Physiology & Behavior, Vol. 48, pp. 19--26. ©Pergamon Press pie, 1990. Printed in the U.S.A.
0031-9384/90 $3.00 + .00
Effects of Drinks Sweetened With Sucrose or Aspartame on Hunger, Thirst and Food Intake in Men B A R B A R A J. R O L L S , 1 S I O N K I M A N D I N G R I D C. F E D O R O F F
Department o f Psychiatry and Behavioral Sciences The Johns Hopkins University School o f Medicine, Baltimore, M D 21205 R e c e i v e d 15 February 1990
ROLLS, B. J., S. KIM AND I. C. FEDOROFF. Effects of drinks sweetened with sucrose or aspartame on hunger, thirst and food intake in men. PHYSIOL BEHAV 48(1) 19-26, 1990.--Forty-two nondieting adult males were given 8 or 16 oz of lemonade, sweetened to equal intensity with either aspartame or sucrose, or the same volumes of water, or no drink. Subjects were separated into three groups receiving the drinks at different times: with a self-selection lunch, or 30, or 60 rain before lunch. Food intakes did not differ when subjects received the drinks with lunch; however, when the calories from the drinks were included, intake was significantly greater with the sucrose-sweetened lemonades than in the other conditions. When subjects received the drinks 30 or 60 min before lunch, food intakes were not significantly different. Appetite ratings were not different among the conditions. When the drinks were consumed with the meal, the 8-oz sucrose-sweetened lemonade differed from the other drinks in that it did not significantly reduce thirst. The results indicate that in nondieting males, aspartame in concentrations similar to those in commercially available drinks did not increase hunger ratings or food intake. However, caloric drinks taken with lunch increased total energy intake in that meal. Also, sucrose-sweetened drinks may decrease thirst less than water or aspartame-sweetened drinks when taken with a meal. Aspartame
Sucrose
Thirst
Hunger
Caloric compensation
Drinks
to five-year-old children, Birch et al. (3) found that fruit-flavored drinks sweetened with aspartame had either no effect or produced a significant suppression, not enhancement, of subsequent food intake compared with the same volume of water. Sucrose-sweetened drinks caused an even greater suppression so that there was compensation for the calories coming from sugar. Anderson et al. (1) studied older children (nine to ten years old) and found that aspartame-sweetened fruit drinks, with or without additional carbohydrate (polycose), did not affect hunger or lunchtime food intake when compared with drinks sweetened with sodium cyclamate or sucrose. Thus, there is no indication that aspartame increases appetite or food intake in children. We (15) conducted a study in adults which examined the effects of commercially available pudding and jello sweetened with either sucrose or aspartame on appetite ratings and food intake. The aspartame-sweetened and sucrose-sweetened foods had similar effects on ratings of hunger and satiety over the two hours after consumption that they were tracked. Despite the difference in caloric intake (up to 206 kcal), subjects showed only a nonsignificant trend towards caloric compensation when presented with a variety of foods two hours later. We reasoned that our failure to f'md increases in ratings of appetite could be because we used foods rather than solutions. Thus, in the present study, we compared two volumes of lemonade
ALTHOUGH intense sweeteners such as saccharin, cyclamate or aspartame have been used for a number of years to reduce the energy density of foods and drinks, studies have only recently begun to clarify how they affect food intake and appetite. When Blundell and Hill (4) suggested that aspartame might increase appetite and that such dietary aids " m a y contribute to disordered patterns of eating prevalent among certain groups of normal weight individuals," there was little data available to contradict this view. Porikos et al. (12) had found that the substitution of aspartame for sucrose over 12-day periods led to a 15% reduction in daily caloric intake. Additionally, Rolls et al. (15) had reported that consumption of jello sweetened with aspartame suppressed hunger to the same extent as did sucrosesweetened jello. Despite the fact that none of these studies had found that aspartame increases food intake, it became widely believed that intense sweeteners might be of no benefit in weight control and, indeed, that they might even lead to weight gain. Blundell and co-workers (13) have continued to report that aspartame and other intense sweeteners in solution increase appetite ratings, but have failed to find an increase in food intake following aspartame. Several other studies have failed to find that aspartame increases appetite ratings or food intake. Two studies have used familiar drinks in children of various ages to study the effects of intense sweeteners. In a study of two-
~Requests for reprints should be addressed to Dr. B. J. Rolls, Department of Psychiatry and Behavioral Sciences, Meyer 1-104, The Johns Hopkins University School of Medicine, 600 N. Wolfe St., Baltimore, MD 21205.
19
20
ROLLS, KIM AND FEDOROFF
sweetened with sucrose or aspartame, the same volumes of water, and a condition in which no drink was consumed. Since the time between consuming the drinks and time of food consumption could affect the results (3,5), we conducted three experiments, each with a different time relationship between the drinks and lunch. A unique feature of the present studies was the investigation of the effects of the different drinks when they were consumed not as a preload, but were taken with the meal. Although drinks are frequently consumed with meals, this design had not been previously employed. Another new aspect of the present research was to determine the effects of the differently sweetened drinks on ratings of thirst. There is no previous work comparing the effects of consumption of sucrose- and aspartame-sweetened drinks on thirst. GENERAL METHOD
Subjects Forty-two healthy male volunteers aged between 21 and 39 (mean age=25.0, S . D . = 4 . 3 ) were recruited for three experiments (14 in each experiment) which were run concurrently. The men were all nonsmoking, nondieting, of normal weight for height (between 90% and 110% of ideal body weight) according to the 1983 Metropolitan Life Weight Tables (mean=99.7%, S.D. = 8.3%), and had no food allergies or restrictions nor took any medication. Furthermore, the volunteers had been screened using the Stunkard Eating Inventory (18) which measures cognitive dietary restraint, the Eating Attitudes Test (10) which detects any symptoms of an eating disorder, and the Zung Self-Rating questionnaire (20) which measures depression. Only those men scoring less than 10 on the Eating Inventory (mean score = 5.2, S.D. = 2.9), less than 15 on the Eating Attitudes Test (mean score=7.0, S.D. =3.4) and less than 50 on the Zung (mean score=34.9, S.D. =4.3) were asked to participate. Finally, to ensure that the subjects liked the foods used in the experiment and would eat the foods, as well as to minimize a novelty effect, each recruited volunteer was required to fast complete a trial session in which he would rate the pleasantness of the taste of all the foods. Only those volunteers rating the taste of all the foods above 50 on 100-mm visual analog scales anchored by "extremely pleasant" on the right and "not at all pleasant" on the left were run in the experiment, When asked whether they always consumed a drink with their lunch, all subjects responded affirmatively. Finally, all subjects signed a consent form at the beginning of their first session which stated that the purpose of the study was to investigate the effects of different drinks (lemonade and water) on thirst and appetite ratings. Subjects, however, were naive as to the sweetener manipulation in the study.
General Procedure All three experiments utihzed a within-subjects, repeated measures design with each subject serving as his own control. Every subject received each of the following drinks: 8 oz sucrosesweetened lemonade, 16 oz sucrose-sweetened lemonade, 8 oz aspartarne-sweetened lemonade, 16 oz aspartame-sweetened lemonade, 8 oz water, and 16 oz water. On one occasion, no drink was given. Each subject was run once in each of the 7 conditions, and the order of presentation of conditions was counterbalanced across subjects. Subjects came to the laboratory at the same designated time for each of the seven sessions with at least three days between sessions. Upon arrival, all subjects presented food diaries on
TABLE 1 DRINK CHARACTERISTICS
Drink
Volume
Lemonade Lemonade Lemonade Lemonade Water Water
8 oz 16 oz 8 oz 16 oz 8 oz 16 oz
Sweetener (g)
Calories
Sucrose (20) Sucrose (40) Aspartame (0.11) Aspartame (0.22) None None
83 166 5 10 0 0
which they had listed all the foods consumed at dinner the night before the session and at breakfast the morning of the session. Having been instructed to eat a normal dinner the night before and to eat the same breakfast on the mornings of all the session days, the food diaries were used to monitor the subjects' compliance. After the final session, all subjects were given a discharge questionnaire to answer and were compensated for their time.
Ratings Subjects rated their hunger, thirst, fullness, desire to eat, desire to drink, how much they thought they could eat and how much they thought they could drink at specific times throughout the sessions by marking their responses on 100-mm visual analog scales. Each 100-mm scale was preceded by a question, i.e., "how hungry are you right now?", and anchored by the prompts "extremely hungry" on the right and "not at all hungry" on the left. All the ratings were done on IBM XT computers which were set up before the subject's arrival with a program that presented each subject with instructions and then one rating on the screen at a time. A subject responded to each question by manipulating a joystick to control the placement of a mark on the scale and by pressing a button on the joystick to record his response. Upon receiving a response, the computer automatically presented the subject with the next question. After a session, the ratings were printed out from the data disk containing the recorded responses. Each subject's baseline appetite ratings were checked at the beginning of each session to ensure that the ratings deviated no more than ___15 mm across all the sessions.
Foods and Macronutrients Lemonade. The test drink developed for use in this study was a lemonade made with chilled water (7°C) and freshly squeezed lemon juice sweetened to equal intensity with either sucrose or aspartame (a ratio of 200 g of sucrose for every 1 g of aspartame was used to match sweetness). The lemonade was made each day by mixing 20 g freshly squeezed lemon juice with every 200 g of chilled water, then adding the sweetener (either 20 g of table sugar or 0.1 g of aspartame for every 200 g water). The sweetener concentrations were comparable to the amounts found in commercially available soft drinks. Caloric densities of each lemonade are given in Table 1. Triangle test. The sucrose- and aspartame-sweetened lemonades were developed to taste so similarly as to be almost indistinguishable in a triangle taste test (17). Tasters were required to choose the two samples most similar in taste from 3 unlabeled samples of lemonade (two samples of one type of lemonade and one sample of the other type). The order of presentation of the samples was counterbalanced. All the subjects were given two
SWEETENERS AND FOOD INTAKE
triangle tests after their last session to determine how well the covert manipulation worked. Self-selection lunch. Foods used in the self-selection lunch consisted of the following: 3 roast beef sandwiches [2.33 kcal/g, 0.085 fat (g)/g, 0.197 carbohydrate (g)/g, 0.179 g protein (g)/g], each sandwich made with 2 slices of Roman Meal whole wheat bread, 5 g of Hellman's Real mayonnaise, 5 g Guldens spicy brown mustard, 50 g lean roast beef and 15 g lettuce and then cut into four equal squares; 3 ham and Swiss cheese sandwiches [2.28 kcal/g, 0.110 fat (g)/g, 0.170 carbohydrate (g)/g, 0.160 protein (g)/g] made exactly as the roast beef sandwiches except each sandwich contained 50 g of extra lean imported ham and 30 g of domestic Swiss cheese; 40 g of Utz potato chips [5.36 kcal/g, 0.350 fat (g)/g, 0.490 carbohydrate (g)/g, 0.070 protein (g)/g]; 10 Lorna Doone shortbread cookies [5.30 kcal/g, 0.250 fat (g)/g, 0.640 carbohydrate (g)/g, 0.014 protein (g)/g]; and 100 g of cherry jelly beans [3.71 kcal/g, 0 g fat, 0.940 carbohydrate (g)/g, 0 g protein]. All foods were weighed before being served and reweighed after the subjects had eaten to obtain the amount consumed (g) of each food.
Statistics Energy intake including and not including the calories from the lemonade was analyzed using two-way analysis of variance with repeated measures (BMDP software) to determine if there were any significant differences among conditions. A post hoc Tukey's HSD test was employed whenever any significant main effects were observed to ascertain which means differed significantly. Macronutrient selection in the different conditions was examined by comparing the percents of the total macronutrients consumed (by weight) derived from fat, carbohydrate, or protein across conditions using two-way analysis of variance. The macronutrient breakdown of all the foods was obtained from two main sources: the actual manufacturers of the food products and Bowes & Church's Food Values of Portions Commonly Used (11). Food selection was also examined by comparing the amount of calories consumed of each food item during lunch across conditions in all three experiments using two-way analysis of variance. Changes in appetite ratings (hunger, thirst, fullness, desire to eat, desire to drink, prospective consumption, prospective drinking) were analyzed using two-way, three-factor analysis of variance with condition (type of load), dose (8 oz or 16 oz), and time as factors. For each question, the initial or baseline rating of each session was subtracted from all subsequent ratings to obtain the difference scores used in the analyses. The baselines for all ratings in all 3 experiments were also analyzed to determine the withinsubject variance; baseline ratings were found not to vary significantly across conditions. The absolute scores for each appetite rating were also analyzed but the results were very similar to those obtained from analyses of the difference scores. Consequently, only the statistical analyses of the change in appetite ratings rather than absolute ratings will be referred to in this publication so as to avoid any confusion or repetition. Data from the no load condition were not included in this analysis; however, each load condition was separately compared to the no load condition using the Bonferoni test for significance. Pearson's product moment tests were used in all three experiments to determine the degree of correlation between energy intake and the change in hunger or thirst ratings (from baseline to immediately after lunch) during each condition. We were interested in examining whether increased or decreased energy intake was reflected in the amount of hunger or thirst suppression. The Greenhouse-Geisserprobability value will be given instead of the tail probability value whenever any statistically significant
21
TABLE 2 TIMELINESFOR VISUALANALOGSCALERATINGS Before Lunch Time (min)
70
60
Experiment 1 Experiment 2 Experiment 3 #1 D #2
50 40
30
After Lunch 20 10
#1 D #2 #3
0 #1 L+D #2 #3 L #4 #4 L #5
. 60 #3 #5 #6
L = lunch. D =drink. #n = visual analog scale rating.
results are discussed so that Type I errors often associated with repeated measures analyses may be minimized. EXPERIMENT 1 This experiment was designed to examine the effect of different drinks on food intake when taken with the meal. METHOD Subjects were instructed to arrive at the laboratory at their usual lunchtime (approx. 12:00 p.m.) for each of the 7 sessions. Upon arrival, subjects completed their initial appetite ratings which served as the baseline ratings (refer to Table 2). Subjects then received their self-selection lunch meal along with the test drink or no drink and were instructed to eat as much as they wanted. They were told to ask for more of any of the foods if they desired. Also, they were instructed to consume all of the drink during the meal and given a maximum of 20 min to eat (no subject found finishing within the allowed time difficult). After having eaten as much as they wanted, subjects completed their second set of appetite ratings. They then sat quietly in their cubicles reading for one hour, after which they completed their third set of appetite ratings. RESULTS
Energy Intake Analysis of variance comparing energy intake during the self-selection meal across all 7 conditions without including the lemonade calories (see Table 3) produced no statistically significant differences. Energy intake in this experiment was not affected by subjects being deprived of a drink when eating their meal. However, when the lemonade calories were added to the calories consumed during lunch (see Table 3), a two-factor, two-way analysis of variance test comparing intake associated with the drink conditions (excluding the no drink condition) showed a significant main effect for condition, F(2,26)--6.70, p
0031-9384/90 $3.00 + .00
Effects of Drinks Sweetened With Sucrose or Aspartame on Hunger, Thirst and Food Intake in Men B A R B A R A J. R O L L S , 1 S I O N K I M A N D I N G R I D C. F E D O R O F F
Department o f Psychiatry and Behavioral Sciences The Johns Hopkins University School o f Medicine, Baltimore, M D 21205 R e c e i v e d 15 February 1990
ROLLS, B. J., S. KIM AND I. C. FEDOROFF. Effects of drinks sweetened with sucrose or aspartame on hunger, thirst and food intake in men. PHYSIOL BEHAV 48(1) 19-26, 1990.--Forty-two nondieting adult males were given 8 or 16 oz of lemonade, sweetened to equal intensity with either aspartame or sucrose, or the same volumes of water, or no drink. Subjects were separated into three groups receiving the drinks at different times: with a self-selection lunch, or 30, or 60 rain before lunch. Food intakes did not differ when subjects received the drinks with lunch; however, when the calories from the drinks were included, intake was significantly greater with the sucrose-sweetened lemonades than in the other conditions. When subjects received the drinks 30 or 60 min before lunch, food intakes were not significantly different. Appetite ratings were not different among the conditions. When the drinks were consumed with the meal, the 8-oz sucrose-sweetened lemonade differed from the other drinks in that it did not significantly reduce thirst. The results indicate that in nondieting males, aspartame in concentrations similar to those in commercially available drinks did not increase hunger ratings or food intake. However, caloric drinks taken with lunch increased total energy intake in that meal. Also, sucrose-sweetened drinks may decrease thirst less than water or aspartame-sweetened drinks when taken with a meal. Aspartame
Sucrose
Thirst
Hunger
Caloric compensation
Drinks
to five-year-old children, Birch et al. (3) found that fruit-flavored drinks sweetened with aspartame had either no effect or produced a significant suppression, not enhancement, of subsequent food intake compared with the same volume of water. Sucrose-sweetened drinks caused an even greater suppression so that there was compensation for the calories coming from sugar. Anderson et al. (1) studied older children (nine to ten years old) and found that aspartame-sweetened fruit drinks, with or without additional carbohydrate (polycose), did not affect hunger or lunchtime food intake when compared with drinks sweetened with sodium cyclamate or sucrose. Thus, there is no indication that aspartame increases appetite or food intake in children. We (15) conducted a study in adults which examined the effects of commercially available pudding and jello sweetened with either sucrose or aspartame on appetite ratings and food intake. The aspartame-sweetened and sucrose-sweetened foods had similar effects on ratings of hunger and satiety over the two hours after consumption that they were tracked. Despite the difference in caloric intake (up to 206 kcal), subjects showed only a nonsignificant trend towards caloric compensation when presented with a variety of foods two hours later. We reasoned that our failure to f'md increases in ratings of appetite could be because we used foods rather than solutions. Thus, in the present study, we compared two volumes of lemonade
ALTHOUGH intense sweeteners such as saccharin, cyclamate or aspartame have been used for a number of years to reduce the energy density of foods and drinks, studies have only recently begun to clarify how they affect food intake and appetite. When Blundell and Hill (4) suggested that aspartame might increase appetite and that such dietary aids " m a y contribute to disordered patterns of eating prevalent among certain groups of normal weight individuals," there was little data available to contradict this view. Porikos et al. (12) had found that the substitution of aspartame for sucrose over 12-day periods led to a 15% reduction in daily caloric intake. Additionally, Rolls et al. (15) had reported that consumption of jello sweetened with aspartame suppressed hunger to the same extent as did sucrosesweetened jello. Despite the fact that none of these studies had found that aspartame increases food intake, it became widely believed that intense sweeteners might be of no benefit in weight control and, indeed, that they might even lead to weight gain. Blundell and co-workers (13) have continued to report that aspartame and other intense sweeteners in solution increase appetite ratings, but have failed to find an increase in food intake following aspartame. Several other studies have failed to find that aspartame increases appetite ratings or food intake. Two studies have used familiar drinks in children of various ages to study the effects of intense sweeteners. In a study of two-
~Requests for reprints should be addressed to Dr. B. J. Rolls, Department of Psychiatry and Behavioral Sciences, Meyer 1-104, The Johns Hopkins University School of Medicine, 600 N. Wolfe St., Baltimore, MD 21205.
19
20
ROLLS, KIM AND FEDOROFF
sweetened with sucrose or aspartame, the same volumes of water, and a condition in which no drink was consumed. Since the time between consuming the drinks and time of food consumption could affect the results (3,5), we conducted three experiments, each with a different time relationship between the drinks and lunch. A unique feature of the present studies was the investigation of the effects of the different drinks when they were consumed not as a preload, but were taken with the meal. Although drinks are frequently consumed with meals, this design had not been previously employed. Another new aspect of the present research was to determine the effects of the differently sweetened drinks on ratings of thirst. There is no previous work comparing the effects of consumption of sucrose- and aspartame-sweetened drinks on thirst. GENERAL METHOD
Subjects Forty-two healthy male volunteers aged between 21 and 39 (mean age=25.0, S . D . = 4 . 3 ) were recruited for three experiments (14 in each experiment) which were run concurrently. The men were all nonsmoking, nondieting, of normal weight for height (between 90% and 110% of ideal body weight) according to the 1983 Metropolitan Life Weight Tables (mean=99.7%, S.D. = 8.3%), and had no food allergies or restrictions nor took any medication. Furthermore, the volunteers had been screened using the Stunkard Eating Inventory (18) which measures cognitive dietary restraint, the Eating Attitudes Test (10) which detects any symptoms of an eating disorder, and the Zung Self-Rating questionnaire (20) which measures depression. Only those men scoring less than 10 on the Eating Inventory (mean score = 5.2, S.D. = 2.9), less than 15 on the Eating Attitudes Test (mean score=7.0, S.D. =3.4) and less than 50 on the Zung (mean score=34.9, S.D. =4.3) were asked to participate. Finally, to ensure that the subjects liked the foods used in the experiment and would eat the foods, as well as to minimize a novelty effect, each recruited volunteer was required to fast complete a trial session in which he would rate the pleasantness of the taste of all the foods. Only those volunteers rating the taste of all the foods above 50 on 100-mm visual analog scales anchored by "extremely pleasant" on the right and "not at all pleasant" on the left were run in the experiment, When asked whether they always consumed a drink with their lunch, all subjects responded affirmatively. Finally, all subjects signed a consent form at the beginning of their first session which stated that the purpose of the study was to investigate the effects of different drinks (lemonade and water) on thirst and appetite ratings. Subjects, however, were naive as to the sweetener manipulation in the study.
General Procedure All three experiments utihzed a within-subjects, repeated measures design with each subject serving as his own control. Every subject received each of the following drinks: 8 oz sucrosesweetened lemonade, 16 oz sucrose-sweetened lemonade, 8 oz aspartarne-sweetened lemonade, 16 oz aspartame-sweetened lemonade, 8 oz water, and 16 oz water. On one occasion, no drink was given. Each subject was run once in each of the 7 conditions, and the order of presentation of conditions was counterbalanced across subjects. Subjects came to the laboratory at the same designated time for each of the seven sessions with at least three days between sessions. Upon arrival, all subjects presented food diaries on
TABLE 1 DRINK CHARACTERISTICS
Drink
Volume
Lemonade Lemonade Lemonade Lemonade Water Water
8 oz 16 oz 8 oz 16 oz 8 oz 16 oz
Sweetener (g)
Calories
Sucrose (20) Sucrose (40) Aspartame (0.11) Aspartame (0.22) None None
83 166 5 10 0 0
which they had listed all the foods consumed at dinner the night before the session and at breakfast the morning of the session. Having been instructed to eat a normal dinner the night before and to eat the same breakfast on the mornings of all the session days, the food diaries were used to monitor the subjects' compliance. After the final session, all subjects were given a discharge questionnaire to answer and were compensated for their time.
Ratings Subjects rated their hunger, thirst, fullness, desire to eat, desire to drink, how much they thought they could eat and how much they thought they could drink at specific times throughout the sessions by marking their responses on 100-mm visual analog scales. Each 100-mm scale was preceded by a question, i.e., "how hungry are you right now?", and anchored by the prompts "extremely hungry" on the right and "not at all hungry" on the left. All the ratings were done on IBM XT computers which were set up before the subject's arrival with a program that presented each subject with instructions and then one rating on the screen at a time. A subject responded to each question by manipulating a joystick to control the placement of a mark on the scale and by pressing a button on the joystick to record his response. Upon receiving a response, the computer automatically presented the subject with the next question. After a session, the ratings were printed out from the data disk containing the recorded responses. Each subject's baseline appetite ratings were checked at the beginning of each session to ensure that the ratings deviated no more than ___15 mm across all the sessions.
Foods and Macronutrients Lemonade. The test drink developed for use in this study was a lemonade made with chilled water (7°C) and freshly squeezed lemon juice sweetened to equal intensity with either sucrose or aspartame (a ratio of 200 g of sucrose for every 1 g of aspartame was used to match sweetness). The lemonade was made each day by mixing 20 g freshly squeezed lemon juice with every 200 g of chilled water, then adding the sweetener (either 20 g of table sugar or 0.1 g of aspartame for every 200 g water). The sweetener concentrations were comparable to the amounts found in commercially available soft drinks. Caloric densities of each lemonade are given in Table 1. Triangle test. The sucrose- and aspartame-sweetened lemonades were developed to taste so similarly as to be almost indistinguishable in a triangle taste test (17). Tasters were required to choose the two samples most similar in taste from 3 unlabeled samples of lemonade (two samples of one type of lemonade and one sample of the other type). The order of presentation of the samples was counterbalanced. All the subjects were given two
SWEETENERS AND FOOD INTAKE
triangle tests after their last session to determine how well the covert manipulation worked. Self-selection lunch. Foods used in the self-selection lunch consisted of the following: 3 roast beef sandwiches [2.33 kcal/g, 0.085 fat (g)/g, 0.197 carbohydrate (g)/g, 0.179 g protein (g)/g], each sandwich made with 2 slices of Roman Meal whole wheat bread, 5 g of Hellman's Real mayonnaise, 5 g Guldens spicy brown mustard, 50 g lean roast beef and 15 g lettuce and then cut into four equal squares; 3 ham and Swiss cheese sandwiches [2.28 kcal/g, 0.110 fat (g)/g, 0.170 carbohydrate (g)/g, 0.160 protein (g)/g] made exactly as the roast beef sandwiches except each sandwich contained 50 g of extra lean imported ham and 30 g of domestic Swiss cheese; 40 g of Utz potato chips [5.36 kcal/g, 0.350 fat (g)/g, 0.490 carbohydrate (g)/g, 0.070 protein (g)/g]; 10 Lorna Doone shortbread cookies [5.30 kcal/g, 0.250 fat (g)/g, 0.640 carbohydrate (g)/g, 0.014 protein (g)/g]; and 100 g of cherry jelly beans [3.71 kcal/g, 0 g fat, 0.940 carbohydrate (g)/g, 0 g protein]. All foods were weighed before being served and reweighed after the subjects had eaten to obtain the amount consumed (g) of each food.
Statistics Energy intake including and not including the calories from the lemonade was analyzed using two-way analysis of variance with repeated measures (BMDP software) to determine if there were any significant differences among conditions. A post hoc Tukey's HSD test was employed whenever any significant main effects were observed to ascertain which means differed significantly. Macronutrient selection in the different conditions was examined by comparing the percents of the total macronutrients consumed (by weight) derived from fat, carbohydrate, or protein across conditions using two-way analysis of variance. The macronutrient breakdown of all the foods was obtained from two main sources: the actual manufacturers of the food products and Bowes & Church's Food Values of Portions Commonly Used (11). Food selection was also examined by comparing the amount of calories consumed of each food item during lunch across conditions in all three experiments using two-way analysis of variance. Changes in appetite ratings (hunger, thirst, fullness, desire to eat, desire to drink, prospective consumption, prospective drinking) were analyzed using two-way, three-factor analysis of variance with condition (type of load), dose (8 oz or 16 oz), and time as factors. For each question, the initial or baseline rating of each session was subtracted from all subsequent ratings to obtain the difference scores used in the analyses. The baselines for all ratings in all 3 experiments were also analyzed to determine the withinsubject variance; baseline ratings were found not to vary significantly across conditions. The absolute scores for each appetite rating were also analyzed but the results were very similar to those obtained from analyses of the difference scores. Consequently, only the statistical analyses of the change in appetite ratings rather than absolute ratings will be referred to in this publication so as to avoid any confusion or repetition. Data from the no load condition were not included in this analysis; however, each load condition was separately compared to the no load condition using the Bonferoni test for significance. Pearson's product moment tests were used in all three experiments to determine the degree of correlation between energy intake and the change in hunger or thirst ratings (from baseline to immediately after lunch) during each condition. We were interested in examining whether increased or decreased energy intake was reflected in the amount of hunger or thirst suppression. The Greenhouse-Geisserprobability value will be given instead of the tail probability value whenever any statistically significant
21
TABLE 2 TIMELINESFOR VISUALANALOGSCALERATINGS Before Lunch Time (min)
70
60
Experiment 1 Experiment 2 Experiment 3 #1 D #2
50 40
30
After Lunch 20 10
#1 D #2 #3
0 #1 L+D #2 #3 L #4 #4 L #5
. 60 #3 #5 #6
L = lunch. D =drink. #n = visual analog scale rating.
results are discussed so that Type I errors often associated with repeated measures analyses may be minimized. EXPERIMENT 1 This experiment was designed to examine the effect of different drinks on food intake when taken with the meal. METHOD Subjects were instructed to arrive at the laboratory at their usual lunchtime (approx. 12:00 p.m.) for each of the 7 sessions. Upon arrival, subjects completed their initial appetite ratings which served as the baseline ratings (refer to Table 2). Subjects then received their self-selection lunch meal along with the test drink or no drink and were instructed to eat as much as they wanted. They were told to ask for more of any of the foods if they desired. Also, they were instructed to consume all of the drink during the meal and given a maximum of 20 min to eat (no subject found finishing within the allowed time difficult). After having eaten as much as they wanted, subjects completed their second set of appetite ratings. They then sat quietly in their cubicles reading for one hour, after which they completed their third set of appetite ratings. RESULTS
Energy Intake Analysis of variance comparing energy intake during the self-selection meal across all 7 conditions without including the lemonade calories (see Table 3) produced no statistically significant differences. Energy intake in this experiment was not affected by subjects being deprived of a drink when eating their meal. However, when the lemonade calories were added to the calories consumed during lunch (see Table 3), a two-factor, two-way analysis of variance test comparing intake associated with the drink conditions (excluding the no drink condition) showed a significant main effect for condition, F(2,26)--6.70, p
