Social and Behavioral
Provision of Foods Differing in Energy Density Affects Long-Term Weight Loss Barbara J. Rolls, Liane S. Roe, Amanda M. Beach, and Penny M. Kris-Etherton
Abstract ROLLS, BARBARA J., LIANE S. ROE, AMANDA M. BEACH, AND PENNY M. KRIS-ETHERTON. Provision of foods differing in energy density affects long-term weight loss. Obes Res. 2005;13:1052–1060. Objective: The energy density (kilocalories per gram) of foods influences short-term energy intake. This 1-year clinical trial tested the effect on weight loss of a diet incorporating one or two servings per day of foods equal in energy but differing in energy density. Research Methods and Procedures: Dietitians instructed 200 overweight and obese women and men to follow an exchange-based energy-restricted diet. Additionally, subjects were randomized to consume daily either one or two servings of low energy-dense soup, two servings of high energy-dense snack foods, or no special food (comparison group). Results: All four groups showed significant weight loss at 6 months that was well maintained at 12 months. The magnitude of weight loss, however, differed by group (p ⫽ 0.006). At 1 year, weight loss in the comparison (8.1 ⫾ 1.1 kg) and two-soup (7.2 ⫾ 0.9 kg) groups was significantly greater than that in the two-snack group (4.8 ⫾ 0.7 kg); weight loss in the one-soup group (6.1 ⫾ 1.1 kg) did not differ significantly from other groups. Weight loss was significantly correlated with the decrease in dietary energy density from baseline at 1 and 2 months (p ⫽ 0.0001) but not at 6 and 12 months. Discussion: On an energy-restricted diet, consuming two servings of low energy-dense soup daily led to 50% greater weight loss than consuming the same amount of energy as high energy-dense snack food. Regularly consuming foods
Received for review July 29, 2004. Accepted in final form March 29, 2005. 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. Department of Nutritional Sciences, The Pennsylvania State University, University Park, Pennsylvania. Address correspondence to Barbara J. Rolls, The Pennsylvania State University, 226 Henderson Building, University Park, PA 16802. E-mail:
[email protected] Copyright © 2005 NAASO
1052
OBESITY RESEARCH Vol. 13 No. 6 June 2005
that are low in energy density can be an effective strategy for weight management.
Introduction It is important to develop practical dietary strategies that are effective for long-term weight loss. Modifications of energy-restricted diets that enhance satiety may improve long-term adherence by helping individuals to feel more satisfied while they are limiting their food intake. Recent research demonstrates that decreasing the energy density (kilocalories per gram) of food reduces short-term energy intake and that this effect is related to increased satiety induced by the greater volume of food (1,2). The long-term effectiveness of this strategy in reducing intake and promoting weight loss, however, is not known. The objective of this randomized, parallel-group trial was to test the effect on weight loss of incorporating one or two servings daily of foods that differed in energy density. Experimental studies have shown that eating a low energy-dense soup or salad as a first course increases satiety and reduces energy intake at a single meal (2,3). Soup, in particular, has been found to increase satiety at a meal to a greater extent than other foods (4,5). In addition to the single-meal experiments, researchers have found that decreasing the energy density of multiple meals results in decreased energy intake over 1 to 2 days (6,7). Some longer term studies suggest that weight loss is improved by incorporating soup in an energy-restricted diet. In one trial of 517 subjects, one-half of whom were instructed to eat soup for lunch at least four times a week (8), there was a positive correlation between the frequency of soup consumption and weight loss after 10 weeks. In another trial (9), subjects who were instructed to follow an energy-restricted diet that included one to two servings of soup daily showed better maintenance of weight loss after 1 year than those instructed in a similar diet without the incorporation of soup. In the present trial, we tested the effect of an energyrestricted diet that incorporated one or two isocaloric servings daily of either low energy-dense soups or high energydense snack foods. Although the prescribed diets were equal in energy content, we hypothesized that the greater volume
Low Energy-Dense Food and Weight Loss, Rolls et al.
of food in the diet that included soup would enhance satiety and subject compliance and that this would lead to greater weight loss at 1 year than the diet that included snacks.
Research Methods and Procedures Experimental Design In this 1-year parallel-group trial, 200 overweight and obese women and men were instructed by dietitians to follow an energy-restricted diet; in addition, subjects were randomized to one of four intervention groups. Subjects in three of the groups were given supplies of commercially available food that was low in energy (100 kcal/serving) and fat (⬍4 g/serving). Subjects in these groups were instructed to consume a specified amount of the provided food daily during the study: one serving of soup (one-soup group), two servings of soup (two-soup group), or two servings of dry snack foods (two-snack group). Subjects in the fourth group were not provided with any specific food to consume (comparison group). The main evaluation of the trial was the contrast between the two-soup group and the two-snack group; subjects in these groups were instructed to consume the same number of daily servings of the provided foods, which were isocaloric but differed in energy density. A contrast between the one- and two-soup groups enabled us to test whether the amount of soup that was provided had an effect on weight loss. The comparison group allowed evaluation of the effects of the energy-restricted diet alone, without any food provision. The clinical trial consisted of a 6-month weight loss phase and a 6-month weight maintenance phase; the main outcomes were weight loss and change in diet composition at the end of 1 year. Participants provided written informed consent and were financially compensated for their participation. The protocol was approved by the Office for Research Protections of The Pennsylvania State University, and the study was conducted at the General Clinical Research Center located at the University Park campus of the university. Subject Recruitment Potential subjects were recruited by flyers and newspaper advertisements, and their eligibility for the trial was determined by a telephone interview, a physical screening session, and questionnaires that assessed symptoms of depression (10), symptoms of eating disorders (11), and ability to safely engage in physical activity (12). Subjects were required to be 20 to 65 years old and have a BMI of 26 to 40 kg/m2. Potential subjects were excluded if they had blood pressure ⬎ 140/90 mm Hg, serum triglycerides ⬎ 400 mg/ dl, or total cholesterol ⬎ 90th percentile for their age and sex; had a serious medical condition that precluded participation or any condition limiting physical activity; were
pregnant or lactating; had symptoms indicative of depression or an eating disorder; followed a sodium-restricted diet; or were unwilling to regularly consume soup or savory snacks. Potential subjects who met the inclusion criteria were instructed in keeping a detailed record of their food and beverage intake for three nonconsecutive days. Individuals who completed a baseline 3-day diet record were assigned to an intervention group; a stratified randomization scheme was used to balance the distribution of subject sex and age across the groups. Subject Characteristics Two hundred subjects (154 women and 46 men) met the inclusion criteria and were enrolled in the trial (Table 1). There were no differences among intervention groups in baseline demographic, physiological, or psychological characteristics, but there were some differences in dietary intakes. At baseline, the two-soup group reported a greater mean daily intake of fruits and vegetables than the other three groups (6.9 vs. a mean of 5.5 servings/d; p ⫽ 0.024) and also a greater sodium intake (4101 vs. a mean of 3619 mg/d; p ⫽ 0.028). At the end of 12 months, 147 subjects (74%) remained in the trial; these subjects will be referred to as study completers. Of the 53 withdrawals from the trial, 12 (23%) occurred in the 1st month, and a total of 29 (55%) occurred in the first 3 months. The most frequent reasons given for withdrawal were inability to attend counseling sessions (due to schedule conflicts or relocating out of the area) and difficulty in complying with the requirements of the diet or the record-keeping. There were no significant differences in baseline characteristics, including BMI, between those who did and did not complete the study. Weight Loss Counseling All subjects were instructed by dietitians in a weight loss program (13,14) that included an energy-restricted low-fat diet, advice to increase physical activity, and strategies for behavior modification. At baseline, subjects were assigned a dietary energy level based on their estimated daily energy expenditure (15) minus 750 kcal/d, to facilitate weight loss at a rate of ⬃0.7 kg/wk during the 6-month weight loss phase. Subjects were given a meal plan with dietary exchanges that provided 55% of energy as carbohydrate, 30% as fat, and 15% as protein; the meal plan incorporated the provided study foods (soup or snacks) for the relevant groups. At the beginning of the 6-month weight maintenance phase, subjects were provided with a new dietary energy level, meal plan, and exchange lists that were appropriate for weight maintenance. During months 1 to 3, subjects participated in a weekly individual counseling session with a dietitian and completed 3-day diet records every 2 weeks; during months 4 to 6, they OBESITY RESEARCH Vol. 13 No. 6 June 2005
1053
Low Energy-Dense Food and Weight Loss, Rolls et al.
Table 1. Baseline characteristics (mean ⫾ SE) of the 200 randomized subjects Intervention group Characteristic
Two snacks (n ⴝ 50)
One soup (n ⴝ 50)
Two soups (n ⴝ 50)
Comparison (n ⴝ 50)
Age (years) Height (cm) Body weight (kg) BMI (kg/m2) Activity score* Energy expenditure (kcal/d)† Energy intake (kcal/d)‡ Energy goal for weight loss (kcal/d)
44.5 ⫾ 1.2 168.4 ⫾ 1.1 89.2 ⫾ 1.9 31.4 ⫾ 0.4 272 ⫾ 29 2073 ⫾ 39 1978 ⫾ 82 1323 ⫾ 39
45.1 ⫾ 1.2 167.4 ⫾ 1.4 86.5 ⫾ 1.9 30.9 ⫾ 0.5 367 ⫾ 42 2049 ⫾ 44 2209 ⫾ 82 1299 ⫾ 34
43.8 ⫾ 1.2 169.5 ⫾ 1.3 88.4 ⫾ 1.8 30.8 ⫾ 0.5 400 ⫾ 40 2046 ⫾ 35 2191 ⫾ 113 1296 ⫾ 35
45.2 ⫾ 1.2 167.8 ⫾ 1.1 88.1 ⫾ 1.7 31.3 ⫾ 0.4 356 ⫾ 44 2052 ⫾ 36 2056 ⫾ 94 1302 ⫾ 36
* From the Leisure-time Exercise Questionnaire (19). † Estimated from sex, age, and activity level (15). ‡ From 3 days of diet records for study completers only (n ⫽ 147).
had a counseling session every 2 weeks and completed 3-day diet records monthly. During the weight maintenance phase in months 7 to 12, subjects met with a dietitian monthly and completed a total of two 3-day diet records. Each subject attended a maximum of 27 counseling sessions that lasted ⬃15 to 30 minutes each. Study Foods Subjects in the soup groups could select from 18 kinds of commercial soups (six of which were also available in a low-sodium version) and were given a mug that held one serving of soup (300 grams; 10.5 fl. oz.). Subjects in the snacks group could choose from 18 varieties of commercial snacks (crackers, baked potato chips, baked tortilla chips, bagel chips, and pretzels) that were provided in packages containing a single serving (24 grams). The study foods had a similar energy content (mean 100 kcal/serving) and fat content (⬍4 g/serving) but differed in energy density (mean 0.35 kcal/gram for soups and 4.2 kcal/gram for snacks). Thus, although the study foods had the same energy content per serving, they had very different weight, volume, and water content per serving. In the dietary exchange system followed by the subjects, each serving of soup or snack replaced one grain exchange, so that the energy level of the assigned diet was not altered. Subjects in the soup and snack intervention groups were asked to complete daily records of their study food intake and were told that it was preferable to consume their servings of provided food just before their lunch and dinner meals. Subjects in all groups were told that the purpose of the trial was to test the inclusion of different foods in a weight loss diet and were not given any specific information related to energy density or other qualities of the soup and snacks. 1054
OBESITY RESEARCH Vol. 13 No. 6 June 2005
Data Collection Body weight was measured at each counseling session, with the subject wearing light clothing without shoes, using a scale that was regularly calibrated. Blood pressure was measured monthly using a standard method (16). Fasting (12-hour) blood specimens were collected at baseline and at 3, 6, and 12 months. Measurements of serum lipids and lipoproteins were performed at the Lipid Laboratory (Mary Imogene Basset Research Institute, Cooperstown, NY), which participates in the Center for Disease Control Lipid Standardization Program. For study completers, the 3-day diet records that were completed at baseline and 1, 2, 6, and 12 months were coded by dietitians who were not involved in the trial. The records were analyzed for nutrients and food group servings using U.S. Department of Agriculture food composition data (17) supplemented with data from commercial sources. Food energy density (kilocalories per gram) was calculated as the ratio between food energy and food weight, both excluding beverages (18); soup was categorized as a food in this calculation. Eating occasions were categorized either as main meals (breakfast, lunch, or dinner) or as between-meal snacks, based on the time of day and the type and amount of food eaten. At each counseling session, the Leisure-Time Exercise Questionnaire (19) was used to assess physical activity during the past week. The Diet Satisfaction Questionnaire, which assessed subjects’ satisfaction with their food plan using seven-point scales, was completed at baseline and 1, 3, 6, 9, and 12 months. The Eating Inventory (20), which measures dietary restraint (the tendency to consciously restrict food intake to control body weight), disinhibition (the loss of control over eating in response to emotional or social
Low Energy-Dense Food and Weight Loss, Rolls et al.
cues), and hunger (tendency to subjective feelings of hunger), was completed at baseline and at 6 and 12 months. The Food Preference questionnaire (21), which assesses preference for the taste of low- and high-fat versions of common foods, was completed at baseline and 3, 6, 9, and 12 months. Statistical Analyses Continuous outcomes were analyzed with a linear mixed model, which took into account the autoregressive nature of the repeated measures. Intervention group, subject gender, and study week were treated as fixed effects, and subjects were treated as random effects. Anthropometric and physiological outcomes were analyzed for subjects who completed the study and also for all randomized subjects (testing the least-squares means from the linear mixed model). Food and nutrient intake data from the diet records were available only for study completers. Multivariate ANOVA was used to test macronutrient intakes as a percentage of energy intake. Categorical outcomes from the questionnaires were analyzed using a 2 test; where there was no effect of study week, outcomes from all available study weeks were analyzed together. Step-wise regression analyses were performed to predict weight loss based on intervention group, subject characteristics, and dietary and activity measures. Differences were considered significant at p ⬍ 0.05, except for post hoc pair-wise comparisons between groups, for which a modified Bonferroni adjustment was used (p ⬍ 0.025).
Results Compliance Study completers attended 97% of their scheduled counseling sessions. For all randomized subjects, including those who withdrew from the trial, overall attendance was 82%. There was no difference in attendance by intervention group or subject gender. The 6- and 12-month diet records showed that, as planned, intake of soup and snacks differed significantly among the intervention groups (Table 2). Intake of soup in both soup groups declined at 12 months compared with at 1, 2, and 6 months (p ⬍ 0.024); intake of snacks in the snacks group did not change significantly over time. Weight Loss and Physical Activity For the 147 study completers, in all intervention groups combined, there was a mean loss of 7.6 ⫾ 0.4 kg (8.7% of initial weight) during the weight loss phase, followed by mean gain of 1.1 ⫾ 0.2 kg (1.3%) during the weight maintenance phase, for a net loss of 6.5 ⫾ 0.5 kg (7.4%) after 1 year (p ⬍ 0.0001). The magnitude of weight loss, however, differed by intervention group (Figure 1; p ⬍ 0.006). At 1 year, mean weight loss in the comparison and two-soup groups was significantly greater than that in the
two-snack group; weight loss in the one-soup group did not differ significantly from that in any other group (Table 2). There was no significant difference in weight loss between women and men. When all 200 randomized subjects were included in the statistical model, weight loss at 1 year (least-squares mean value) was 7.2 ⫾ 1.0 kg (8.2%) in the comparison group, 6.6 ⫾ 1.1 kg (7.2%) in the two-soup group, 4.3 ⫾ 1.0 kg (5.0%) in the two-snack group, and 5.4 ⫾ 0.9 kg (6.1%) in the one-soup group. The same group differences were significant as were significant in the analysis of study completers only. The physical activity score increased 57% in all intervention groups, from a mean of 350 ⫾ 20 at baseline to 549 ⫾ 34 at 12 months, but the magnitude of the increase did not differ among groups. Blood Pressure and Serum Lipids Subjects in all intervention groups showed significant decreases in both systolic and diastolic blood pressure after intervention. Mean blood pressure decreased from 116/77 mm Hg at baseline to 111/73 mm Hg at 6 months (p ⬍ 0.0001) and remained unchanged at 12 months. All intervention groups also showed a decrease in total and lowdensity lipoprotein-cholesterol and triglycerides and an increase in high-density lipoprotein-cholesterol at 12 months compared with baseline (Figure 2; p ⬍ 0.0001). The magnitude of the changes in blood pressure and serum lipids did not differ significantly among the intervention groups. Dietary Intakes Analysis of the diet records of study completers showed that reported energy intake decreased significantly from baseline to 6 months and increased slightly at 12 months (Table 2; p ⬍ 0.005). There were no significant differences in reported energy intake among the intervention groups at any time-points. The variability in energy intake was large, however; the SE of the difference between groups was ⬃100 kcal/d. At the end of the 6-month weight loss phase, the maximum difference in weight loss between groups corresponded to an estimated difference in daily energy intake of 120 kcal/d. Thus, there was insufficient statistical power to detect the differences in daily energy intake that would account for the differences in weight loss among the groups. As planned, during the intervention there were significant differences among groups in dietary energy density (Table 2; p ⬍ 0.0001). Food energy density decreased from baseline in all groups, but significantly more so in the soup groups than in the comparison and snacks groups. The energy density of individual meals also differed among the groups (p ⬍ 0.0001); at midmorning and midafternoon snack meals and lunch, energy density was greater in the snacks group than in the other three groups (Figure 3). Food intake by weight also differed among groups during interOBESITY RESEARCH Vol. 13 No. 6 June 2005
1055
1056
OBESITY RESEARCH Vol. 13 No. 6 June 2005
9.1 ⫾ 0.9ab
7.0 ⫾ 0.8a
NA
9.0 ⫾ 1.1b 5.5 ⫾ 0.8a
4.8 ⫾ 0.7a
Comparison Two snacks (n ⴝ 37) (n ⴝ 37)
8.4 ⫾ 0.9ab 10.2 ⫾ 1.2b
7.6 ⫾ 0.8ab
Two soups (n ⴝ 34)
6.9 ⫾ 1.2ab
6.1 ⫾ 1.1ab
One soup (n ⴝ 39)
7.9 ⫾ 1.0b
7.2 ⫾ 0.9b
Two soups (n ⴝ 34)
12 months
9.2 ⫾ 1.2b
8.1 ⫾ 1.1b
Comparison (n ⴝ 37)
4.7 ⫾ 0.3a 6.5 ⫾ 0.3a
5.8 ⫾ 0.2‡ 7.5 ⫾ 0.2
6.1 ⫾ 0.5a
6.1 ⫾ 0.5b 4.8 ⫾ 0.3b
6.4 ⫾ 0.3b
5.8 ⫾ 0.4ab
5.8 ⫾ 0.3ab
6.7 ⫾ 0.3
5.2 ⫾ 0.3a
5.8 ⫾ 0.5
5.9 ⫾ 0.5ab
6.2 ⫾ 0.5
6.8 ⫾ 0.4b
5.8 ⫾ 0.4
6.0 ⫾ 0.5ab
1.86 ⫾ 0.03 1.61 ⫾ 0.05a 1.23 ⫾ 0.05b 1.03 ⫾ 0.04c 1.40 ⫾ 0.05d 1.73 ⫾ 0.06a 1.32 ⫾ 0.06b 1.23 ⫾ 0.06b 1.54 ⫾ 0.05c
0.16 ⫾ 0.03 1.93 ⫾ 0.06a 0.20 ⫾ 0.06b 0.13 ⫾ 0.04b 0.20 ⫾ 0.05b 1.90 ⫾ 0.15a 0.26 ⫾ 0.07b 0.18 ⫾ 0.06b 0.13 ⫾ 0.05b 2110 ⫾ 47 1430 ⫾ 45 1557 ⫾ 82 1390 ⫾ 57 1538 ⫾ 90 1441 ⫾ 26 1496 ⫾ 42 1443 ⫾ 32 1496 ⫾ 42 1013 ⫾ 25 830 ⫾ 33a 1214 ⫾ 65b 1226 ⫾ 47b 1001 ⫾ 44c 859 ⫾ 30a 1125 ⫾ 55b 1244 ⫾ 67b 955 ⫾ 41a
0.10 ⫾ 0.02 0.25 ⫾ 0.06a 1.09 ⫾ 0.09b 1.72 ⫾ 0.09c 0.40 ⫾ 0.07a 0.11 ⫾ 0.04a 0.90 ⫾ 0.09b 1.33 ⫾ 0.12c 0.10 ⫾ 0.04a
7.9 ⫾ 0.9ab
6.1 ⫾ 0.7a
One soup (n ⴝ 39)
NA*
Two snacks (n ⴝ 37)
6 months
At a given time-point, means with different superscript letters are significantly different (modified Bonferroni comparison, p ⬍ 0.025). * NA, not applicable. † Energy density of food only, excluding beverages. ‡ Fruit and vegetable intake differed by group at baseline; subjects in the two-soups group had a greater intake than the other groups (p ⫽ 0.024).
Weight loss Weight loss (kg) Percentage weight loss (% initial weight) Dietary intakes Soup intake (servings/d) Snack food intake (servings/d) Total energy (kcal/d) Food weight (g/d) Dietary energy density (kcal/g)† Fruit and vegetable group (servings/d) Grain and cereal group (servings/d)
Baseline: all groups (n ⴝ 147)
Table 2. Mean (⫾SE) weight loss and reported dietary intakes for 147 study completers at major time-points
Low Energy-Dense Food and Weight Loss, Rolls et al.
Low Energy-Dense Food and Weight Loss, Rolls et al.
Figure 1: Mean weight loss (⫾SE) over time in 147 study completers. Significant differences among intervention groups at the end of the weight loss and weight maintenance phases are indicated by different letters (p ⬍ 0.025).
vention (p ⬍ 0.0001; Table 2); it increased from baseline in the soup groups, decreased from baseline in the snacks group, and did not change significantly from baseline in the comparison group. There was a small but significant difference in macronutrient content among groups; the snacks group had a higher intake of energy as carbohydrate (50 ⫾ 0.5%) than the other
Figure 2: Mean change in blood lipids (⫾SE) over time for 147 study completers. Because there were no significant differences among intervention groups, data are shown for all four groups combined.
Figure 3: Mean energy density of foods (⫾SE) at morning and afternoon snack meals and lunch over time in 147 study completers. Significant differences among intervention groups are indicated by different letters (p ⬍ 0.025).
three groups (mean 48 ⫾ 0.3%; p ⫽ 0.017). Intakes of fat (30 ⫾ 0.3%), protein (17.6 ⫾ 0.1%), and alcohol (1.3 ⫾ 0.1%) did not differ significantly among groups. Dietary fiber intakes during intervention did not differ among groups or from baseline values (data not shown). During intervention, the decrease in sodium intake from baseline values did not differ significantly among intervention groups; both at baseline and during intervention, absolute intake of sodium was higher in the two-soup group than in the other groups (p ⬍ 0.0001). An analysis of food group intakes showed differences among intervention groups in the fruits and vegetables group and the grain and cereals group (Table 2 shows data for months 6 and 12). During the study, subjects in the comparison group had a significant increase in fruit and vegetable intake compared with those in the two-soup group, who had a decrease from their high baseline intake (increase of 0.7 serving/d vs. decrease of 0.5 serving/d; p ⫽ 0.023). The resulting absolute intake of fruits and vegetables was significantly greater in the comparison group and both soup groups than in the snacks group (p ⫽ 0.005). Intake of very low energy-dense fruits and vegetables (⬍0.6 kcal/gram) increased more in the comparison group than in the snacks group (1.3 vs. 0.2 serving/d; p ⫽ 0.006). The snacks group had a greater intake of grain and cereal foods than did the other three groups (p ⫽ 0.007). Factors Predicting Weight Loss Percentage weight loss at 1 and 2 months was significantly correlated with the decrease in food energy density from baseline to these time-points (Figure 4; p ⫽ 0.0001). At 6 and 12 months, however, this correlation was not OBESITY RESEARCH Vol. 13 No. 6 June 2005
1057
Low Energy-Dense Food and Weight Loss, Rolls et al.
Figure 4: Correlation between decrease in food energy density (from baseline value) and percentage weight loss in 147 study completers at 1 and 2 months (r ⫽ 0.36 and 0.33, respectively; p ⫽ 0.0001).
significant. Regression analysis showed that when all dietary and subject characteristics were considered, the strongest predictor of weight loss at 1 and 2 months was the decrease in food energy density from the baseline value, regardless of intervention group. The next strongest predictor at 1 and 2 months was the baseline score for disinhibition. Together, these factors accounted for 18% of the variability in weight loss at 1 month and 15% at 2 months. At 6 and 12 months, the decrease in the disinhibition score was the strongest predictor of weight loss, followed by the increase in the restraint score (p ⬍ 0.0001; R2 ⫽ 0.22), and none of the dietary variables was significant in the regression analysis. Questionnaire Responses On the Diet Satisfaction Questionnaire, during the study there were no differences among groups in the answers to the question “How healthy do you feel on this food plan?”; 83% of subjects answered “very healthy” or “extremely healthy.” There were, however, differences among groups in response to the question “How full do you feel after your meals while on this food plan?” (p ⬍ 0.003). A greater proportion of subjects in the two-soup and one-soup groups answered “very full” or “extremely full” (55% and 47%, respectively) than in the snacks (40%) and comparison (36%) groups. The responses to the question “How hungry do you feel while on this food plan?” also differed by group (p ⬍ 0.029); fewer subjects in the snacks group answered “not at all hungry” or “somewhat hungry” (59%) than in the other three groups (mean 70%). Scores on the Eating Inventory showed that all groups had a decrease in disinhibition (from 9.9 ⫾ 0.2 to 8.0 ⫾ 0.3) and tendency toward hunger (from 6.6 ⫾ 0.2 to 4.5 ⫾ 0.3) and an increase in dietary restraint (from 8.6 ⫾ 0.3 to 13.9 ⫾ 0.3) between baseline and 1 year (p ⬍ 0.0001). On the Food Preference Questionnaire, subjects reported that 1058
OBESITY RESEARCH Vol. 13 No. 6 June 2005
the high-fat choice tasted better for 63 ⫾ 1% of food items at baseline; this percentage decreased significantly to 56 ⫾ 1% at 3 months (p ⬍ 0.0001) and remained at this level for the duration of the trial. There were no significant differences among intervention groups in any of these scores.
Discussion This randomized trial of modification of an energy-restricted diet resulted in substantial weight loss in all intervention groups after 6 months, which was well-maintained for a further 6 months and resulted in improvements in blood pressure and blood lipids. The magnitude of weight loss, however, varied among the groups. The main evaluation of the trial demonstrated that consuming two servings of soup daily led to greater weight loss than consuming the same amount of energy as two servings of a dry snack food daily; after 1 year, weight loss in the two-soup group was 50% greater than that in the snacks group (7.2 vs. 4.8 kg). Thus, substituting a type of food that was low in energy density for one that was high in energy density increased the magnitude of the weight loss. These results extend the findings of single-meal experiments showing that consuming a food low in energy density increases satiety and decreases energy intake compared with consuming a food high in energy density (2,3). Subjects in all intervention groups were instructed to follow diets that had a similar energy restriction, and analysis of the diet records showed that reported energy intakes did not differ among the groups. Although we expected differences in satiety to affect compliance with the diet, we were unable to demonstrate differences in reported energy intake that would account for the differences in weight loss among the groups. The most likely reason for this finding is that the diet records were insufficiently precise to measure the differences in daily energy intake that were involved (⬍120 kcal/d).
Low Energy-Dense Food and Weight Loss, Rolls et al.
Our hypothesis, that changes in daily energy intake were caused by differences in satiety attributable to the dietary modifications, remains feasible. The variations in reported food energy density and food weight that resulted from changing only two servings of food per day were striking: at 6 months, food energy density was 1.6 kcal/d in the twosnack group and 1.0 kcal/d in the two-soup group. The responses to the Diet Satisfaction questionnaire provided evidence that satiety played a role in the differential weight loss among groups; a greater percentage of subjects in the soup groups than in the snacks group reported feeling very full and not very hungry during the intervention. The regression analysis showed that the decrease in food energy density was the most significant predictor of weight loss at 1 and 2 months, when subjects were the most motivated and compliant and achieved most of their weight loss. The diet followed by subjects in the comparison group, who were not provided with any special food, was a standard one used for treating obesity (13,14). This exchangebased diet led to substantial weight loss and healthier eating patterns, including low energy-dense snack meals and an increase in fruit and vegetable intake. The incorporation of two daily servings of soup into this diet led to similar weight loss and healthy eating patterns and an additional decrease in energy density and an increase in the weight of food consumed. The changes in energy density and food weight might be expected to further increase satiety and decrease energy intake, but although we observed differences in reported fullness between the comparison and two-soup groups, there were no differences in weight loss. Because subjects in the comparison group were not required to consume a special food and record this intake each day, it is possible that they were able to concentrate on the dietary advice with fewer distractions. It is also noteworthy that both the comparison and two soups groups had a similar low-energy density of snack meals. Research has suggested that snacks may be particularly problematic for weight management (22,23); it may be that the energy density of snack meals has an influence on satiety that is disproportionate to the amount of energy that they provide. Although the energy-restricted diet led to substantial weight loss, incorporating two servings daily of snack foods that were low in energy but high in energy density resulted in less weight loss and less healthy eating patterns. Subjects in the snacks group, unlike the other groups, experienced a decrease in food weight (mean 17%) from their baseline amount. It is likely that this decreased food weight was responsible for their reduced feelings of satiety during the intervention, which, in turn, attenuated their weight loss. In incorporating the snack foods in the diet, subjects in this group displaced fruit and vegetable servings (particularly those very low in energy density), rather than grain servings, as instructed. The resulting decrease in fruit and vegetable intake may have had an additional detrimental effect
on the amount of weight loss. A recent review concluded that fruits and vegetables have a measurable beneficial effect on satiety and weight loss when included in an energy-restricted diet (24). The diet records showed that reported soup intake decreased in the soup groups during the maintenance phase of the intervention, whereas snack intake remained stable in the snack group. This difference in compliance with the diet over time may be related to the fact that the snacks, which required no preparation, were more convenient to eat than the soups. In addition, although subjects could choose from 18 varieties of both soups and snacks, the provided food was always of the same general type. A wider range of food choices would be likely to help alleviate monotony effects associated with a single type of food and improve compliance to the diet and sustainability in the longer term. Education and training in reducing the energy density of the diet is a more practical general strategy than provision of low energy-dense food. A recent study showed that subjects taught to lower the energy density of their diet in addition to reducing the fat content had greater weight loss at 6 months than those taught to reduce the fat content alone (25). Further research is needed on the effects on weight loss of reducing the energy density of the diet, both by manipulating multiple foods and by nutrition education. Weight loss in this trial was not significantly affected by subject characteristics such as sex, age, and initial body weight, but it was predicted by scores from the Eating Inventory. The decrease in the disinhibition score and the increase in the dietary restraint score predicted the amount of weight loss at 6 and 12 months, confirming results of a previous clinical trial (26). Changes in both disinhibition and dietary restraint can be predictive of long-term weight loss, and these characteristics should be assessed in weight loss trials and addressed in weight loss programs. In this trial, an energy-restricted diet based on food exchanges led to significant weight loss that was well-maintained for 6 months. The energy density of two types of food incorporated in this diet significantly affected both the energy density of the total diet and the magnitude of weight loss. In the first 2 months of the trial, when most of the weight loss was achieved, the change in dietary energy density was the strongest predictor of the change in body weight. These findings provide support for regular consumption of food low in energy density, such as soup, as a strategy for weight management.
Acknowledgments This research was supported by Campbell Soup Co. (Camden, NJ) and by the General Clinical Research Center of The Pennsylvania State University (NIH Grant M01 RR 10732). OBESITY RESEARCH Vol. 13 No. 6 June 2005
1059
Low Energy-Dense Food and Weight Loss, Rolls et al.
References 1. Rolls BJ, Castellanos VH, Halford JC, et al. Volume of food consumed affects satiety in men. Am J Clin Nutr. 1998; 67:1170 –7. 2. Rolls BJ, Bell EA, Thorwart ML. Water incorporated into a food but not served with a food decreases energy intake in lean women. Am J Clin Nutr. 1999;70:448 –55. 3. Rolls BJ, Roe LS, Meengs JS. Salad and satiety: energy density and portion size of a first-course salad affect energy intake at lunch. J Am Diet Assoc. 2004;104:1570 – 6. 4. Himaya A, Louis-Sylvestre J. The effect of soup on satiation. Appetite. 1998;30:199 –210. 5. Rolls BJ, Fedoroff IC, Guthrie JF, Laster LJ. Foods with different satiating effects in humans. Appetite. 1990;15:115– 26. 6. Bell EA, Castellanos VH, Pelkman CL, Thorwart ML, Rolls BJ. Energy density of foods affects energy intake in normal-weight women. Am J Clin Nutr. 1998;67:412–20. 7. Bell EA, Rolls BJ. Energy density of foods affects energy intake across multiple levels of fat content in lean and obese women. Am J Clin Nutr. 2001;73:1010 – 8. 8. Jordan HA, Levitz LS, Utgoff KL, Lee HL. Role of food characteristics in behavioral change and weight loss. J Am Diet Assoc. 1981;79:24 –9. 9. Foreyt JP, Reeves RS, Darnell LS, Wohlleb JC, Gotto AM. Soup consumption as a behavioral weight loss strategy. J Am Diet Assoc. 1986;86:524 – 6. 10. Zung WWK. A self-rating depression scale. Arch Gen Psychiatry. 1965;12:63–70. 11. Garner DM, Garfinkel PE. The Eating Attitudes Test: an index of the symptoms of anorexia nervosa. Psychol Med. 1979;9:273– 80. 12. Thomas S, Reading J, Shephard RJ. Revision of the Physical Activity Readiness Questionnaire (PAR-Q). Can J Sport Sci. 1992;17:338 – 45. 13. National Institutes of Health. The Practical Guide: Identification, Evaluation, and Treatment of Overweight and Obesity in Adults: NIH Publication No. 02-4084. Bethesda, MD: Department of Health and Human Services, National Institutes of Health, National Heart, Lung, and Blood Institute; 2002. 14. American Dietetic Association and American Diabetes Association. Exchange Lists for Weight Management. 2nd ed. Chicago, IL: American Dietetic Association; 1995.
1060
OBESITY RESEARCH Vol. 13 No. 6 June 2005
15. St. Jeor ST, Stumbo PJ. Energy needs and weight maintenance in controlled feeding studies. In: Dennis BH, Ershow AG, Obarzanek E, Clevidence BA, eds. Well-Controlled Diet Studies in Humans. Chicago, IL: American Dietetic Association; 1997, p. 258. 16. Joint National Committee on Prevention Detection, Evaluation, and Treatment of High Blood Pressure. The sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Arch Intern Med. 1997;157:2413– 46. 17. U.S. Department of Agriculture, Agricultural Research Service. USDA National Nutrient Database for Standard Reference, Release 16-1. http://www.nal.usda.gov/fnic/foodcomp (accessed June 6, 2004). 18. Ledikwe JH, Blanck HM, Kettel Khan L, et al. Dietary energy density determined by eight calculation methods in a nationally representative United States population. J Nutr. 2005;135:273– 8. 19. Godin G, Shephard RJ. A simple method to assess exercise behavior in the community. Can J Appl Sport Sci. 1985;10: 141– 6. 20. Stunkard AJ, Messick S. The three-factor eating questionnaire to measure dietary restraint, disinhibition, and hunger. J Psychosom Res. 1985;29:71– 83. 21. Ledikwe JH, Pelkman CL, Roe LS, Mannino SL, Rolls BJ. Validation of a questionnaire to assess preference for fat. FASEB J. 2004;18:A117. 22. Marmonier C, Chapelot D, Fantino M, Louis-Sylvestre J. Snacks consumed in a nonhungry state have poor satiating efficiency: influence of snack composition on substrate utilization and hunger. Am J Clin Nutr. 2002;76:518 –28. 23. Marmonier C, Chapelot D, Louis-Sylvestre J. Metabolic and behavioral consequences of a snack consumed in a satiety state. Am J Clin Nutr. 1999;70:854 – 66. 24. Rolls BJ, Ello-Martin JA, Tohill BC. What can intervention studies tell us about the relationship between fruit and vegetable consumption and weight management? Nutr Rev. 2004; 62:1–17. 25. Ello-Martin JA, Roe L, Rolls B. A diet reduced in energy density results in greater weight loss than a diet reduced in fat. Obes Res. 2004;12:A23. 26. Foster GD, Wadden TA, Swain RM, Stunkard AJ, Platte P, Vogt RA. The Eating Inventory in obese women: clinical correlates and relationship to weight loss. Int J Obes Relat Metab Disord. 1998;22:778 – 85.