Physiology& Behavior, Vol. 47, pp. 287-291. ©PergamonPress pie, 1990. Printedin the U.S.A.
0031-9384/90$3.00 + .00
Variations in Food Preference and Consumption Across the Menstrual Cycle D E B O R A H J. B O W E N
Division of Public Health Sciences, Fred Hutchinson Cancer Research Center 1124 Columbia Street, Seattle, WA 98104 AND N E I L E. G R U N B E R G
Department of Medical Psychology, Uniformed Services University of the Health Sciences Bethesda, MD 20814 Received 6 D e c e m b e r 1988
BOWEN, D. J. AND N. E. GRUNBERG. Variations in food preference and consumption across the menstrual cycle. PHYSIOL BEHAV 47(2) 287-291, 1990.--There is some evidence that food consumption changes across the menstrual cycle. However, archived studies tend to rely on self-report data and do not differentiate among types of food eaten. The present laboratory study was designed to measure women's taste preferences and specific food consumption across the menstrual cycle. Women came into the laboratory, consumed everyday foods, and rated them on taste judgment scales. The foods were weighed before and after the tasting session to determine amount eaten. After the tasting session, subjects completed several questionnaires (e.g., measures of affect, restrained eating, menstrual cycle and menstrual symptoms). Physical measurements (e.g., height, weight) were taken at the end of the session. Sweet food consumption and preference ratings were significantly higher during the premenstrual period. Consumption and ratings of other foods did not differ. There were no differences in levels of mood, obesity, or restrained eating across the menstrual cycle. Low restraint was associated with reporting fewer menstrual symptoms, in general, and slightly better mood. These results support the idea that specific taste preferences change over the menstrual cycle, and that endocrinological factors could be involved in this phenomenon. Food consumption
Menstrual cycle
Taste
Gonadal steroids
ANECDOTAL and empirical reports indicate that women's food consumption changes across the menstrual cycle. Anecdotally, women report increased hunger immediately before the onset of the menstrual period. Some empirical evidence indicates greater caloric intake, snack food consumption, and body weight during several premenstrual days compared with the postmenstrual period (5, 12, 18). Because gonadal steroids clearly affect food consumption in animals (2, 15, 16), perhaps food consumption or taste preference also is altered in humans by steroid levels. Human 24-hr energy expenditure increases during the premenstrual period (17), supporting the existence of some metabolic premenstrual change that could modify food consumption. One possibility that has not been fully investigated in studies of food intake across the menstrual cycle is a possible change in preference for a particular taste or taste class. Positive hedonic judgment or pleasantness of food taste can lead to a preference for a specific food or taste and thereby increase specific food consumption. Caloric need can cause humans to report increased preferences for specific foods, particularly sweet foods, and to consume more of those foods (4). These changes in caloric need can be caused by environmental changes, such as food deprivation, and by internal changes, such as hormone levels.
The human menstrual cycle is a naturally occurring fluctuation of circulating levels of gonadal steroids. If circulating steroid levels do affect food consumption and food preference, then these variables should vary over the course of the menstrual cycle. Specifically, we predict that when levels of progestins and estrogens are at their highest, then food consumption, particularly sweet food consumption, will be high. When these hormones are at low levels, or when only estrogen is high, then food consumption will be low. The present study was designed to measure human taste preference and food consumption across the menstrual cycle. Women came into the laboratory, consumed everyday foods, and rated them on taste judgment scales. The foods were weighed before and after the tasting session to determine amount eaten. After the tasting session, subjects completed several questionnaires (e.g., measures of affect, restrained eating). Day of menstrual cycle and menstrual symptoms were assessed. Physical measurements (e.g., height, weight) were taken at the end of the session. METHOD
Subjects Subjects for this study were 39 women recruited through a 287
288
newspaper ad asking for women interested in participating in a study of sensory sensitivity changes and stress. Subjects were unaware of the interest in stage of menstrual cycle when they were recruited for the present study. The research team scheduled each subject for a laboratory appointment at a mutually convenient time between 11:15 a.m. and 2:30 p.m. Stage of menstrual cycle was assessed after the visit using information from a one-page questionnaire and by a follow-up phone call close to the reported beginning date of the subsequent period to confirm the questionnaire answers. Subjects ranged in age from 19-40 years (mean = 28.6; s.d. --=5.2) and were healthy nonsmokers who were not using birth control pills at the time of the study. There were no significant differences for any demographic variable over any of the stages of the cycle (described below). Subjects were asked to eat their regular breakfast and to refrain from eating for three hours before the experimental session. Procedure The female experimenter greeted the subject and escorted her to the experimental suite where several rooms were set up with equipment to test for perception of different stimuli. Subjects were told that the study involved tests of sensation and perception, including sight, sound, and taste. The experimenter consulted a chart on her clipboard and informed every subject that, "We don't have time in one experimental session to test everyone on all of their senses. You are in the taste condition. I'd like you to taste some everyday foods and to fill out some taste judgment questionnaires." This procedure was designed to minimize selfconsciousness about eating. The experimenter then escorted the subject into the taste room and explained to her the procedure for testing the foods and filling out the taste judgment forms. After the explanations, the experimenter said, "It's important that your ratings are as accurate as possible, so eat as much of each food as you need in making your judgments. After you leave, I ' m going to throw the foods out anyway, so you might as well eat what you want." The experimenter left the subject in the room for 20 minutes to taste the foods. Previous research (3,7) and pretests found that this amount of time was ample for the subjects to taste all the foods and then to nibble. After the 20-minute tasting session, the experimenter took the subject to a nearby room in the suite and gave her a packet of questionnaires to complete. When she was finished, the physical measures were taken, and the subject was paid $15 and debriefed. Measurements Each subject was asked to taste nine different foods. These foods were weighed before and after the taste test to determine consumption of each food, but the subject was unaware of the weighing. These procedures were based on previous research (3,7) which found these laboratory measures of food preferences to be valid and sensitive. Three sweet (coffee cake, chocolate, gum drops), three salty (ham, salted peanuts, salted pretzel sticks), and three bland foods (cheese, unsalted crackers, unsalted peanuts), representing both meal and snack choices, were provided in the study. The mean caloric density was 4.27, 3.90, and 4.37 kcal/gram for sweet, salty, and bland categories, respectively. Thus, consumption in grams is approximately analogous to caloric consumption when values were averaged by taste class. All foods were presented in bite-size portions in beige bowls. The order of foods was counterbalanced for food taste class. Cold water was also available. The goal of this taste test was to provide a naturalistic setting where subjects could consume freely. The taste
BOWEN AND GRUNBERG
test was conducted in a 5' × 9' room with a table and chair for the subject and a chair for the experimenter to use while explaining the taste test. For each food that the subject tasted, she was asked to complete one taste judgment sheet. On each sheet was a list of taste dimensions. The subject was asked to circle a number between 1 (not at all) and 7 (extremely) for each of the following taste dimensions: sweet, sour, salty, bitter, spicy, flavorful/bland, smooth/grainy, light/heavy, strong/mild, and fresh/stale. The judgments for each type of food were used as manipulation check (i. e., did the "sweet" foods really taste sweet?) and as an evaluation of taste perception. Each subject was given a menstrual cycle questionnaire assessing the stage of cycle she was in on the day of the study, and was called near the projected starting day of her next menstrual cycle to verify her previous self-report. Height and weight were measured at the end of the laboratory session. A body mass index (weight/height 2) was calculated for each woman. Subjects completed the Multiple Affect Adjective Checklist (MAACL) to assess affect (19). The Restraint Scale (10) was used to measure restraint of eating. Restrained eating is defined as consciously holding back from eating everything one wishes. Restraint has been shown to play an important role in the determinants of eating (10). Data Analysis Data from two of 39 subjects were dropped from the analyses because they were inaccurate in their projected day of menstrual cycle by more than two days and because their periods exceeded a 32-day cycle. Analyses of the data gathered from 37 subjects were performed comparing levels of several dependent variables between stages of the menstrual cycle. Subjects were grouped into stage of menstrual cycle in two ways. In the first method of staging (the "phase" method) subjects were assigned to either follicular phase (days 1-14; 18 subjects) or luteal phase (days 15-28; 19 subjects), based on reported day of menstrual cycle. This twophase staging is commonly used in menstrual cycle research [e.g. (1,11)] and was used in the present study to enable direct comparisons of data to the findings of other studies. The second method of staging (the "period" method) involved assigning subjects to a period of their cycle based on predicted endogenous hormone levels. Hormone levels of interest were based on study hypotheses. Estrogens and progestins are low during days 1-12, estrogens are high but progestins remain low during days 12-14 and both estrogens and progestins are high during days 20-26 (9). Subjects were reassigned to one of these periods (14, 4, and 10 subjects, respectively), based again on reported day of the menstrual cycle. Both methods of menstrual cycle staging were used to analyze the data collected during the laboratory session. RESULTS
The first set of analyses examined food consumption in the laboratory during different phases and periods of the menstrual cycle. Figure 1 presents means of food consumption by taste classes (sweet, salty, and bland) for the subjects assigned to follicular and luteal phases. Subjects in both phases consumed significantly more sweet food than either salty or bland food [paired t(36)=2.54, 2.73, p < 0 . 0 5 for salty and bland food, respectively]. Subjects in the luteal phase consumed significantly more sweet food than did subjects in the follicular phase, t(35) = 3.01, p
0031-9384/90$3.00 + .00
Variations in Food Preference and Consumption Across the Menstrual Cycle D E B O R A H J. B O W E N
Division of Public Health Sciences, Fred Hutchinson Cancer Research Center 1124 Columbia Street, Seattle, WA 98104 AND N E I L E. G R U N B E R G
Department of Medical Psychology, Uniformed Services University of the Health Sciences Bethesda, MD 20814 Received 6 D e c e m b e r 1988
BOWEN, D. J. AND N. E. GRUNBERG. Variations in food preference and consumption across the menstrual cycle. PHYSIOL BEHAV 47(2) 287-291, 1990.--There is some evidence that food consumption changes across the menstrual cycle. However, archived studies tend to rely on self-report data and do not differentiate among types of food eaten. The present laboratory study was designed to measure women's taste preferences and specific food consumption across the menstrual cycle. Women came into the laboratory, consumed everyday foods, and rated them on taste judgment scales. The foods were weighed before and after the tasting session to determine amount eaten. After the tasting session, subjects completed several questionnaires (e.g., measures of affect, restrained eating, menstrual cycle and menstrual symptoms). Physical measurements (e.g., height, weight) were taken at the end of the session. Sweet food consumption and preference ratings were significantly higher during the premenstrual period. Consumption and ratings of other foods did not differ. There were no differences in levels of mood, obesity, or restrained eating across the menstrual cycle. Low restraint was associated with reporting fewer menstrual symptoms, in general, and slightly better mood. These results support the idea that specific taste preferences change over the menstrual cycle, and that endocrinological factors could be involved in this phenomenon. Food consumption
Menstrual cycle
Taste
Gonadal steroids
ANECDOTAL and empirical reports indicate that women's food consumption changes across the menstrual cycle. Anecdotally, women report increased hunger immediately before the onset of the menstrual period. Some empirical evidence indicates greater caloric intake, snack food consumption, and body weight during several premenstrual days compared with the postmenstrual period (5, 12, 18). Because gonadal steroids clearly affect food consumption in animals (2, 15, 16), perhaps food consumption or taste preference also is altered in humans by steroid levels. Human 24-hr energy expenditure increases during the premenstrual period (17), supporting the existence of some metabolic premenstrual change that could modify food consumption. One possibility that has not been fully investigated in studies of food intake across the menstrual cycle is a possible change in preference for a particular taste or taste class. Positive hedonic judgment or pleasantness of food taste can lead to a preference for a specific food or taste and thereby increase specific food consumption. Caloric need can cause humans to report increased preferences for specific foods, particularly sweet foods, and to consume more of those foods (4). These changes in caloric need can be caused by environmental changes, such as food deprivation, and by internal changes, such as hormone levels.
The human menstrual cycle is a naturally occurring fluctuation of circulating levels of gonadal steroids. If circulating steroid levels do affect food consumption and food preference, then these variables should vary over the course of the menstrual cycle. Specifically, we predict that when levels of progestins and estrogens are at their highest, then food consumption, particularly sweet food consumption, will be high. When these hormones are at low levels, or when only estrogen is high, then food consumption will be low. The present study was designed to measure human taste preference and food consumption across the menstrual cycle. Women came into the laboratory, consumed everyday foods, and rated them on taste judgment scales. The foods were weighed before and after the tasting session to determine amount eaten. After the tasting session, subjects completed several questionnaires (e.g., measures of affect, restrained eating). Day of menstrual cycle and menstrual symptoms were assessed. Physical measurements (e.g., height, weight) were taken at the end of the session. METHOD
Subjects Subjects for this study were 39 women recruited through a 287
288
newspaper ad asking for women interested in participating in a study of sensory sensitivity changes and stress. Subjects were unaware of the interest in stage of menstrual cycle when they were recruited for the present study. The research team scheduled each subject for a laboratory appointment at a mutually convenient time between 11:15 a.m. and 2:30 p.m. Stage of menstrual cycle was assessed after the visit using information from a one-page questionnaire and by a follow-up phone call close to the reported beginning date of the subsequent period to confirm the questionnaire answers. Subjects ranged in age from 19-40 years (mean = 28.6; s.d. --=5.2) and were healthy nonsmokers who were not using birth control pills at the time of the study. There were no significant differences for any demographic variable over any of the stages of the cycle (described below). Subjects were asked to eat their regular breakfast and to refrain from eating for three hours before the experimental session. Procedure The female experimenter greeted the subject and escorted her to the experimental suite where several rooms were set up with equipment to test for perception of different stimuli. Subjects were told that the study involved tests of sensation and perception, including sight, sound, and taste. The experimenter consulted a chart on her clipboard and informed every subject that, "We don't have time in one experimental session to test everyone on all of their senses. You are in the taste condition. I'd like you to taste some everyday foods and to fill out some taste judgment questionnaires." This procedure was designed to minimize selfconsciousness about eating. The experimenter then escorted the subject into the taste room and explained to her the procedure for testing the foods and filling out the taste judgment forms. After the explanations, the experimenter said, "It's important that your ratings are as accurate as possible, so eat as much of each food as you need in making your judgments. After you leave, I ' m going to throw the foods out anyway, so you might as well eat what you want." The experimenter left the subject in the room for 20 minutes to taste the foods. Previous research (3,7) and pretests found that this amount of time was ample for the subjects to taste all the foods and then to nibble. After the 20-minute tasting session, the experimenter took the subject to a nearby room in the suite and gave her a packet of questionnaires to complete. When she was finished, the physical measures were taken, and the subject was paid $15 and debriefed. Measurements Each subject was asked to taste nine different foods. These foods were weighed before and after the taste test to determine consumption of each food, but the subject was unaware of the weighing. These procedures were based on previous research (3,7) which found these laboratory measures of food preferences to be valid and sensitive. Three sweet (coffee cake, chocolate, gum drops), three salty (ham, salted peanuts, salted pretzel sticks), and three bland foods (cheese, unsalted crackers, unsalted peanuts), representing both meal and snack choices, were provided in the study. The mean caloric density was 4.27, 3.90, and 4.37 kcal/gram for sweet, salty, and bland categories, respectively. Thus, consumption in grams is approximately analogous to caloric consumption when values were averaged by taste class. All foods were presented in bite-size portions in beige bowls. The order of foods was counterbalanced for food taste class. Cold water was also available. The goal of this taste test was to provide a naturalistic setting where subjects could consume freely. The taste
BOWEN AND GRUNBERG
test was conducted in a 5' × 9' room with a table and chair for the subject and a chair for the experimenter to use while explaining the taste test. For each food that the subject tasted, she was asked to complete one taste judgment sheet. On each sheet was a list of taste dimensions. The subject was asked to circle a number between 1 (not at all) and 7 (extremely) for each of the following taste dimensions: sweet, sour, salty, bitter, spicy, flavorful/bland, smooth/grainy, light/heavy, strong/mild, and fresh/stale. The judgments for each type of food were used as manipulation check (i. e., did the "sweet" foods really taste sweet?) and as an evaluation of taste perception. Each subject was given a menstrual cycle questionnaire assessing the stage of cycle she was in on the day of the study, and was called near the projected starting day of her next menstrual cycle to verify her previous self-report. Height and weight were measured at the end of the laboratory session. A body mass index (weight/height 2) was calculated for each woman. Subjects completed the Multiple Affect Adjective Checklist (MAACL) to assess affect (19). The Restraint Scale (10) was used to measure restraint of eating. Restrained eating is defined as consciously holding back from eating everything one wishes. Restraint has been shown to play an important role in the determinants of eating (10). Data Analysis Data from two of 39 subjects were dropped from the analyses because they were inaccurate in their projected day of menstrual cycle by more than two days and because their periods exceeded a 32-day cycle. Analyses of the data gathered from 37 subjects were performed comparing levels of several dependent variables between stages of the menstrual cycle. Subjects were grouped into stage of menstrual cycle in two ways. In the first method of staging (the "phase" method) subjects were assigned to either follicular phase (days 1-14; 18 subjects) or luteal phase (days 15-28; 19 subjects), based on reported day of menstrual cycle. This twophase staging is commonly used in menstrual cycle research [e.g. (1,11)] and was used in the present study to enable direct comparisons of data to the findings of other studies. The second method of staging (the "period" method) involved assigning subjects to a period of their cycle based on predicted endogenous hormone levels. Hormone levels of interest were based on study hypotheses. Estrogens and progestins are low during days 1-12, estrogens are high but progestins remain low during days 12-14 and both estrogens and progestins are high during days 20-26 (9). Subjects were reassigned to one of these periods (14, 4, and 10 subjects, respectively), based again on reported day of the menstrual cycle. Both methods of menstrual cycle staging were used to analyze the data collected during the laboratory session. RESULTS
The first set of analyses examined food consumption in the laboratory during different phases and periods of the menstrual cycle. Figure 1 presents means of food consumption by taste classes (sweet, salty, and bland) for the subjects assigned to follicular and luteal phases. Subjects in both phases consumed significantly more sweet food than either salty or bland food [paired t(36)=2.54, 2.73, p < 0 . 0 5 for salty and bland food, respectively]. Subjects in the luteal phase consumed significantly more sweet food than did subjects in the follicular phase, t(35) = 3.01, p
