grain, refined-grain, or total cereal intake. Cox proportional hazards models were used to estimate age- and multivariate-adjusted hazard rate ratios (RRs) for each intake category, as compared with the reference category (rarely or never), for both total mortality and CVD-specific mortality. The multivariate analyses were adjusted for age (in years); body mass index; smoking; alcohol intake; physical activity; history of high blood cholesterol, hypertension, and diabetes; and use of multivitamins. We then conducted stratified analyses according to baseline risk conditions including diabetes (yes or no), hypertension (yes or no), and high cholesterol (yes or no). Tests for a linear trend across increasing categories of breakfast cereal intake were conducted by treating the median intake (servings/d) in each category as a continuous variable. All analyses were conducted with SAS (version 8; SAS Institute Inc, Cary, NC). All P values were two-sided.
RESULTS
At baseline in 1982, 19% of men reported consuming, on average, 1 serving of breakfast cereal/d and 12% reported consuming 1 serving of whole-grain breakfast cereal/d. Men who had a greater intake of cereal products (regardless of type) were older, more physically active, and had a lower prevalence of heavy smoking or overweight. Use of multivitamin supplements and history of diabetes, hypertension, or high cholesterol did not vary appreciably across categories of breakfast cereal intake (Table 1).
View this table:
TABLE 1 . Baseline characteristics according to intakes of whole-grain, refined-grain, and total breakfast cereals in the Physicians’ Health Study enrollment cohort1
During an average of 5.5 y (range: 1–6.6 y) of follow-up, we identified 3114 deaths from all causes, including 1381 due to CVD (488 due to myocardial infarctions and 146 due to strokes). There was a graded inverse relation of intake of whole-grain breakfast cereals to total mortality (Table 2). Relative to the men in the lowest category of whole-grain breakfast cereal intake (rarely), the age-adjusted RR of total mortality for men in the highest category ( 1 serving/d) was 0.73 (95% CI: 0.65, 0.82; P for trend = 0.0001). Higher intakes of whole-grain breakfast cereals were also associated with lower risks of mortality from CVD (age-adjusted RR: 0.72; 95% CI: 0.61, 0.85; P for trend < 0.001) or myocardial infarctions (RR: 0.77; 95% CI: 0.57, 1.01; P for trend = 0.01) (Table 3). In multivariate models that also adjusted for cigarette smoking; alcohol intake; physical activity; body mass index; history of diabetes, high cholesterol, and hypertension; and use of multivitamins, these associations remained significant. When comparing the highest category of whole-grain breakfast-cereal intake to the lowest category, the multivariate RRs were 0.83 (95% CI: 0.73, 0.94; P for trend < 0.001) for total mortality, 0.80 (95% CI: 0.66, 0.97; P for trend = 0.008) for CVD mortality, and 0.71 (95% CI: 0.51, 0.98; P for trend = 0.01) for myocardial infarction mortality. In contrast, intakes of total and refined-grain breakfast cereals were not significantly associated with total and CVD-specific mortality (Tables 2 and 3).
View this table:
TABLE 2 . Adjusted hazard rate ratios and 95% CIs of total mortality by cereal intake in the Physicians’ Health Study enrollment cohort1
View this table:
TABLE 3 . Adjusted hazard rate ratios and 95% CIs of cardiovascular disease (CVD)–specific mortality by cereal intake in the Physicians’ Health Study enrollment cohort1
We further examined the associations between the intake of breakfast cereals and mortality risk in the men who did not have a history of diabetes, high cholesterol, or hypertension at baseline. In these subgroup analyses in which participants with a history of diabetes, high cholesterol, or hypertension were further excluded, whole-grain cereal intake remained inversely associated with total and CVD-specific mortality, whereas total or refined-grain cereal intake remained unrelated to mortality (Figure 1). For example, when the highest and lowest categories of whole-grain cereal intake were compared, the multivariate RRs among participants without a history of diabetes were 0.76 (95% CI: 0.65, 0.90; P for trend = 0.0001) for total mortality and 0.73 (95% CI: 0.55, 0.96; P for trend = 0.004) for CVD-specific mortality. When the highest and lowest categories of refined-grain cereal intake were compared, the multivariate RRs among participants with a history of diabetes were 1.39 (95% CI: 0.97, 1.98; P for trend = 0.07) for total mortality and 1.47 (95% CI: 0.96, 2.24; P for trend = 0.08) for CVD-specific mortality.
FIGURE 1. . Multivariate-adjusted hazard rate ratios (RRs) of total mortality by category of intake of whole-grain () and refined-grain () breakfast cereals in 65 390 participants in the Physicians’ Health Study enrollment cohort who did not have a history of diabetes, hypertension, or high cholesterol at baseline. For whole-grain and refined-grain cereals, P for trend = 0.0001 and 0.31.
DISCUSSION
In this large prospective study of US male physicians who were followed for 5.5 y, we found a modest, graded association between higher intakes of whole-grain breakfast cereals and lower risks of total and CVD-specific mortality that was independent of known CVD risk factors. In contrast, intake of refined-grain cereals was not associated with total and CVD-specific mortality.
Several large prospective studies that examined the relation between whole-grain intake and the risk of chronic disease found that increased intakes of whole-grain products are associated with reduced risks of type 2 diabetes (7, 8), hypertension (6, 9), and CVD (10–12). Our findings are generally consistent with the results of these previous prospective studies and extend those results with an assessment of the relation between types of cereal grains and total and CVD-specific mortality in men. The magnitude of the association between whole-grain intake and mortality, with the exception of stroke mortality, is in general agreement with that of the study by Jacobs et al (14), which was conducted in a large cohort of women. Whereas neither total nor refined-grain cereal intake appeared to be significantly related to total and CVD-specific mortality in the entire cohort of male physicians, we found a positive, albeit not significant, association between the intake of refined-grain cereals and CVD-specific mortality among those participants who had a history of diabetes. Although these findings in subgroup analysis should be interpreted cautiously, they nevertheless suggest that a higher intake of refined-grain cereal products may have harmful effects on CVD endpoints among individuals with glucose intolerance, a suggestion that needs to be confirmed by future studies.
Several potential limitations are worthy of discussion. First, the observed inverse association between whole-grain breakfast cereal intake and total mortality could be due to confounding by other heart-healthy lifestyle factors (eg, smoked less, exercised more, and had lower body mass index) and dietary factors associated with higher whole-grain intake. However, the apparent protective association with whole-grain cereal intake persisted in multivariate models accounting for known coronary risk factors. Also, the homogeneity in education and occupation of our study population of male physicians should also have minimized confounding by socioeconomic variables that may have affected the opportunity for screening and treatments. As discussed in Subjects and Methods, the SFFQ used in this study was an abbreviated one that did not provide a comprehensive assessment of the participants’ usual diet; thus, we could not adjust for other dietary intakes. However, previous dietary studies did not identify any dietary factors that may confound the association between grain intake and total mortality in multivariate analyses in which multiple dietary variables are included in the same model (14). Vegetables and liver or meat are the only other food groups that were assessed in the SFFQ. Even after further adjustment for intakes of vegetables and liver, the RRs for the comparison between the highest and lowest categories of whole-grain intake were 0.82 (95% CI: 0.73, 0.94) for total mortality and 0.79 (95% CI: 0.65, 0.96) for CVD-specific mortality.
Second, some high-risk conditions such as hypercholesterolemia, diabetes, and hypertension may lead to changes in dietary habits and therefore confound the association between the intake of whole-grain breakfast cereals and total mortality. However, any biases from these conditions would tend to attenuate the protective effect of whole-grain cereal intake because the tendency would be for men to increase their intake of whole grains if they perceived themselves to have an increased risk of CVD. Moreover, the inverse association persisted when men with these conditions at baseline in 1982 were excluded from the main analysis. Finally, we observed a specific relation of lower total and CVD-specific mortality with greater intakes of whole-grain cereals but not with greater intakes of either total or refined-grain cereals even though greater intakes of total or refined-grain cereals were also related to heart-healthy lifestyle factors in this study. The specificity of the whole-grain and mortality relation argues against confounding as a full explanation for our findings.
As in any observational study of diet and disease, one major concern is the measurement error inherent in dietary assessment. Our SFFQ only assessed a limited number of foods, which hampered our ability to adjust for total energy intake and other dietary factors that may be important in affecting mortality. In addition, the definition of whole grain has always been contentious, and some of the difficulties were pointed out by Jenkins et al (18). In contrast with intact grain products or stone-ground flour, most "whole grains" products are reconstituted after milling from the 3 original components: the starchy endosperm, the wheat bran milled to a specific particle size, and the wheat germ. To prevent rancidity of the polyunsaturated fats and to lower the fat content of the products, wheat germs are often heat-treated and, in some instances, may not be added back in the production process (19). The final whole-grain products, however, remain high in dietary fiber. For this reason, we and others (12, 14) have consistently defined whole grains as those products that contain 25% whole grain or bran by weight. In dealing with the issue of misclassification, our approach was to compare findings from sensitivity analyses that either keep bran in or leave it out of the classification system. Results from these sensitivity analyses were similar. Furthermore, the SFFQ was designed to rank and contrast long-term average cereal intakes among participants rather than to assess absolute intakes. Thus, total cereal intakes may have been underestimated. In a prospective setting, however, such an underestimation of intake would be unlikely to differ by survival status and would thus tend to cause an underestimation of the diet-disease relation rather than a spurious relation. Therefore, the 20% lower total mortality observed with greater intakes of whole-grain cereals may be a conservative estimate.
Although the exact mechanisms responsible for the benefits of whole grains remain to be elucidated, multiple mechanistic pathways, including reduction in lipids and blood pressure and improvement of insulin sensitivity and glucose tolerance, have been suggested. Much attention has focused on individual components of whole grains, such as dietary antioxidants, minerals, enzyme inhibitors, dietary fiber, folate, and B vitamins, which may independently or jointly contribute to the lower mortality associated with greater intakes of whole grains (6, 12, 20–22).
Identifying a direct relation between the intake of specific types of cereal and mortality can provide a scientific rationale for formulating dietary guidelines. Until recently, however, the prevailing health advice has been to increase the intake of total grains rather than of specific types of grains to prevent chronic diseases, a recommendation primarily intended to lower the intake of saturated fat and cholesterol. Several developments have drawn attention to the importance of whole grains. One of the goals in Healthy People 2010 is to "increase the proportion of persons aged 2 y and older who consume at least 6 daily servings of grain products, with at least 3 being whole grains" (goal 19-7; 2). The 2000 Dietary Guidelines for Americans makes the following recommendation: "Choose a variety of grains daily, especially whole grains. . . eating plenty of whole grains. . . as part of the healthful eating patterns. . . may help protect you against many chronic diseases" (1). Nevertheless, these guidelines still implicitly advocate the intake of a large amount of refined grains, even though large prospective studies have consistently shown no benefit or harmful effects of refined-grain products. Of significant concern is the fact that most grain products consumed in the United States are highly refined (3). Although data directly comparing the effects of refined-grain products to those of saturated fats on CVD mortality are sparse, our data indicate that whole-grain products should be a good substitute either for products with a high saturated fat content or for refined-grain cereals. Compared with whole-grain products, refined-grain products often contain lower amounts of many potentially beneficial micronutrients, antioxidants, minerals, phytochemicals, and fiber. In our cohort, the men in the highest category of whole-grain breakfast-cereal intake (only 12% of total participants) consumed 1 serving of whole-grain breakfast cereals/d and had an 20% lower risk of total mortality than did the men in the lowest category, who rarely or never consumed whole-grain breakfast cereals. Thus, substituting high-fiber products (ie, whole-grain cereals, fruit, and vegetables) for low-fiber refined-grain products may have a significant effect on public health (22–26).
In conclusion, in this large population of men, both total mortality and CVD-specific mortality were inversely associated with intakes of whole-grain breakfast cereals but not with intakes of refined-grain breakfast cereals. These prospective data highlight the importance of distinguishing whole-grain cereal products from refined-grain ones for the prevention of premature death.
ACKNOWLEDGMENTS
SL participated in the study design, data collection, data analysis, and the writing of the first draft of the manuscript. HDS, WCW, JEB, and JAEM participated in the study design and data collection. All authors participated in the writing of the manuscript. None of the authors had any financial conflict of interest. In 2001 SL received honoraria from General Mills Co for a presentation unrelated to this article.
REFERENCES