点击显示 收起
1 From the Department of Food Science, Cook College, Rutgers University, New Brunswick, NJ
2 Supported by a Busch Biomedical Research Grant from Rutgers University. 3 Address reprint requests to BJ Tepper, Department of Food Science, Rutgers University, 65 Dudley Road, New Brunswick, NJ 08901-8520. E-mail: tepper{at}aesop.rutgers.edu.
ABSTRACT
Background: Genetic variation in sensitivity to the bitterness of 6-n-propylthiouracil (PROP) is thought to play a role in the acceptance and rejection of bitter-tasting vegetables by young children.
Objectives: We investigated the relation between the PROP bitter-taste phenotype and acceptance and consumption of vegetables by young children. We hypothesized that nontasters of PROP would give higher hedonic ratings to bitter-tasting vegetables and would consume more bitter vegetables than would tasters of PROP.
Design: Sixty-five preschool children were identified as tasters (n = 24) or nontasters (n = 41) of PROP. Children were allowed to select from among 5 types of vegetables (black olives, cucumbers, carrots, red pepper, and raw broccoli) to consume in a free-choice intake test and to give hedonic ratings to the vegetables.
Results: The nontaster children consumed more vegetables than did the taster children during the free-choice test (0.91 servings compared with 0.48 servings; P < 0.05). This difference reflected the higher consumption of the more bitter-tasting vegetables (olives, cucumber, and broccoli) by the nontaster children (P < 0.05). Only 8% of the nontaster children consumed no vegetables in the free-choice test compared with 32% of taster children (P < 0.03). The nontaster children also liked raw broccoli more than did the taster children in the hedonic test (P < 0.05).
Conclusions: The nontaster children consumed more vegetables, particularly the vegetables that were bitter tasting, than did the taster children during a free-choice intake test. These novel findings suggest that the PROP bitter-taste phenotype contributes to the development of vegetable acceptance and consumption patterns during early childhood.
Key Words: Bitter taste • genetics • 6-n-propylthiouracil • vegetable intake • children
INTRODUCTION
Consumption of fruit and vegetables by young children in the United States is low, with only 1 in 5 children consuming the recommended amount, ie, 5 servings/d (1). Although diet quality for fruit and vegetable intakes increased slightly for American children between 1977 and 1998, mean vegetable intake for 4–5-y-olds remains at 2 servings/d (2). These data are troubling because vegetables are rich in nutrients that promote optimum health and that may protect against the development of obesity, cardiovascular disease, and cancer (3-5).
Eating habits are acquired at a young age and, once established, tend to track across childhood and beyond (6, 7). Thus, it may be important to identify the factors that either promote or hinder fruit and vegetable intake during this sensitive developmental window. First and foremost, parents and other caregivers define the eating environment of their children. Parental feeding strategies such as restriction, monitoring, and pressure to eat have been shown to alter children's eating patterns but not always in the desired direction (8). Encouraging children to make healthy choices with options to choose from among different foods (a so-called "authoritative" feeding style) is associated with higher fruit and vegetable intakes (9, 10). Other work has shown that parental consumption is a strong predictor of fruit and vegetable intakes in children (11, 12). Thus, the availability of fruit and vegetables in the home and the example parents set with their own eating behavior may serve as powerful predictors of children's food habits. On the other hand, personal traits such as neophobia (the fear of trying new foods) and pickiness have been shown to reduce fruit and vegetable intakes in young children (13, 14).
Sociocultural factors notwithstanding, taste is the ultimate determinant of what children will eat (15), and preferences for fruit and vegetables were among the strongest correlates of fruit and vegetable intakes in a large study conducted in Norway (16). However, many fruit and vegetables are bitter tasting and are disliked for this reason (17). Thioureas and related chemical compounds are prevalent in cruciferous and noncruciferous vegetables and other edible plants. Variation in sensitivity to the bitter taste of thiourea compounds is a well-known genetic trait that may be a marker for individual differences in acceptance of bitter tasting vegetables (both cruciferous and noncruciferous), bitter citrus fruits, and other bitter foods (18, 19). Phenylthiocarbamide (PTC) and 6-n-propylthiouracil (PROP) are the most frequently studied of this class of compounds. Most of the population perceives PTC and PROP as bitter tasting. The remainder of the population perceives these compounds as weak, and a small number describe them as tasteless (20). The approximate distribution of tasters and nontasters among whites in the United States and Western Europe is 70% and 30%, respectively, but this distribution varies in other race and ethnic groups (21). Tasters can be further divided into medium tasters and super tasters; the latter group exhibits extreme taste sensitivity to PTC and PROP (20). A major gene controlling this trait was recently identified (22, 23).
Observations in adults suggest that PROP tasters dislike bitter and strong tasting foods (24-29), although other studies disagree with this conclusion (30-32). More consistent findings have been reported in children; nontaster children gave higher hedonic ratings than did taster children to raw broccoli (33), grapefruit-orange juice (19), and raw spinach (34). An earlier study in children showed no difference in hedonic ratings of raw broccoli between tasters and nontasters (35). However, nontaster children in the aforementioned study preferred a more bitter food (cheese) in a forced-choice preference test, whereas taster children preferred a less bitter food (milk).
PROP-tasting studies in children have typically used a variety of techniques to assess children's food acceptance, including hedonic ratings, rank-order preference procedures, verbally administered food-preference questionnaires, and parental reports of children's food preferences (33-35). However, no studies have directly examined the selection and consumption of vegetables by taster and nontaster children under free-choice conditions. We developed a quasinaturalistic, free-choice snack test that mimics the real-world experience of selecting foods from a variety of choices. In addition to the free-choice selection and intake test, the present study also used a hedonic test of tasted food and a child food-preference questionnaire completed by mothers. We hypothesized that nontaster children would like bitter vegetables more than would taster children and would select and consume more bitter vegetables when offered a choice of vegetables as a snack. An additional objective was to compare the 3 measures of acceptance and consumption to assess the concordance among measures. Finally, mothers completed the Child Feeding Questionnaire (CFQ) (36) for the assessment of the influence of parental feeding style on the children's vegetable acceptance and intake.
SUBJECTS AND METHODS
Subjects
The subjects were 65 preschool children (n = 33 girls and 32 boys) who were 3.5–4.5 y of age at the time of the study and were enrolled at the Rutgers University Nutritional Sciences Preschool, New Brunswick, NJ. The study was conducted over a 2-y period with approximately equal numbers of children participating in each study year. Parents completed a screening questionnaire designed to assess their child's general health and suitability for the study. Children who had food allergies, had dietary restrictions, had recent illnesses, or were taking medications that altered taste or food intake were excluded from participating. The Rutgers University Institutional Review Board approved the study. Parents provided written informed consent for their child's participation, and verbal assent was obtained from each child before testing.
Subject characteristics
The children's weight was measured to the nearest 0.1 kg with an electronic scale (WB-800; Best Wright, Brooklyn, NY), and their height was measured to the nearest 0.5 cm with a standard tape measure. Children were not wearing shoes during the height measurement and were instructed to stand straight against a wall while their height was being taken. Body weights were converted to weight-for-height percentiles and corresponding z scores with the use of EPI INFO 2000 (Internet: www.cdc.gov/epiinfo).
Testing location
Before testing, the investigator visited the preschool during playtime for several days to develop familiarity with the children. Children in the 4-y-old class were tested during playtime in a separate kitchen area equipped with child-sized tables and chairs. Because children in the 3-y-old class were restricted from leaving the classroom, they were tested at an isolated table in the corner of the classroom.
PROP classification
Children's PROP taster status was determined by using a procedure developed by Keller et al (33, 37). Subjects were classified as either tasters or nontasters of PROP based on their ability to taste a 0.56-mmol/L solution of PROP (6-propyl-2-thiouracil, #P3755; Aldrich Chemical Inc, Milwaukee, WI). Children were presented with a 60-mL plastic soufflé cup containing 10 mL PROP solution in spring water. After tasting the solution they were asked, "Do you taste anything?" If their response was no, they were asked whether it tasted like water. If they responded that it had no taste and tasted like water, they were classified as nontasters. If they responded that it tasted yucky or bad they were classified as tasters. If they responded that they did not know what it tasted like or provided an ambiguous response, they were retested at a later time. Facial expressions were observed during the tasting to support the verbal response and to identify any ambiguous or conflicting responses. Test-retest reliability of this method is high (r = 0.92) (33).
Food stimuli
The Nutritional Sciences Preschool exposes children to a wide range of vegetables, fruit, and other foods during daily snack time. The food stimuli chosen for this study were familiar foods that the children had previously been offered at the preschool. The vegetables included raw broccoli, cucumber, pitted ripe black olives, minicarrots, and red bell pepper. All foods were purchased fresh from a local grocery store, except for the olives, which were canned (Shop-Rite Brand, Rahway, NJ). The vegetables fell into 2 general taste categories: bitter (black olives, broccoli, and cucumber) and nonbitter (carrots and red peppers). These categories are supported by a factor analysis of vegetable preferences by adults, which identified bitter (turnips and greens) and nonbitter (carrots) subgroups for tasters. The model for nontasters did not separate vegetables by bitter taste, which suggests that nontasters do not view these foods as bitter (29). Technically, olives are the fruit of the tree Olea europaea. Because of their high fat content, olives are typically considered a fat rather than a vegetable (38). However, culinary uses of olives include their addition to salads and vegetable trays, and the children were previously offered olives in this context at the preschool.
Free-choice selection and intake test
Preweighed samples were placed in 120-mL clear-plastic soufflé cups that were arranged in individual rows on a serving table. The samples were offered in 25-g portions, except for olives, which were offered in a 17-g portion (3 items). The children were instructed to choose their own food, place it onto their own tray, and to not share their food items with the other children. They were allowed to select as many food cups as they wanted, and they could eat as much or as little as they wanted. They were not encouraged or praised for their selections. Children were tested in groups of 2–3 to simulate their normal snack time. The order of presentation of the foods was randomized across groups of children to minimize presentation bias. The children were read stories to hold their attention and to minimize talking and distractions. The investigator monitored the snack session to ensure adherence to the study protocol. Snack tests were conducted on separate days from the hedonic tests.
After the session, the weight consumed (in g) of each food item was calculated by subtracting its postconsumption weight from its preconsumption weight. Gram weights were then converted to servings by using the 1999 US Department of Agriculture Food Guide Pyramid serving sizes for 3–6 y-old children (39). The following conversions were used: one serving (equivalent to 0.5 cups coarsely chopped raw vegetables) was 75 g for cucumber and red pepper, 81 g for carrots, and 44 g for broccoli. Three olives constituted one serving and weighed 17 g.
Hedonic test
Vegetable samples were prepared immediately before testing. Samples were presented in 10-g portions and served in 60-mL clear-plastic soufflé cups. They were presented one at a time in randomized order to each child. In between tasting each sample, the child was asked to rinse with water. Liking was assessed according to a 5-point facial hedonic scale, which is a valid instrument for obtaining liking data from children aged 47–59 mo (40). The scale categories range from "super good" (smiling face; scale value of 5) to "super bad" (frowning face; scale value of 1). The scale was verbally explained to each child and his or her comprehension of the task was evaluated by using a standard role-playing exercise. During this exercise, the child was asked to name his or her favorite food and describe how it tastes. The researcher pointed to the "super good" facial descriptor and explained that the child's favorite food would be rated as that face. The same procedure was completed for the child's least favorite food.
A poster board with colorful pictures of each vegetable was used to help maintain the interest of the child during testing. After the child tasted and rated a vegetable using the facial hedonic scale, he or she was allowed to select a sticker and place it onto the picture of the vegetable item that he or she had tasted. The poster board methodology was used in our previous studies (33) and was adapted from a procedure developed by Kimmel et al (41). The entire session took 10–15 min for each child to complete.
Reported food preferences and child-feeding questionnaire
The parent who was primarily responsible for the child's food preparation (the mother in most cases) completed a food-preference questionnaire for her child and the CFQ (36). The child food-preference questionnaire consists of 42 food items that vary in bitter, sweet, fat, and spicy taste qualities. The questionnaire was a shortened version of an instrument used previously (33). The vegetables tested in this study were also embedded in this questionnaire. Responses were collected with a 9-point category liking scale that ranged from "like extremely" to "dislike extremely." Two additional categories labeled "never tried" and "do not eat" were also provided. These boxes were checked if the child had never tried the food or if the child did not eat an item because of religious reasons or a food allergy. The CFQ is a 31-item, validated instrument designed to measure parental attitudes, perceptions, and practices regarding child feeding. The CFQ has 7 subscales for perceived responsibility, monitoring, restriction, pressure to eat, perceived parent weight, perceived child weight, and concern about child weight.
Statistical analyses
Analysis of variance (ANOVA) was used to examine the influence of PROP taster status on the children's hedonic ratings, consumption of vegetables in the free-choice snack test, maternal-reported child food preferences, and CFQ responses. Vegetable consumption was analyzed for number of servings consumed of each item, total number of vegetable servings, and total number of bitter and nonbitter vegetable servings (as described above). Post hoc comparisons were conducted by using Tukey's test. Pearson's correlations examined relations among the outcome measures. Associations between child feeding practices and the outcome measures were examined by using Pearson's correlations and partial correlations to control for the influence of PROP status in the analysis. A chi-square test was used to determine the proportion of taster and nontaster children who declined to eat any vegetables during the intake test. All statistical analyses were conducted by using PC-SAS (version 10.0; SAS Institute, Cary, NC) with = 0.05.
RESULTS
Subject characteristics
All 65 children completed the study. The ethnicity of the children was 79.4% white (non-Hispanic), 11.1% Asian, 3.3% black, 3.3% Hispanic, and 3.3% other. Ninety-two percent of the children lived in households that met or exceeded the median annual household income for the state of New Jersey (42). Most of the mothers were college-educated.
The percentage of PROP taster and nontaster children was 63% and 37%, respectively. PROP status did not vary with sex. These data agree with previous observations in the literature for young children (33, 37, 43). Among girls, there were 21 tasters and 12 nontasters. Among boys, there were 20 tasters and 12 nontasters. The children were mostly lean and had a BMI-percentile-for-age of 51.2%. BMI%-for-age did not vary with taster status.
Free-choice intake test
As shown in Table 1, intakes of individual vegetables were low, and there were no significant differences between taster groups for the consumption of individual vegetables. As shown in Figure 1 A, significant differences were observed when the vegetables were grouped into bitter (cucumber, broccoli, and olive) and nonbitter (carrots and red pepper) categories. The nontaster children consumed more servings of bitter vegetables than did the taster children (P < 0.01). The nontaster children also consumed more servings of bitter vegetables than of nonbitter vegetable (P < 0.01). Consumption of nonbitter vegetables was low and did not vary between taster groups. Overall, the nontaster children consumed slightly less than 1 vegetable serving (0.91 serving) compared with the taster children, who consumed 0.48 of a vegetable serving (P < 0.02). Black olives were excluded, and the analysis was repeated (see Figure 1B). Intakes were lower when olives were excluded, but the pattern of results was the same (P < 0.05). The nontaster children consumed more bitter vegetables than did the taster children (P < 0.01). The nontaster children also consumed more vegetables overall (0.52 servings) than did the taster children (0.26 servings) (P < 0.05). Chi-square analysis showed that 8.3% of the nontaster children selected no vegetables during the test compared with 32.5% of the taster children (P 0.03).
View this table:
TABLE 1. Consumption of individual vegetables (in partial servings) by 6-n-propylthiouracil nontaster and taster children during the free-choice snack test1
FIGURE 1.. Mean (±SE) number of vegetable servings consumed by 6-n-propylthiouracil nontaster (; n = 24) and taster (; n = 41) children during the free-choice snack test. A: Vegetables were grouped into bitter (black olives, cucumber, and raw broccoli) and nonbitter (red peppers and carrots) categories. Tukey's test: *P < 0.05, **P < 0.01. There was a significant taster x vegetable type interaction by ANOVA (P < 0.01) and a main taster effect for total vegetable servings (P < 0.05). B: The same analysis as in A was conducted with black olives removed. Tukey's test: *P < 0.05. There was a significant taster x vegetable type interaction by ANOVA (P < 0.05).
Children's hedonic ratings
Liking ratings for the tasted vegetables are shown in Figure 2. The results showed that the nontaster children liked raw broccoli more than did the taster children (P < 0.05). There were no significant differences in the liking of the other vegetables, although the ratings for black olives and cucumber tended to be higher for nontasters than for tasters.
FIGURE 2.. Mean (±SE) liking ratings in 6-n-propylthiouracil nontaster (; n = 24) and taster (; n = 41) children during the hedonic test. *Significant difference between the nontaster and taster children by ANOVA (P < 0.05). 1 = dislike extremely; 5 = like extremely.
Maternal-reported child food preferences
The mean reported liking ratings for selected vegetables from the child food-preference questionnaire that matched those examined in the laboratory hedonic tests are shown in Table 2. There were no taster-group differences in the liking ratings for these vegetables as reported by mothers. The percentage of children who had not tried these vegetables did not vary by PROP taste sensitivity.
View this table:
TABLE 2. Maternal-reported child liking of selected vegetables and reported percentages of children who never tried these vegetables, categorized by 6-n-propylthiouracil nontaster and taster status
Correlations between the liking and intake measures
As shown in Table 3, the children's hedonic ratings of the vegetables were correlated with maternal reported child preferences for those same vegetables (r = 0.41–0.86, P 0.01). Children's hedonic ratings of olives, cucumbers, and red peppers were modestly correlated with the consumption of those same items (r = 0.42–0.56, P 0.01). Significant, but weak, associations were observed between children's consumption and maternal reports of children's liking of black olives and cucumber (r = 0.35–0.54, P 0.01).
View this table:
TABLE 3. Pearson's correlation coefficients relating measures of child liking and consumption of individual vegetables1
Child Feeding Questionnaire
CFQ subscale scores are shown in Table 4. No differences in subscale scores were observed as a function of children's PROP taster phenotype. Pearson's correlations showed a few isolated, significant associations between CFQ subscale scores and the liking and intake measures. However, the r values were uniformly low (r = 0.26–0.30, P < 0.05), and, consequently, these relations were not considered meaningful.
View this table:
TABLE 4. Child Feeding Questionnaire subscale scores as a function of the children's 6-n propylthiouracil taster status1
Partial correlations were computed between CFQ scores and children's hedonic ratings and the number of vegetable servings consumed while controlling for the children's PROP taster status. The pressure to eat subscale showed modest negative correlations with liking of the bitter vegetables, including raw broccoli (r = –0.40, P < 0.02), black olives (r = –0.37, P < 0.04), and cucumbers (r = –0.36, P < 0.04). No other significant correlations were observed between CFQ subscale scores and any other outcome measures.
DISCUSSION
The results of this study showed that the nontaster children consumed more bitter vegetables and more vegetables overall during a free-choice intake test than did the taster children. The nontaster children consumed almost one serving of vegetables, whereas the taster children consumed approximately 0.5 servings of vegetables. These data suggest that young children will freely consume vegetables as a snack when such foods are familiar and made accessible to them. But the types and amounts of vegetables consumed reflect the children's genetic taste predispositions. To our knowledge, this is the first study to report differences in short-term vegetable intake by children classified by PROP bitter-taste phenotype. If these differences persisted across meals and days, they would be nutritionally important to the diets of children. Future studies will have to address this critical question.
It is not known whether olives are specifically bitter tasting to young children. The black olives used in the present study have complex sensory characteristics, including low levels of bitterness and brininess, savoriness, and a fatty mouthfeel (44). It is possible that the higher consumption of olives by the nontaster children reflects their greater liking of the fattiness of olives. Previous studies have shown an association between the nontaster phenotype and increased preferences for other high-fat foods, such as full-fat milk, high-fat salad dressing, and high-fat meats and cheeses (33, 45, 46). When olives were excluded from the current analysis, overall consumption dropped by half, but the pattern of intake was the same, ie, the nontaster children consumed more bitter vegetables and more vegetables overall than did the taster children. Thus, the influence of bitter-taste phenotype on the pattern of children's vegetable intake remained constant regardless of the presence or absence of the olives.
Another finding in our study was that the nontaster children gave raw broccoli higher liking ratings than did the taster children in the hedonic test, which confirmed the earlier observations of Keller et al (33). Our data conflict with other studies that reported no difference in children's liking of broccoli between the PROP taster groups (35). However, another study did find that nontaster children liked another bitter vegetable, spinach, more than did taster children (34).
Even though the nontaster children liked raw broccoli more than did the taster children, the nontaster children did not consume more broccoli in the selection test. One explanation for this inconsistency could be that hedonic ratings provide a valid measure of the degree of liking of a single food item, but they might not predict food choices when different alternatives are available (47). Thus, these 2 measures capture different information about food acceptance behavior. Because the free-choice test used in the present study directly measures food consumption, it could be considered a more nutritionally relevant measure of food acceptability.
Previous studies have shown that mothers reliably report the food preferences of their children (48), and we observed relatively strong correlations between the children's hedonic ratings of tasted vegetable samples and maternal reports of children's preferences for those same vegetables (Table 2). As for the children's hedonic ratings, maternal reported food preferences were not good indicators of the children's vegetable selections and intakes.
The present study showed only weak associations between CFQ responses and the various outcome measures. Control for taster status by using partial correlations improved the strength of some of these associations. Specifically, parental pressure to eat was negatively associated with the liking of bitter vegetables. Our data agree with those of previous studies that reported a negative association between parental pressure to eat and fruit and vegetable intakes in young girls (11, 14).
Our study population was somewhat ideal because the children were very familiar with different types of vegetables and were frequently offered vegetables as a snack at the preschool. Thus, common barriers to vegetable intake, such as neophobia and limited accessibility to vegetables, were virtually absent in the children we studied. This could explain why the hedonic ratings for minicarrots were high but intakes of minicarrots were low when children were given a choice of vegetables to consume.
The PROP screening method we used has high test-retest reliability (r = 0.92) (33) and was used in previous studies from our laboratory yielding results similar to those obtained here, ie, we identified 36% (33) and 34% (37) nontasters in both studies, respectively. TAS2R38, the gene that controls PTC taste sensitivity, was identified after the current study began (22). Thus, genotyping for this trait was not available at the time of this study. However, 37% of the children studied were identified as nontasters by PROP phenotype, which was virtually identical to the percentage (36.5%) of white children identified by TAS2R38 genotype as nontasters by Mennella et al (43). The study by Mennella et al (43) also included a high percentage of Hispanic children who constitute a different admixture for this trait than do whites (21). Our data are consistent with the notion that PTC and PROP tasting may be controlled by distinct but overlapping genes (23).
Finally, our study sample was relatively small and not demographically diverse. The extent to which our findings will generalize to children in the population at large is unknown and needs to be investigated. Nevertheless, even under the close-to-ideal conditions observed in the present study, nearly one-third of the taster children selected no vegetables in the snack test compared with only 8% of the nontaster children. These data suggest that the PROP bitter-taste phenotype plays a substantive role in shaping patterns of vegetable acceptance and rejection by young children and warrants further study.
ACKNOWLEDGMENTS
We thank Harriet and John Worobey for permission to recruit children at the Nutritional Sciences Preschool and Christina Tani for her assistance.
KIB was responsible for subject recruitment and testing and the data analysis and cowrote the manuscript. BJT was responsible for the experimental design, data interpretation, and writing of the manuscript. Neither author had any personal or financial conflict of interest.
REFERENCES