点击显示 收起
1 From the Division of Preventive Medicine and the Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, and the Departments of Epidemiology and Nutrition, Harvard School of Public Health, Boston.
2 Supported by National Institutes of Health grant CA87969 and by Clinical Scientist Development Award K08-DK 02767 from the National Institute of Diabetes and Digestive and Kidney Diseases (to SL). In addition, for activities related to the Nurses' Health Study, we received modest resources at various times and for various periods since 1 January 1993 from the Alcoholic Beverage Medical Research Foundation, the American Cancer Society, Amgen, the California Prune Board, the Centers for Disease Control and Prevention, the Ellison Medical Foundation, the Florida Citrus Growers, the Glaucoma Medical Research Foundation, Hoffmann-La Roche, Kellogg's, General Mills Company, Lederle, the Massachusetts Department of Public Health, Mission Pharmacal, the National Dairy Council, Rhone Poulenc Rorer, the Robert Wood Johnson Foundation, Roche, Sandoz, the US Department of Defense, the US Department of Agriculture, the Wallace Genetics Fund, Wyeth-Ayerst, Merck, and private contributors. 3 Address reprint requests to S Liu, Division of Preventive Medicine, Brigham and Women's Hospital, 900 Commonwealth Avenue East, Boston, MA 02215. E-mail: siminliu{at}hsph.harvard.edu.
ABSTRACT
Background: Although increased consumption of dietary fiber
and grain products is widely recommended to maintain healthy
body weight, little is known about the relation of whole grains to
body weight and long-term weight changes.
Objective: We examined the associations between the intakes of dietary fiber and whole- or refined-grain products and weight gain over time.
Design: In a prospective cohort study, 74 091 US female nurses, aged 38-63 y in 1984 and free of known cardiovascular disease, cancer, and diabetes at baseline, were followed from 1984 to 1996; their dietary habits were assessed in 1984, 1986, 1990, and 1994 with validated food-frequency questionnaires. Using multiple models to adjust for covariates, we calculated average weight, body mass index (BMI; in kg/m2), long-term weight changes, and the odds ratio of developing obesity (BMI =" BORDER="0"> 30) according to change in dietary intake.
Results: Women who consumed more whole grains consistently weighed less than did women who consumed less whole grains (P for trend < 0.0001). Over 12 y, those with the greatest increase in intake of dietary fiber gained an average of 1.52 kg less than did those with the smallest increase in intake of dietary fiber (P for trend < 0.0001) independent of body weight at baseline, age, and changes in covariate status. Women in the highest quintile of dietary fiber intake had a 49% lower risk of major weight gain than did women in the highest quintile (OR = 0.51; 95% CI: 0.39, 0.67; P < 0.0001 for trend).
Conclusion: Weight gain was inversely associated with the intake of high-fiber, whole-grain foods but positively related to the intake of refined-grain foods, which indicated the importance of distinguishing whole-grain products from refined-grain products to aid in weight control.
Key Words: Whole grains • refined grains • dietary fiber • body weight • prospective study • obesity • weight gain • women
INTRODUCTION
According to the third National Health and Nutrition Examination Survey (NHANES III), which ended in 1994, nearly
100 million US adults were overweight or obese (1), which
puts them at increased risk of hypertension, dyslipidemia, type
2 diabetes, heart disease, and many other chronic disorders (2).
Weight control is clearly an important goal for many US adults.
At any given time in the United States, 44% of women and
29% of men report that they are attempting to lose weight, and
5 million of these adults have used prescription drugs for
treatment of obesity (3, 4). Despite the public's increased
awareness of the health hazards associated with overweight and
obesity, recent reports from the Centers for Disease Control
and Prevention indicate that the prevalence of overweight and
obesity in the United States continues to increase (5, 6).
At both individual and population levels, strategies that improve nutrition and increase physical activity are fundamental to the control of the epidemic of overweight and obesity (2, 7). Yet the long-term efficacy of any specific dietary approach to weight control remains to be determined (8-10). Because of the belief that diets rich in fiber are generally low in saturated fat, many national authorities have long recommended greater consumption of grain products to control weight (11-14). Whole grains may have beneficial effects on weight control through promoting satiety (15-17). The intake of whole grains may also slow starch digestion or absorption, which leads to relatively lower insulin and glucose responses that favor the oxidation and lipolysis of fat rather than its storage (15-17). However, most grain products consumed in the United States are highly refined (17, 18). Refined-grain products have a higher starch content but a lower fiber content (ie, greater energy density) than do whole grains. Concentrations of vitamins, minerals, essential fatty acids, and phytochemicals that are important in carbohydrate metabolism are also lower in refined grains (19).
Indirect evidence from both epidemiologic and short-term experimental studies suggests a beneficial role of a high-fiber diet in weight control (19-21). However, few trials have examined directly the effects of whole grains, as opposed to those of refined grains, on body weight and weight changes. Nor is there any large prospective epidemiologic study directly linking the intake of whole grains to changes in weight over time. To examine further the associations between grain intake and body weight and weight changes, we analyzed prospective data from the Nurses' Health Study (NHS) from 1984 to 1996 using repeated measurements of grain intake and body weight.
SUBJECTS AND METHODS
Study population
The NHS was initiated in 1976 when 121 700 female registered nurses aged 30-55 y answered a mailed questionnaire on
their medical history and lifestyle. The cohort has been followed every 2 y since 1976 to ascertain exposures and incident
diseases. In 1980, we assessed diet with a 61-item semiquantitative food-frequency questionnaire (FFQ) (22). The subjects
who did not satisfy the a priori criterion of reported daily
energy intake between 2514 KJ (600 kcal) and 14 665 KJ
(3500 kcal) were excluded. In 1984 the semiquantitative FFQ
was expanded to include 126 items. Because the expanded
questionnaire contained additional food items that are important in assessments of the details of carbohydrate intake and
because it accorded with our previous studies in this area, we
considered 1984 as the baseline for the analytic cohort of
81 757 women. We further excluded respondents with previously diagnosed diabetes (n = 2518); cardiovascular disease
(CVD), including angina, myocardial infarction, stroke, and
other CVDs (n = 690); and cancers (n = 4458). The final
baseline population consisted of 74 091 women who were
38-63 y old in 1984.
Assessment of whole- and refined-grain consumption
Measurements of whole-grain and refined-grain foods, along
with other aspects of diet, were repeated in 1986, 1990, and
1994 using FFQs nearly identical to the questionnaire used in
1984. For each food, a commonly used unit or portion or
serving size (eg, 1 slice of bread) was specified, and participants were asked how often, on average, during the previous
year they had consumed that amount. Nine responses were
possible, ranging from "never" to "=" BORDER="0"> 6 times/d." On the basis
of the responses to these questions, we calculated the average
grain intake (servings/d) for each participant. The type and
brand of breakfast cereal also were ascertained. We used a
procedure developed by Jacobs et al (23) and Liu et al (24) to
classify foods into whole and refined grains. This classification
scheme was previously used in this cohort for studying the
relations of grain intake to the risk of type 2 diabetes (25) and
CVD (24, 26). Specifically, whole-grain foods included dark
bread, whole-grain breakfast cereal, popcorn, cooked oatmeal,
wheat germ, brown rice, bran, and other grains (eg, bulgur,
kasha, couscous). Refined-grain foods included sweet rolls and
cakes or desserts, white bread, pasta, English muffins, muffins
or biscuits, refined-grain breakfast cereal, white rice, pancakes
or waffles, and pizza. The list of breakfast cereals reported in
the semiquantitative FFQ was evaluated for whole-grain and
bran content; breakfast cereals with =" BORDER="0"> 25% whole grain or bran
content by weight were classified as whole grain. A full description of the semiquantitative FFQs and data on reproducibility and validity in this cohort were published previously
(22, 27). The performance of the semiquantitative FFQ in
assessing the individual grain products was documented to be
high (28). For example, between the FFQ and detailed diet
records in a sample of the participants, correlation coefficients
were 0.75 for cold breakfast cereal, 0.71 for white bread, and
0.77 for dark bread. Overall, these data indicate that the semiquantitative FFQ provides a reasonably valid measure of average long-term dietary intake.
Measurements of body weight, body mass index, and
changes in weight
Participants self-reported their body weight every 2 y from
1984 to 1996. Height was reported in 1976. In this cohort, body
mass index (BMI; in kg/m2) is minimally correlated with
height (r = -0.03) and highly correlated with height-adjusted
fat mass measured by underwater weighing of young and
middle-aged women (r = 0.8-0.9) (29). We calculated weight
changes from 2-4-y intervals between 1984 and 1996 as well
as from 1984 to 1996. In a sample of 184 participants described
previously (30), the correlation between self-reported and directly measured weight was 0.96.
Statistical analysis
We first conducted a cross-sectional analysis of the associations between baseline covariates and the intakes of whole grains
and refined grains in 1984. Unless otherwise indicated, intake of
whole or refined grains was categorized into quintiles, and we
calculated mean body weight and BMI according to these quintiles. We then plotted the mean attained BMI according to the
intake of whole or refined grains at each 2-y interval from 1984
to 1996. Second, we used generalized estimating equations to
examine the direct relation between changes in intake of whole
or refined grains in relation to changes in weight in the same
period with adjustment for changes in covariates. Changes in
weight were defined as the difference between weight assessed
in each dietary period (1984-1986, 1986-1990, and 1990-1994). Covariates, including age; years of follow-up; change in
exercise (ie, differences in metabolic score); change in smoking status (ie, increase, decrease, no change); change in hormone replacement status status (ie, from current use to no use,
same use, from no use to current use); and changes in intakes
of alcohol, caffeine, and total energy were defined by taking
the differences between 1984 and 1986, between 1986 and
1990, and between 1990 and 1994. Because exercise status was
not assessed in 1984, we carried forward information from
1982. Tests of linear trend across increasing categories of
changes in grain consumption were conducted by treating the
median of intake in categories (servings/d) as a continuous
variable. The model has the following form:
RESULTS
The median intake of energy-adjusted whole grain ranged
from virtually no consumption in the lowest quintile to 1.62
servings · 1000 kcal-1 · d-1 in the highest quintile (Table 1). Women with high intake of whole grains smoked less, exercised more, and were more likely to use postmenopausal hormones. Greater whole-grain intake was also associated with
higher intakes of carbohydrate, protein, dietary fiber, and fruit
and vegetables but with lower intakes of fats, cholesterol, and
alcohol. These baseline associations were generally the opposite for women with greater intake of refined-grain products.
View this table:
TABLE 1. . Age-standardized baseline distributions of covariates according to quintiles (Q) of intake of whole or refined grains among 74,091 US female nurses
aged 38-63 y in 19841
At baseline in 1984, women in the highest quintile of whole-grain intake weighed 0.9 kg less than did women in the
lowest quintile of intake (BMI of 24.5 compared with BMI of
24.9; P for trend < 0.0001) (Figure 1 A), whereas women in
the highest quintile of refined-grain intake weighed 1.2 kg
more than did women in the lowest quintile of intake (BMI of
25.2 compared with BMI of 24.6; P for trend < 0.0001)
(Figure 1B). During a 10-y period from 1984 to 1994, the mean
daily intakes of whole grain increased from 0.64 servings per
1000 kcal to 0.77 servings per 1000 kcal. The correlation
coefficients between any 2 assessments of whole-grain intake
ranged from 0.46 to 0.67, depending on the closeness in time.
Although BMI was inversely associated with whole-grain intake and positively associated with refined-grain intake, average BMI increased with time regardless of levels of grain
intake (Figure 1).
FIGURE 1.. A: Mean BMI from 1984 to 1996 according to quintiles
of whole-grain intake. BMI differences by quintiles of intake, P < 0.0001.
Whole-grain intake x years of follow-up interaction, P = 0.02. B: Mean
BMI from 1984 to 1996 according to quintiles of refined grain intake. BMI
differences by quintiles of intake, P < 0.0001. Refined-grain intake x
years of follow-up interaction, P = 0.01.
Changes in the intakes of whole grains, refined grains, or
dietary fiber in relation to average changes in weight or BMI in
2-4-y intervals (between 2 dietary assessments) are shown in
Table 2. In general, increases in the intake of whole grains
were associated with less weight gain (mean weight gain of
1.58 kg in 2-4 y in the lowest quintile and of 1.07 kg in the
highest quintile; P for trend < 0.0001). In contrast, increases in
the intake of refined grains were related to greater weight gain
(from 0.99 kg to 1.65 kg; P for trend < 0.0001). Similarly,
greater increases in dietary fiber intake were associated with
less weight gain every 2-4 y.
View this table:
TABLE 2. . Change in BMI or in weight according to quintiles (Q) of change in intakes of whole or refined grains in the Nurses' Health Study from 1984 to 19961
We observed similar patterns when we examined changes in
grain intake from 1984 to 1994 in relation to the multivariate-adjusted mean long-term weight changes from 1984 to 1996.
As shown in Table 3, over the entire follow-up period, an
increased intake of dietary fiber and whole grains was associated with significantly less weight gain, whereas the opposite
was evident for increased intake of refined grains. Adjustment
for total energy intake had little effect on these results. During
12 y of follow-up, those who had the greatest increase in
dietary fiber intake gained an average of 1.52 kg less than did
those who had the smallest increase in intake (P for trend
< 0.0001), independent of body weight at baseline; age; and
changes in physical activity, smoking status, alcohol consumption, caffeine intake, and hormone replacement therapy use;
and intakes of different types of fats, protein, and total energy.
After further correction for measurement errors in change in
dietary fiber intake, we estimated that an increase of 12 g in
dietary fiber intake was associated with 3.5 kg (8 lb) less
weight gain in 12 y. Overall, women who consumed a larger
amount of whole grains consistently weighed less in each
interval than did those who consumed a smaller amount of
whole grains; in contrast, women who consumed a larger
amount of refined grains consistently weighed more than did
those who consumed a smaller amount of refined grains (P <
0.0001).
View this table:
TABLE 3. . Average weight gain (in kg) according to quintiles (Q) of change in intake in the Nurses' Health Study from 1984 to 1996
The relations between intake of high-fiber whole-grain foods
and weight change also appeared to differ by BMI status at
baseline (P < 0.0001 for interaction; Figure 2). Specifically,
the beneficial effects of an increased intake of dietary fiber on
weight gain were greater among those who were overweight at
baseline. Among this overweight group, those in the highest
quintile of intake experienced only about one-half (2.65 kg) as
much weight gain as did those in the lowest quintile of intake
(5.10 kg) during the entire follow-up period (P for trend
< 0.0001).
FIGURE 2.. Multivariate-adjusted weight gain in kilograms by quintiles (Q) of change in dietary fiber intake from 1984 to 1994 according to
BMI in 1984. BMI =" BORDER="0"> 25 (), P for trend < 0.0001; BMI < 25 (), P for
trend = 0.01; BMI at baseline x change in fiber intake interaction, P <
0.001. Covariates adjusted for were age; changes in exercise (differences in
metabolic score); changes in smoking status (ie, increase, decrease, no
change); change in hormone replacement therapy status (ie, increase,
decrease, no change); changes in intakes of alcohol, caffeine, and total
energy; changes in intakes of saturated fat, polyunsaturated fat, monounsaturated fat, trans fat, and protein; and BMI at baseline. Median values in
each quintile of change in dietary fiber intake (g/d) were -3.4, 0, 2.2, 4.6,
and 8.9 for quintiles 1, 2, 3, 4, and 5, respectively.
To examine further the association between grain intake and
major weight gain and the development of obesity, we calculated odds ratios for obesity (BMI =" BORDER="0"> 30) and for major weight
gain (=" BORDER="0"> 25 kg) in 12 y according to changes in intake in the
same period. During the 12-y follow-up, 6400 women became
obese and 657 had a major weight gain (Table 4). The odds of
developing obesity or having a major weight gain were generally lower with larger intakes of dietary fiber and whole grains
but not with a larger intake of refined grains. Compared with
women who were in the lowest quintile of fiber intake, women
who increased their intake of dietary fiber to the highest quintile reduced their risk of major weight gain by 49% (OR =
0.51; 95% CI: 0.39, 0.67; P < 0.0001 for trend).
View this table:
TABLE 4. . Adjusted relative odds ratios for attained BMI =" BORDER="0"> 30 or major weight gain (=" BORDER="0"> 25 kg) in 12 y according to quintiles (Q) of change in grain intake in the Nurses' Health Study from 1984 to 1996
DISCUSSION
This 12-y prospective study of middle-aged women provides
both cross-sectional and longitudinal estimates of the effects of
whole or refined grains on body weight and changes in weight.
In the cross-sectional analysis at baseline, body weight was
inversely associated with whole-grain intake but positively
associated with refined-grain intake. Longitudinally, participants who had greater intake of whole grains tended to gain
less weight than did those with greater intake of refined grains.
Compared with women who had decreased intake of high-fiber
or whole-grain products, those who had the greatest increase in
intake of high fiber or whole grains had half the odds of
becoming obese over a 12-y period.
These findings should be interpreted in the context of the study's strengths and limitations. The NHS data are unique in that multiple measurements of diet and body weight are available, which makes it possible to observe the temporal relation between the intake of whole or refined grains and changes in weight. By using each woman as her own control and directly defining changes in grain intake through multiple assessments of diet, we can estimate the amount of weight change associated with changes in the intake of whole or refined grains.
The observational nature of our study design, however, still hampers straightforward interpretation. First, because both intake of whole grain and body weight are time-dependent variables, we cannot determine with complete certainty that changes in grain intake preceded changes in weight. For example, those who had recently gained weight might, in an attempt to lose weight, increase their intake of grain products. Thus, it is not clear whether increased intake of whole grains prevented weight gain, or whether it is just a marker for attempted weight loss among people with recent weight gain. In addition, the general health consciousness that might be associated with both increased whole-grain intake and caloric restriction remains a potential threat to the validity of our findings. To address this issue, we defined changes in weight as well as changes in grain intake by taking the difference between 2 measurements from 1984 to 1996 and with adjustment for previously attained weight in the model so that any systematic biases due to attained weight could be minimized. However, calculating the differences between 2 dietary assessments would increase measurement error by incorporating errors of both assessments, which suggests that our estimates of the potential effects of dietary fiber or whole grains on weight changes might be conservative.
Further complicating the picture is the possibility not only that BMI can change in response to diet but also that attained BMI may lead both to real changes in diet and to changes in self-reporting of diet (ie, a misclassification of dietary intake that is dependent on body weight). For example, it has been reported that those who are obese are more likely to underreport their energy intake, which makes the study of diet and obesity in an observational setting difficult (33-35). However, assessment of whole- grain foods with the questionnaires used in this study had relatively high validity (r > 0.70 compared with dietary record for most whole-grain foods). In multivariate analyses, we adjusted for many potential predictors of underreporting, including age, smoking, physical inactivity, and other known predictors of weight gain, that might minimize this bias. Moreover, results from these multivariate analyses showed that weight gain over time was inversely associated with increases in whole-grain intake but positively related to increases in refined-grain intake. Other dietary factors may have accounted for the observed relations between grain intake and body weight. However, the association between the intake of grain products and weight changes remains almost identical when we conducted a series of sensitivity analyses in which the intakes of red meats, fruit and vegetables, and dairy products were simultaneously included in the same model along with whole grains or refined grains (data not shown). The similar magnitudes of associations compared with the age-adjusted analyses and the specific findings regarding the different effects of whole grain versus refined grain intake on weight gain argue against confounding as a full explanation for our findings.
Diet may affect body weight through multiple pathways, including control of satiety and metabolic efficiency, or through modulation of insulin secretion and action. Total and saturated fats have been the focus of intense scientific scrutiny as potential causes for overweight and obesity, but as yet there is no conclusive evidence directly linking dietary fats to body fat. Whereas there is no doubt that overfeeding animals with a large percentage of energy from fat can cause obesity and insulin resistance (36, 37), the magnitude and long-term significance of the effect of a low-fat diet on weight control in humans remain uncertain (9, 38). In contrast, energy-restricted low-fat diets have been shown to achieve 10% reduction in weight in many feeding trials (39, 40). But much of the weight thus lost is regained within 12 mo, and no long-term efficacy has ever been convincingly shown (41). Moreover, the effects of changes in dietary composition (percentage of energy from fat rather than from carbohydrates) alone on weight loss appeared minimal (41).
Reports based on NHANES III and a recent time-trend analysis indicate that intakes of both total and saturated fat, in terms of the percentage of total energy intake, have been declining in the United States since 1960s (42, 43). In contrast, in the same period, an increase in the intake of refined carbohydrates in the form of processed grains, soft drinks, sugars, and refined flours in the US food supply has been reported to parallel the increased prevalence of obesity and diabetes (44). Consistent with these reports, our data indicate that the intake of refined grains is directly associated with dietary glycemic load and index, which suggests that an overemphasis on a low-fat diet might have contributed to an increase in intake of high-energy and high-glycemic-load foods. Long-term intervention trials may be necessary if we are to better estimate the relation between changes in the intake of refined grains and weight control, but the feasibility of such long-term trials remain unclear.
Experimental data indicate that refined-grain products, unlike whole-grain products, can induce an increase in fat synthesis in animal feeding trials even when the total energy intake is unchanged and body weight remains constant (45). Results from several short-term (a single meal or a single day) feeding trials in humans suggest that the consumption of whole-grain products with a low glycemic index might increase satiety and reduce energy consumption and thus contribute to weight loss (20). Our data are consistent with results from these animal and human experiments. In addition, attained body weight may function as an important modifier for the effects of whole grains, as indicated by the observation that the relation between dietary fiber or whole-grain food intake and weight change also appeared to differ by BMI status at baseline (Figure 2). The beneficial effects of increased intake of dietary fiber on weight gain were most significant among subjects who were overweight at baseline. Several epidemiologic studies of dietary fiber also suggest that the intake of whole grains but not of refined grains is inversely associated with body weight and fat distribution (21, 46). The inherent high-fiber content of most whole-grain foods may help prevent weight gain by increasing appetite control through producing a delay in carbohydrate absorption (47). Moreover, multiple enzyme inhibitors that exist in a whole-grains-fiber complex might directly affect metabolic efficiency (19). This mechanism might serve as yet another way whereby whole grains could have beneficial effects on body weight. Finally, high concentrations of insulin associated with low-fiber refined grains may over the long term lead to weight gain by directing metabolic fuels from oxidation to storage.
In conclusion, in this prospective study of apparently healthy middle-aged women, weight gain was inversely associated with increases in the intake of whole grains but positively associated with increases in the intake of refined grains. This suggests the potential importance of increasing the intake of dietary fiber as well as of distinguishing whole-grain from refined-grain products to control weight.
ACKNOWLEDGMENTS
SL participated in the study design, data collection, and data analysis
and wrote the first draft of the manuscript. WCW, GC, FBH, JEM, and BR
participated in the study design, data collection, and data analysis. The
authors had no conflicts of interest.
REFERENCES