SEARCH

SEARCH BY CITATION

Keywords:

  • body weight changes;
  • weight gain;
  • diet;
  • food patterns;
  • risk factors

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Objective: Our objective was to examine the association between adherence to dietary patterns and weight change in women.

Research Methods and Procedures: Women (51,670, 26 to 46 years old) in the Nurses’ Health Study II were followed from 1991 to 1999. Dietary intake and body weight were ascertained in 1991, 1995, and 1999. A Western pattern, characterized by high intakes of red and processed meats, refined grains, sweets and desserts, and potatoes, and a prudent pattern, characterized by high intakes of fruits, vegetables, whole grains, fish, poultry, and salad dressing, were identified with principal component analysis, and associations between patterns and change in body weight were estimated.

Results: Women who increased their Western pattern score had greater weight gain (multivariate adjusted means, 4.55 kg for 1991 to 1995 and 2.86 kg for 1995 to 1999) than women who decreased their Western pattern score (2.70 and 1.37 kg for the two time periods), adjusting for baseline lifestyle and dietary confounders and changes in confounders over time (p < 0.001 for both time periods). Furthermore, among women who increased their prudent pattern score, weight gain was smaller (multivariate-adjusted means, 1.93 kg for 1991 to 1995 and 0.66 kg for 1995 to 1999) than among women who decreased their prudent pattern score (4.83 and 3.35 kg for the two time periods) (p < 0.001). The largest weight gain between 1991 and 1995 and between 1995 and 1999 was observed among women who decreased their prudent pattern score while increasing their Western pattern score (multivariate adjusted means, 6.80 and 4.99 kg), whereas it was smallest for the opposite change in patterns (0.87 and −0.64 kg) (p < 0.001).

Discussion: Adoption of a Western dietary pattern is associated with larger weight gain in women, whereas a prudent dietary pattern may facilitate weight maintenance.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The prevalence of obesity in the U.S. has increased rapidly during the last decades (1, 2, 3). Two-thirds of American adults were overweight or obese, and 30% were obese in 2001 to 2002 according to the National Health and Nutrition Examination Survey (3). Although weight gain results from an imbalance between energy intake and expenditure, little is known about dietary factors contributing to weight gain and the development of obesity. Numerous clinical trials, mostly short term, have examined the role of individual macronutrients such as fat, carbohydrates, and protein in weight loss among overweight and obese individuals, but few studies are conclusive, and the effects of dietary modification of the macronutrient composition on weight loss remain controversial (4). In addition, little is known about what dietary characteristics determine long-term energy balance, which is a different question from which dietary intervention may be effective for short-term weight loss. Most likely, the overall eating pattern, rather than intakes of single nutrients or foods, affects long-term weight gain or maintenance because dietary patterns reflect cumulative effects of the diet. However, few studies evaluated the role of dietary patterns in long-term body weight regulation (5, 6). Therefore, we examined the association between dietary patterns and weight change over 8 years in a large cohort of young and middle-aged women.

Research Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Study Population

The Nurses’ Health Study II is a prospective cohort study of 116,671 female U.S. nurses. Participants were 24 to 44 years of age at study initiation in 1989. This cohort is followed using biennial mailed questionnaires with a follow-up rate exceeding 90% for every 2-year period, and we estimate that there is almost complete (98%) ascertainment of mortality. Participants completed self-administered food frequency questionnaires (FFQs)1 in 1991, 1995, and 1999. For the analyses presented here, women were excluded from the baseline population if they did not complete dietary questionnaires, if more than nine food items were left blank on it, if the reported dietary intake was implausible with regard to total energy intake (i.e., <500 or >3500 kcal/d), if they had a history of diabetes or cardiovascular disease before 1995 or reported the diagnosis of cancer (except non-melanoma skin cancer) on any questionnaire, if they did not report body weight on any questionnaire, if they had no data on physical activity assessed in 1991 or 1997, or if they were pregnant at the time of the 1991, 1995, or 1999 questionnaire administration. These exclusions left a total of 51,670 women for the analyses. The study was approved by the institutional review boards at the Harvard School of Public Health and the Brigham and Women's Hospital; completion of the self-administered questionnaire was considered to imply informed consent.

Dietary Assessment

In 1991, the mailed questionnaire included a 133-food item semiquantitative FFQ to obtain dietary information. Women were asked how often they had consumed a commonly used unit or portion size of each food on average over the previous year. There were nine possible responses, ranging from never to six or more times per day. Similar FFQs were used to collect dietary information in 1995 and 1999. Foods from the FFQs were classified into 39 food groups based on nutrient profiles or culinary usage. This classification follows that of previous studies in similar cohorts among U.S. health professionals that used similar dietary assessment instruments (7, 8). Nutrient intakes were computed by multiplying the frequency response by the nutrient content of the specified portion sizes. Values for nutrients were derived from the U.S. Department of Agriculture sources (9) and supplemented with information from manufacturers. Intake of dietary fiber and the glycemic index (based on glucose as reference) were energy-adjusted using the residuals method (10). Intake of fat, carbohydrates, and protein was expressed as nutrient density (percentage of total energy intake) (10). The validity and reliability of FFQs similar to those used in the Nurses’ Health Study II have been described elsewhere (11, 12, 13).

Assessment of Non-dietary Exposures

Information on age, weight, smoking status, contraceptive use, postmenopausal hormone replacement therapy, and pregnancies was collected by biennial questionnaires. We calculated BMI as the ratio of weight (in kilograms) to squared height (in meters squared), height being assessed at baseline only. Self-reports of body weight were highly correlated with technician-measured weights (r = 0.96) in the Nurses’ Health Study I (14). Physical activity was assessed with the 1991 and 1997 questionnaires and was computed as metabolic equivalents per week using the duration per week of various forms of exercise, weighting each activity by its intensity level. Correlations between physical activity reported on the questionnaire and recorded in diaries or by recalls were high in our cohort (0.62 and 0.79) (15). Because physical activity was not assessed in 1995 and 1999, we used the 1997 estimate for both time-points instead.

Statistical Analysis

Dietary patterns were defined from principal component analysis based on the 39 predefined food groups using the PROC FACTOR procedure in SAS and carried out for each dietary questionnaire separately (16). The factors are linear composites of the optimally weighted observed variables. To identify the number of factors to be retained, we used the eigenvalue > 1.0 criterion and a plot of the eigenvalues (16). The identified two-pattern structure was rotated with the varimax rotation to increase its interpretability while maintaining the orthogonality of patterns. Factor scores, i.e., the individual values of the factors, were saved for the two factors for each study participant. They were computed by weighting each factor loading by the factor's eigenvalue, multiplying these weights with subject's corresponding standardized food group intake, and summing these products. Because factor scores are strongly correlated with total energy intake, they were energy-adjusted using the residuals method (10). We categorized the scores based on quintiles and classified participants according to change in pattern scores over time, considering scores within the lower two quintiles as low and scores within the upper two quintiles as high scores. Because all women have scores for both patterns and at all time-points, all groupings are based on the same number of women. We calculated the mean weight changes for groups defined by change in pattern scores from 1991 to 1995, 1995 to 1999, and 1991 to 1999 adjusting for age, alcohol intake, physical activity, smoking, and other lifestyle and dietary confounders at baseline for each time period. We adjusted for changes in these covariates in a separate model to account for variation in lifestyle and dietary behavior. We also adjusted in separate models for changes in soft drink intake, which have been shown to be associated with weight change in this cohort (17), and changes in the score of the dietary pattern not considered as the main exposure. We furthermore cross-classified participants according to changes in both pattern scores and estimated the mean weight change for these groups.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We identified a two-pattern structure with principal component analysis for each time-point, 1991, 1995, and 1999. Correlations with individual foods were consistent over time (Table 1). The pattern explaining the largest part of variance among food items was labeled the prudent pattern and was characterized by higher intake of fruits, vegetables, whole grains, fish, poultry, and salad dressing, whereas the other pattern, labeled the Western pattern, was characterized by higher intakes of red and processed meats, refined grains, sweets and desserts, and potatoes.

Table 1.  Factor loading matrix for two dietary patterns from FFQs completed in 1991, 1995, and 1999 by 51,670 women
 199119951999
Food or food groupsPrudentWesternPrudentWesternPrudentWestern
  1. Factor loadings ≥ 0.30 are marked in bold. FFQ, food frequency questionnaire.

Other vegetables0.680.110.700.110.690.11
Green, leafy vegetables0.67−0.020.650.010.610.03
Dark-yellow vegetables0.63−0.010.59−0.030.55−0.06
Fruit0.62−0.040.61−0.070.59−0.13
Cruciferous vegetables0.580.020.580.020.57−0.01
Tomatoes0.540.140.440.330.460.31
Legumes0.500.170.530.100.500.05
Fish and other seafood0.440.010.390.100.430.04
Oil and vinegar salad dressing0.43−0.010.500.050.550.02
Whole grains0.430.050.430.030.41−0.03
Poultry0.400.050.240.200.310.22
Garlic0.38<−0.010.41<−0.010.43−0.02
Water0.38−0.110.39−0.130.39−0.17
Red meats−0.100.61−0.060.55−0.040.62
Processed meats−0.120.58−0.100.54−0.050.49
French fries−0.210.50−0.220.51−0.200.54
Refined grains0.140.470.200.430.100.51
Sweets and desserts−0.070.42−0.080.43−0.040.39
Potatoes0.160.410.200.370.130.43
Eggs0.100.390.060.390.210.20
Snacks0.010.390.100.330.100.33
High-fat dairy products0.040.360.010.340.090.31
Margarine0.070.36−0.020.38<−0.010.32
Pizza−0.060.34−0.090.39−0.080.40
Mayonnaise0.140.340.170.360.210.31
Sugar-sweetened soft drinks−0.190.30−0.190.28−0.190.30
Diet soft drinks<0.010.04−0.040.13−0.070.22
Low-fat dairy products0.27<0.010.240.010.22−0.05
Condiments0.220.220.310.200.300.19
Soups0.060.270.020.310.050.31
Fruit juices0.220.150.220.100.170.08
Beer0.020.050.030.060.030.07
Wine0.13−0.070.17−0.090.19−0.10
Liquor0.030.020.030.020.030.03
Coffee0.12−0.010.11−0.010.120.01
Tea0.020.140.070.110.170.03
Butter0.010.180.030.180.100.15
Nuts0.200.250.150.220.240.11
Organ meats−0.020.05−0.040.04−0.020.04

Women with a low prudent pattern score in both 1991 and 1995 were slightly younger, were less physically active, more likely to smoke, and had a diet with a relatively larger proportion of fat but lower proportion of carbohydrates and a higher glycemic index but lower cereal fiber intake at baseline, compared with women with a consistently high prudent pattern score (Table 2). Differences in baseline characteristics between women who either increased their pattern score or decreased it between 1991 and 1995 were relatively minor. Similarly, baseline characteristics for women who increased their Western pattern score between 1991 and 1995 were relatively similar to women who decreased their Western pattern score (Table 3). However, women with consistently high Western pattern score had higher body weight and BMI, were less physically active, more likely to smoke, had a higher proportion of dietary fat and a lower proportion of carbohydrates, a higher glycemic index, and lower cereal fiber intake compared with women who had a low pattern score at both time-points.

Table 2.  Age-standardized 1991 characteristics according to time trends in energy-adjusted prudent dietary pattern in 51,670 women
 Change* in prudent dietary pattern between 1991 and 1995
VariableLow-lowHigh-highLow-highHigh-lowOther
  • MET, metabolic equivalent. Means for low-low and high-high were significantly (p < 0.001) different; means for low-high and high-low were significantly (p < 0.001) different except for age (p = 0.04) and carbohydrates (p = 0.91); χ2 tests for categorical variables were significant (p < 0.001) except for hormone replacement therapy (p = 0.54).

  • *

    Low, lower two quintiles of pattern score; high, upper two quintiles of pattern score.

  • BMI is calculated as weight in kilograms divided by the square of the height in meters.

  • Physical activity was computed as metabolic equivalents per week using the duration per week of various forms of exercise, weighting each activity by its intensity level.

N14,65414,5392,5032,44017,534
Age (years), mean35.737.336.136.436.5
Weight (kg), mean66.765.467.868.965.8
BMI (kg/m2), mean24.624.024.925.324.2
Physical activity (METs), mean14.427.918.321.219.6
Currently smoking (%)13.89.212.311.810.5
Currently using oral contraceptives (%)10.910.212.010.710.9
Currently receiving hormone replacement therapy (%)2.22.52.31.92.4
Diet, mean     
 Total energy (kcal/d)1,7911,8121,8581,8091,699
 Alcohol (g/d)2.54.03.42.93.2
 Total fat (energy percentage)33.829.032.831.331.7
 Carbohydrates (energy percentage)48.851.549.149.149.3
 Protein (energy percentage)18.020.318.520.519.6
 Glycemic index55.352.454.652.953.8
 Cereal fiber (grams)4.96.55.35.75.8
Table 3.  Age-standardized 1991 characteristics according to time trends in energy-adjusted Western dietary pattern in 51,670 women
 Change* in Western dietary pattern between 1991 and 1995
VariableLow-lowHigh-highLow-highHigh-lowOther
  • MET, metabolic equivalent. Means for low-low and high-high were significantly (p < 0.001) different except for total energy (p = 0.29); means for low-high and high-low were significantly (p < 0.001) different except for alcohol (p = 0.31); χ2 tests for categorical variables were significant (p < 0.001) for smoking, but not for oral contraceptive use (p = 0.03) and hormone replacement therapy (p = 0.16).

  • *

    Low, lower two quintiles of pattern score; high, upper two quintiles of pattern score.

  • BMI is calculated as weight in kilograms divided by the square of the height in meters.

  • Physical activity was computed as metabolic equivalents per week using the duration per week of various forms of exercise, weighting each activity by its intensity level.

N13,41713,4283,4343,45917,932
Age (years), mean36.836.336.136.736.4
Weight (kg), mean63.768.468.266.465.8
BMI (kg/m2), mean23.425.225.124.324.2
Physical activity (METs), mean28.014.821.118.119.2
Currently smoking (%)7.015.510.612.511.0
Currently using oral contraceptives (%)10.710.510.211.111.0
Currently receiving hormone replacement therapy (%)2.22.32.92.02.4
Diet, mean     
 Total energy (kcal/d)1,8101,8031,8591,8111,693
 Alcohol (g/d)4.02.63.43.23.1
 Total fat (energy percentage)27.435.429.634.131.6
 Carbohydrates (energy percentage)53.446.650.647.949.6
 Protein (energy percentage)19.918.520.518.719.4
 Glycemic index52.655.052.854.454.0
 Cereal fiber (grams)6.94.76.05.25.6

Weight change in women with a consistently low prudent pattern score appeared to be not different from weight change in women with consistently high prudent pattern score (Table 4). However, average weight gain adjusted for baseline age was smallest among those women who increased their prudent pattern score (1991 to 1995, 1.46 kg; 1995 to 1999, 0.25 kg; 1991 to 1999, 3.75 kg) and largest among women who decreased their prudent pattern score (1991 to 1995, 5.55 kg; 1995 to 1999, 3.73 kg; 1991 to 1999, 7.98 kg) (p < 0.001). Adjustment for baseline confounders and changes in confounders over time only slightly attenuated this association. Weight gain appeared to be larger in women who had a consistently high Western pattern score across time, compared with women with low Western pattern score (Table 5). However, this association was attenuated after adjustment for confounding factors, even though it remained statistically significant. The multivariate-adjusted weight gain was 4.90 kg for women with low-low Western pattern score and 5.62 kg for women with high-high Western pattern score for the time period 1991 to 1999 (p < 0.001). However, average weight gain adjusted for baseline age was largest among those women who increased their Western pattern score (1991 to 1995, 5.20 kg; 1995 to 1999, 3.23 kg; 1991 to 1999, 7.81 kg) and smallest among women who decreased their Western pattern score (1991 to 1995, 2.10 kg; 1995 to 1999, 0.85 kg; 1991 to 1999, 4.04 kg) (p < 0.001). This difference largely persisted after adjustment for baseline confounders and changes in confounders over time.

Table 4.  Mean weight change (kilograms) according to changes in energy-adjusted prudent dietary pattern over the same time periods in 51,670 women
 Change in energy-adjusted prudent dietary pattern*
Time periodLow-lowHigh-highLow-highHigh-lowOther
  • Data are means ± standard error. Means for low-high and high-low were significantly (p < 0.001) different; means for low-low, high-high, and other were all significantly (p < 0.001) different from low-high and high-low.

  • *

    Low, lower two quintiles of pattern score; high, upper two quintiles of pattern score; change in pattern corresponds to the same time period as weight change.

  • Model 1, adjusted for baseline age (continuous).

  • Model 2, model 1 + baseline alcohol intake (0, 0.1 to 4.9, 5.0 to 9.9, 10+ g/d), physical activity (quintiles metabolic equivalent score), smoking (never, past, current, missing), postmenopausal hormone use (no, current or past, missing), oral contraceptive use (no, current, missing), cereal fiber intake (quintiles), total fat intake (quintiles), and baseline BMI (continuous).

  • §

    Model 3, model 2 + changes in confounders between time periods (except BMI and except physical activity for 1995 to 1999).

  • Model 4, model 3 + change in intake of sugar-sweetened soft drinks, change in intake of diet soft drinks, and change in Western pattern.

1991 to 1995     
 N14,65414,5392,5032,44017,534
 Model 13.45 ± 0.053.17 ± 0.051.46 ± 0.125.55 ± 0.123.22 ± 0.04
 Model 23.45 ± 0.053.17 ± 0.051.42 ± 0.125.44 ± 0.123.23 ± 0.04
 Model 3§3.20 ± 0.053.41 ± 0.051.70 ± 0.125.20 ± 0.123.24 ± 0.04
 Model 43.20 ± 0.053.46 ± 0.051.93 ± 0.124.83 ± 0.123.22 ± 0.04
1995 to 1999     
 N14,73914,6312,4152,40117,484
 Model 12.39 ± 0.052.01 ± 0.050.25 ± 0.123.73 ± 0.122.05 ± 0.04
 Model 22.33 ± 0.052.08 ± 0.050.32 ± 0.123.69 ± 0.122.05 ± 0.04
 Model 3§2.33 ± 0.052.07 ± 0.050.42 ± 0.123.66 ± 0.122.05 ± 0.04
 Model 42.17 ± 0.052.24 ± 0.050.66 ± 0.123.35 ± 0.122.05 ± 0.04
1991 to 1999     
 N14,01913,9812,9433,00417,723
 Model 15.71 ± 0.065.15 ± 0.063.75 ± 0.147.98 ± 0.135.27 ± 0.06
 Model 25.63 ± 0.075.24 ± 0.073.68 ± 0.147.84 ± 0.135.30 ± 0.06
 Model 3§5.44 ± 0.065.41 ± 0.063.84 ± 0.137.70 ± 0.135.31 ± 0.05
 Model 45.30 ± 0.075.62 ± 0.074.12 ± 0.157.16 ± 0.155.30 ± 0.05
Table 5.  Mean weight change (kilograms) according to changes in energy-adjusted Western dietary pattern over the same time periods in 51,670 women
 Change in energy-adjusted Western dietary pattern*
Time periodLow-lowHigh-highLow-highHigh-lowOther
  • Data are means ± standard error. Means (Model 4) for low-high and high-low were significantly (p < 0.001) different; means for low-low, high-high, and other were all significantly (p < 0.001) different from low-high in all time intervals and significantly different from high-low in 1995 to 1999 (p = 0.0018 for 1991 to 1995, p = 0.30 for 1991 to 1999).

  • *

    Low, lower two quintiles of pattern score; high, upper two quintiles of pattern score; change in pattern corresponds to the same time period as weight change.

  • Model 1, adjusted for baseline age (continuous).

  • Model 2, model 1 + baseline alcohol intake (0, 0.1 to 4.9, 5.0 to 9.9, 10+ g/d), physical activity (quintiles metabolic equivalent score), smoking (never, past, current, missing), postmenopausal hormone use (no, current or past, missing), oral contraceptive use (no, current, missing), cereal fiber intake (quintiles), total fat intake (quintiles), and BMI (continuous).

  • §

    Model 3, model 2 + changes in confounders between time periods (except BMI and except physical activity for 1995 to 1999).

  • Model 4, model 3 + change in intake of sugar-sweetened soft drinks, change in intake of diet soft drinks, and change in prudent pattern.

1991 to 1995     
 N13,41713,4283,4343,45917,932
 Model 12.79 ± 0.053.58 ± 0.055.20 ± 0.102.10 ± 0.103.32 ± 0.04
 Model 22.80 ± 0.063.57 ± 0.065.10 ± 0.102.14 ± 0.103.33 ± 0.04
 Model 3§3.07 ± 0.063.30 ± 0.064.85 ± 0.102.46 ± 0.103.32 ± 0.04
 Model 43.08 ± 0.063.31 ± 0.064.55 ± 0.102.70 ± 0.103.32 ± 0.04
1995 to 1999     
 N13,78713,6943,2323,18317,774
 Model 11.83 ± 0.052.46 ± 0.053.23 ± 0.100.85 ± 0.102.15 ± 0.04
 Model 21.70 ± 0.062.58 ± 0.063.16 ± 0.100.96 ± 0.112.15 ± 0.04
 Model 3§1.78 ± 0.062.47 ± 0.063.01 ± 0.101.21 ± 0.112.16 ± 0.04
 Model 41.88 ± 0.062.35 ± 0.062.86 ± 0.101.37 ± 0.112.17 ± 0.04
1991 to 1999     
 N12,97612,8573,9383,87918,020
 Model 14.45 ± 0.066.06 ± 0.077.81 ± 0.124.04 ± 0.125.46 ± 0.05
 Model 24.59 ± 0.075.93 ± 0.077.68 ± 0.124.11 ± 0.125.47 ± 0.05
 Model 3§4.84 ± 0.075.68 ± 0.077.45 ± 0.124.43 ± 0.125.45 ± 0.05
 Model 44.90 ± 0.075.62 ± 0.077.03 ± 0.124.75 ± 0.125.46 ± 0.05

We cross-classified participants according to both changes in prudent pattern and Western pattern scores and estimated the weight change for the time period 1991 to 1995 and 1995 to 1999. Weight gain was not significantly different between women who had a consistently low prudent and a high Western pattern score compared with the opposite extreme. However, weight gain was largest among women who increased their Western pattern score but decreased their prudent pattern score (1991 to 1995, 6.80 kg; 1995 to 1999, 4.99 kg) and lowest for the opposite change in pattern scores (1991 to 1995, 0.87 kg; 1995 to 1999, −0.64 kg; p < 0.001) (Figures 1 and 2).

image

Figure 1. Mean weight change (kilograms) between 1991 and 1995 according to joint classifications in prudent and Western pattern scores (low and high pattern scores were defined as lower two quintiles and upper two quintiles), adjusted for baseline age, alcohol intake, physical activity, smoking, postmenopausal hormone use, oral contraceptive use, cereal fiber intake, total fat intake, and BMI and changes in these confounders (except BMI) and change in intake of sugar-sweetened soft drinks and diet soft drinks between time-points. Sample sizes were: A, n = 520; B, n = 61; C, n = 62; and D, n = 504.

Download figure to PowerPoint

image

Figure 2. Mean weight change (kilograms) between 1995 and 1999 according to joint classifications in prudent and Western pattern scores (low and high pattern scores were defined as lower two quintiles and upper two quintiles), adjusted for baseline age, alcohol intake, physical activity, smoking, postmenopausal hormone use, oral contraceptive use, cereal fiber intake, total fat intake, and BMI and changes in these confounders (except BMI and activity) and change in intake of sugar-sweetened soft drinks and diet soft drinks between time-points. Sample sizes were: A, n = 418; B, n = 78; C, n = 59; and D, n = 465.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Our results suggest that a dietary pattern including high intakes of red and processed meats, refined grains, sweets and desserts, and potatoes may contribute to long-term weight gain, whereas a dietary pattern characterized by high intakes of fruits, vegetables, whole grains, fish, poultry, and salad dressing may facilitate weight maintenance. We observed no difference in weight change between women with consistently low or high prudent pattern score and relatively small differences between women with consistently low or high Western pattern score. The lower weight gain associated with reduction in Western pattern score but increase in prudent pattern score compared with women with stable patterns suggests that women do benefit from adopting a prudent dietary pattern but that a new steady state is achieved, and weight trajectories do not continue to diverge with time.

Higher intakes of meat and sweets were predictive of weight gain over a 2-year follow-up period among men and women in the European Prospective Investigation into Cancer and Nutrition-Potsdam study (18). Ten-year changes in BMI were positively associated with meat consumption and negatively with vegetable consumption (19). A self-reported decrease in the consumption of fatty food or an increase in consumption of fruit predicted a lower increase in body weight and adiposity among 248 participants of the Quebec Family Study (20). In the Finnish Diabetes Prevention Study (21), subjects in an intensive lifestyle intervention group were advised to increase physical activity and to reduce calorie and total fat intake by frequent ingestion of whole-grain products, vegetables, fruits, low-fat milk and meat products, soft margarines, and vegetable oils rich in monounsaturated fatty acids. Weight loss over 2 years was significantly larger (3.5 kg) in the intervention group compared with a control group (0.8 kg), which received general nutrition and lifestyle advice. Also, the prudent dietary pattern was similar to the Mediterranean-type diet that reduced weight over 18 months in a randomized trial (22).

Relatively few studies have examined associations between overall dietary patterns and prospective weight change. Results from the Women's Health Initiative suggest that a low-fat diet, high in fruits, vegetables, and grain products, results in an only modest lower weight gain (0.4 kg) over 7.5 years compared with a regular U.S. diet (23). However, differences in fruit and vegetable consumption and fiber intake between the intervention and control group were only modest. In the Baltimore Longitudinal Study of Aging, a dietary pattern derived by factor analysis sharing components of both the prudent and the Western pattern of our study (low in white bread, refined grains, processed meats, potatoes, meat, and sugar-sweetened soft drinks and high in low-fat dairy products, cereal, fruit, fruit juice, non-white bread, nuts and seeds, and legumes) was inversely associated with annual change in BMI and waist circumference among 219 women and inversely with change in waist circumference but not BMI among 240 men (6). In the same cohort, the largest annual increase in BMI was observed for a cluster of subjects with a diet high in potatoes and meat, whereas it was smallest for a healthy diet cluster showing high contributions to energy intake from fruits, cereals, low-fat dairy products, and low contributions from fast food, sugar-sweetened soft drinks, and salty snacks (5). In the Framingham Offspring/Spouse cohort, women who ate an empty calorie diet that was rich in sweets and fats but low in fruits, vegetables, and lean food choices were at higher risk for developing overweight compared with women who ate a lower fat heart healthy diet, although this association gained no statistical significance (24). In the European Prospective Investigation into Cancer and Nutrition-Potsdam Study, a high-fiber, high-carbohydrate, low-fat dietary pattern, characterized by a high consumption of whole grain bread, fruits, fruit juices, grain flakes/muesli, and raw vegetables and by a low consumption of processed meat, butter, high-fat cheese, margarine, and red meat, was associated with lower weight gain over 4 years in non-obese individuals but not among obese (25). Togo et al. (26) evaluated the association between baseline dietary patterns and changes in pattern scores and prospective weight change in the Danish part of the MONICA Study. Although a pattern characterized by high intakes of meat and processed meats, potatoes, white bread, butter, and eggs (also included sweets, cakes, soft drinks or ice cream, and jam in women) was inversely associated with weight gain, this association appeared not to be independent of other dietary patterns, previous weight change, and other covariates. In a study among 1143 men, a healthy dietary pattern (defined as high intake of fruit and vegetables, use of margarine instead of butter, and preference for low-fat dairy products instead of high-fat) was not significantly associated with 10-year weight change (27). Using a step-wise selection procedure, Drapeau et al. (20) observed that decreasing fat consumption and increasing fruit consumption were associated with lower weight gain over a mean follow-up of 6 years. More recently, Sanchez-Villegas et al. (28) evaluated whether adherence to the Mediterranean diet is related to weight gain in the Sun cohort study. Although those with a low Mediterranean diet score had the largest weight gain over 28 months, this association was not independent of other lifestyle characteristics. The relatively large proportion of negative findings among these longitudinal studies support our observation that there is no difference in weight gain between those who maintain Western or prudent patterns. Other studies evaluating associations between dietary patterns and anthropometric characteristics used a cross-sectional study design. Overall, they provide inconclusive results (29), possibly because of the variability of dietary patterns derived, the variability of dietary assessment methods applied, and the potential for uncontrolled confounding. Because cross-sectional analyses on diet and body weight are error prone due to the potential for reverse causation, they have limited utility in testing the association between dietary pattern and body weight.

Women who decreased their prudent pattern score and increased their Western pattern score had the highest weight gain in our study. Such a dietary change represents changes in several food groups and nutrients, but it remains unclear which components or nutritive factors might mediate the association between the observed dietary pattern and weight change. A similar combination of food groups was observed as one single dietary pattern in the Baltimore Longitudinal Study of Aging based on the contribution of food items to total energy intake (6). Thus, such a pattern may facilitate larger weight gain particularly because of its energy-dense components. Other characteristics may also be important, e.g., the higher glycemic index (30, 31, 32) or the lower fiber content (33, 34, 35).

One major limitation of our study is its observational design. In particular, dietary patterns may strongly interact with other lifestyle characteristics or be part of specific lifestyles as previously suggested (36). Residual confounding by other dietary and lifestyle factors may, therefore, account for the association observed. Also, we cannot determine with certainty that changes in dietary patterns preceded changes in body weight. For example, an increase in prudent pattern score may reflect attempted weight loss among people who recently gained weight, rather than that an increase in prudent pattern score prevented weight gain. A further limitation of our study is the reliance on self-reported body weight. It is possible that under-reporting of body weight, particularly among heavier women, may have led to an underestimation of weight gain. However, correlation between self-reported and technician-measured body weight was found to be high in a similar cohort of older female nurses (14), and under-reporting may be less prevalent among relatively young women (37). Imprecise dietary measurement could potentially have influenced our observed associations. However, random errors in dietary assessment measures might have accounted for a lack of association but not the reverse (38). Furthermore, data from a similar cohort among male health professionals indicate reasonable reproducibility of the major dietary patterns defined by factor analysis (39). The reliability correlations for the factor scores between two FFQs administered 1 year apart were 0.70 for the prudent pattern and 0.67 for the Western pattern. Because correction for measurement error would tend to strengthen associations between changes in dietary exposures and weight change (40), our results likely underestimate the effects of dietary patterns on weight change. Still, error in self-reports of body weight might systematically relate to dietary patterns or changes in patterns, and we cannot rule out that this might have biased our observation.

In conclusion, our findings suggest that a dietary pattern characterized by frequent consumption of red and processed meats, refined grains, sweets and desserts, French fries, and potatoes may be associated with larger weight gain but that increased consumption of fruits, vegetables, whole grains, fish, poultry, legumes, and oil and vinegar salad dressing may be associated with less weight gain.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

This study was funded by NIH Research Grant CA50385. M.B.S. was also supported by a grant from Deutsche Krebshilfe. F.B.H. is the recipient of an American Heart Association Established Investigator Award. The funding organizations had no role in the design and conduct of the study, in the collection, analysis, and interpretation of the data, and in the preparation, review, or approval of the manuscript.

Footnotes
  • 1

    Nonstandard abbreviation: FFQ, food frequency questionnaire.

  • The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  • 1
    Mokdad, A. H., Bowman, B. A., Ford, E. S., Vinicor, F., Marks, J. S., Koplan, JP (2001) The continuing epidemics of obesity and diabetes in the United States. JAMA 286: 11951200.
  • 2
    Mokdad, A. H., Ford, E. S., Bowman, B. A., et al (2003) Prevalence of obesity, diabetes, and obesity-related health risk factors, 2001. JAMA 289: 7679.
  • 3
    Hedley, A. A., Ogden, C. L., Johnson, C. L., Carroll, M. D., Curtin, L. R., Flegal, KM (2004) Prevalence of overweight and obesity among US children, adolescents, and adults, 1999–2002. JAMA 291: 28472850.
  • 4
    Astrup, A., Larsen, Meinert T., Harper, A. (2004) Atkins and other low-carbohydrate diets: hoax or an effective tool for weight loss? Lancet 364: 897899.
  • 5
    Newby, P. K., Muller, D., Hallfrisch, J., Qiao, N., Andres, R., Tucker, KL (2003) Dietary patterns and changes in body mass index and waist circumference in adults. Am J Clin Nutr. 77: 14171425.
  • 6
    Newby, P. K., Muller, D., Hallfrisch, J., Andres, R., Tucker, KL (2004) Food patterns measured by factor analysis and anthropometric changes in adults. Am J Clin Nutr. 80: 504513.
  • 7
    Fung, T. T., Schulze, M., Manson, J. E., Willett, W. C., Hu, FB (2004) Dietary patterns, meat intake, and the risk of type 2 diabetes in women. Arch Intern Med. 164: 22352240.
  • 8
    van Dam, R. M., Rimm, E. B., Willett, W. C., Stampfer, M. J., Hu, FB (2002) Dietary patterns and risk for type 2 diabetes mellitus in U.S. men. Ann Intern Med. 136: 201209.
  • 9
    U.S. Department of Agriculture (1992) Composition of Foods: Raw, Processed, Prepared, 1963–1991 U.S. Government Printing Office Washington, DC.
  • 10
    Willett, W., Stampfer, MJ (1986) Total energy intake: implications for epidemiologic analyses. Am J Epidemiol. 124: 1727.
  • 11
    Willett, W. C., Lenart, E. (1998) Reproducibility and validity of food-frequency questionnaires. Willett, WC eds. Nutritional Epidemiology 101156. Oxford University Press New York.
  • 12
    Feskanich, D., Rimm, E. B., Giovannucci, E. L., et al (1993) Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire. J Am Diet Assoc. 93: 790796.
  • 13
    Salvini, S., Hunter, D. J., Sampson, L., et al (1989) Food-based validation of a dietary questionnaire: the effects of week-to-week variation in food consumption. Int J Epidemiol. 18: 858867.
  • 14
    Willett, W., Stampfer, M. J., Bain, C., et al (1983) Cigarette smoking, relative weight, and menopause. Am J Epidemiol. 117: 651658.
  • 15
    Wolf, A. M., Hunter, D. J., Colditz, G. A., et al (1994) Reproducibility and validity of a self-administered physical activity questionnaire. Int J Epidemiol. 23: 991999.
  • 16
    Hatcher, L. (1994) A Step-by-Step Approach to Using the SAS System for Factor Analysis and Structural Equation Modeling. Cary, NC: SAS Institute Inc.
  • 17
    Schulze, M. B., Manson, J. E., Ludwig, D. S., et al (2004) Sugar-sweetened beverages, weight gain, and incidence of type 2 diabetes in young and middle-aged women. JAMA 292: 927934.
  • 18
    Schulz, M., Kroke, A., Liese, A. D., Hoffmann, K., Bergmann, M. M., Boeing, H. (2002) Food groups as predictors for short-term weight changes in men and women of the EPIC-Potsdam cohort. J Nutr. 132: 13351340.
  • 19
    Kahn, H. S., Tatham, L. M., Rodriguez, C., Calle, E. E., Thun, M. J., Heath, C. W., Jr (1997) Stable behaviors associated with adults’ 10-year change in body mass index and likelihood of gain at the waist. Am J Public Health 87: 747754.
  • 20
    Drapeau, V., Despres, J. P., Bouchard, C., et al (2004) Modifications in food-group consumption are related to long-term body-weight changes. Am J Clin Nutr. 80: 2937.
  • 21
    Tuomilehto, J., Lindstrom, J., Eriksson, J. G., et al (2001) Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N Engl J Med. 344: 13431350.
  • 22
    McManus, K., Antinoro, L., Sacks, F. (2001) A randomized controlled trial of a moderate-fat, low-energy diet compared with a low fat, low-energy diet for weight loss in overweight adults. Int J Obes Relat Metab Disord. 25: 15031511.
  • 23
    Howard, B. V., Manson, J. E., Stefanick, M. L., et al (2006) Low-fat dietary pattern and weight change over 7 years: the Women's Health Initiative Dietary Modification Trial. JAMA 295: 3949.
  • 24
    Quatromoni, P. A., Copenhafer, D. L., D'Agostino, R. B., Millen, BE (2002) Dietary patterns predict the development of overweight in women: The Framingham Nutrition Studies. J Am Diet Assoc. 102: 12391246.
  • 25
    Schulz, M., Nothlings, U., Hoffmann, K., Bergmann, M. M., Boeing, H. (2005) Identification of a food pattern characterized by high-fiber and low-fat food choices associated with low prospective weight change in the EPIC-Potsdam cohort. J Nutr. 135: 11831189.
  • 26
    Togo, P., Osler, M., Sorensen, T. I., Heitmann, BL (2004) A longitudinal study of food intake patterns and obesity in adult Danish men and women. Int J Obes Relat Metab Disord. 28: 583593.
  • 27
    Fogelholm, M., Kujala, U., Kaprio, J., Sarna, S. (2000) Predictors of weight change in middle-aged and old men. Obes Res. 8: 367373.
  • 28
    Sanchez-Villegas, A., Bes-Rastrollo, M., Martinez-Gonzalez, M. A., Serra-Majem, L. (2006) Adherence to a Mediterranean dietary pattern and weight gain in a follow-up study: the SUN cohort. Int J Obes (Lond) 30: 350358.
  • 29
    Togo, P., Osler, M., Sorensen, T. I., Heitmann, BL (2001) Food intake patterns and body mass index in observational studies. Int J Obes Relat Metab Disord. 25: 17411751.
  • 30
    Brand-Miller, J. C., Holt, S. H., Pawlak, D. B., McMillan, J. (2002) Glycemic index and obesity. Am J Clin Nutr. 76: 281S285S.
  • 31
    Ludwig, DS (2003) Dietary glycemic index and the regulation of body weight. Lipids 38: 117121.
  • 32
    Roberts, SB (2003) Glycemic index and satiety. Nutr Clin Care 6: 2026.
  • 33
    Howarth, N. C., Saltzman, E., Roberts, SB (2001) Dietary fiber and weight regulation. Nutr Rev. 59: 129139.
  • 34
    Pereira, M. A., Ludwig, DS (2001) Dietary fiber and body-weight regulation: observations and mechanisms. Pediatr Clin North Am. 48: 969980.
  • 35
    Koh-Banerjee, P., Rimm, EB (2003) Whole grain consumption and weight gain: a review of the epidemiological evidence, potential mechanisms and opportunities for future research. Proc Nutr Soc. 62: 2529.
  • 36
    Martinez, M. E., Marshall, J. R., Sechrest, L. (1998) Invited commentary: factor analysis and the search for objectivity. Am J Epidemiol. 148: 1719.
  • 37
    Kuczmarski, M. F., Kuczmarski, R. J., Najjar, M. (2001) Effects of age on validity of self-reported height, weight, and body mass index: findings from the Third National Health and Nutrition Examination Survey, 1988–1994. J Am Diet Assoc. 101: 2834. quiz 35–36.
  • 38
    Hu, F. B., Stampfer, M. J., Rimm, E., et al (1999) Dietary fat and coronary heart disease: a comparison of approaches for adjusting for total energy intake and modeling repeated dietary measurements. Am J Epidemiol. 149: 531540.
  • 39
    Hu, F. B., Rimm, E., Smith-Warner, S. A., et al (1999) Reproducibility and validity of dietary patterns assessed with a food-frequency questionnaire. Am J Clin Nutr. 69: 243249.
  • 40
    Liu, S., Willett, W. C., Manson, J. E., Hu, F. B., Rosner, B., Colditz, G. (2003) Relation between changes in intakes of dietary fiber and grain products and changes in weight and development of obesity among middle-aged women. Am J Clin Nutr. 78: 920927.