点击显示 收起
1 From the Department of Pediatrics, University of Colorado Health Sciences Center, Denver, CO
2 Supported by NICHD HD33511 and a grant from The National Dairy Council. 3 Reprints not available. Address correspondence to SL Johnson, University of Colorado Health Sciences Center, Department of Pediatrics C225, The Center for Human Nutrition, 4200 E 9th Avenue, Denver, CO 80231. E-mail: susan.johnson{at}uchsc.edu.
ABSTRACT
Background:Preschool-aged children consistently self-regulate energy intake by adjusting food intake to reflect the energy density of the diet. However, only select adults demonstrate self-regulation skills related to energy density cues.
Objective:The objective was to examine the developmental characteristics of non-Hispanic white and Hispanic children’s self-regulation of energy intake and to determine whether individual differences in the precision of food intake regulation were related to the children’s anthropometric measures.
Design:Three hundred forty-two 5–12-y-old children were recruited from 2 schools, one school that enrolled predominantly non-Hispanic white children (n = 148) and another school in a Hispanic neighborhood (n = 194). The children ate 2-part meals, which consisted of a preload and a main meal. The preloads were fruit-flavored drinks that were either high (625 kJ) or low (12.5 kJ) in energy density. The children’s food intake at the meal was measured and energy intake was estimated. An eating Compensation Index (COMPX) score, which reflected the children’s ability to regulate energy intake, was used to predict the children’s adiposity.
Results:Both non-Hispanic white and Hispanic children showed evidence of incomplete compensation, with a mean (±SEM) COMPX score of 48.6 ± 6.4%. No significant ethnicity or sex differences in the children’s overall COMPX scores were observed; however, the children’s COMPX scores decreased with age (P < 0.05). Maternal body mass index was the best predictor of the children’s weight status.
Conclusion:Children aged 5–12 y show individual variation in their responsiveness to energy density cues; the responsiveness declines with increasing age.
Key Words: Self-regulation • compensation • adiposity • children • ethnicity • Hispanic
INTRODUCTION
Under laboratory conditions, young children exhibit the ability to self-regulate energy intake by adjusting energy intake to reflect changes in the energy density of the diet. The first evidence of such regulation was provided by Fomon et al (1), who studied regulation of energy intake in infants. Infants adjusted their total energy intake to reflect the energy density of the formula offered to them; they consumed a consistent amount of energy despite differences in the energy content of the formula offered. Additional evidence provided by Birch et al (2, 3) suggests that preschoolers can consistently adjust their energy intake to reflect the energy density of the diet during short-term single-meal protocols. Preschoolers show this ability to self-regulate energy intake in response to energy density manipulations achieved by changes in both carbohydrate and fat content (3, 4). Previously, we studied variation in young children’s energy intake control and reported that preschoolers, although showing evidence of compensation as a group, vary in their individual ability to regulate energy intake (5). This variation in self-regulation was related to the children’s degree of adiposity: children with greater adiposity showed less evidence of self-regulation of energy intake than did those with lower adiposity.
The evidence in support for self-regulation of energy intake in adults is equivocal. Rolls et al (6-8) have consistently reported that young, lean, non-restrained, non-dieting men and women respond to changes in energy content of the diet by adjusting consumption to reflect the energy content of the foods offered. In subsequent research, Rolls et al (7-10, 11) focused on the effects of energy density on adults’ regulation of food intake and noted that energy density and food volume directly influence energy intake. However, contrasting findings have been reported by other investigators with respect to self-regulation characteristics for men and women—particularly for obese and dieting subjects (12-14).
Little is known regarding the developmental aspects of self-regulation of energy intake. Although preschoolers exhibit self-regulation, many adults do not, and the self-regulation skills of girls and boys aged between 6 and 17 y have not been well-described. During this transitional period, many biological and social events occur, including increased growth, puberty, and changes in the effect and importance of peer and family influences (15).
Furthermore, self-regulation characteristics that have been reported relate primarily to non-Hispanic white subjects. The prevalence of overweight in Hispanic adults is 73%, and 39% of Hispanic children aged 6–11 y are at risk of overweight [body mass index (BMI) > 85th percentile] (16). This high prevalence of childhood obesity has been suggested to be due, in part, to differences in food intake (17-19), physical activity levels (20), and sedentary pastimes (21, 22). The relation between adiposity and socioeconomic factors (23, 24) and acculturation level (24-26) have yielded discrepant findings.
Our study examined whether differences in elementary school children’s self-regulation characteristics exist and whether individual differences may account for some of the variance in adiposity between non-Hispanic white and Hispanic children We used a modified single-meal protocol developed previously for preschoolers and collected data regarding the children’s age, sex, and ethnicity. Our focus was to document the developmental course of non-Hispanic white and Hispanic 5–11-y-old girls’ and boys’ Compensation Index (COMPX), which is an individual measure of children’s ability to self-regulate energy intake. We related the children’s COMPX scores to their weight status (BMI z scores) and to parent characteristics (education, Hollingshead Index, acculturation, and weight status). We expected to find that children with a higher COMPX score (ie, better evidence of self-regulation) would be leaner and that the children’s ability to self-regulate (as indicated by their COMPX score) would decrease with age. Last, we hypothesized that Hispanic children would have greater adiposity and would consequently show less evidence of self-regulation than would non-Hispanic white children.
SUBJECTS AND METHODS
Overview
Children were recruited from 2 Denver area elementary schools. The first cohort was predominantly composed of non-Hispanic white children, whereas the second cohort consisted of children who were primarily of Hispanic (Mexican-American) origin.
Children participated in one pair of caloric compensation trials with each trial consisting of a preload followed by a main meal. Preloads were fruit-flavored drinks that were matched for taste and other hedonic characteristics but differed in carbohydrate and, therefore, energy content. Shortly after consuming the preload, the children were invited to eat an ad libitum lunch that was weighed to determine the children’s food intake. The children were said to compensate if they adjusted their intake to reflect the energy content of the drinks (eg, by eating less for lunch after consuming the high-energy drink). The percentage of compensation was calculated for each child and subsequently used as the COMPX score of the child’s responsiveness to energy density cues. Anthropometric measures were obtained and correlated with the children’s COMPX score to determine whether any systematic relations existed between the children’s adiposity and their COMPX score.
Parents provided information regarding demographic characteristics, their self-reported height and weight, and, for our Hispanic cohort, acculturation to further characterize the sample. The protocol was reviewed and approved by the Colorado Multiple Institutional Review Board, and parents gave informed consent for their child’s participation. The children were asked to assent each day they participated in the study.
Subjects
A total of 342 children aged 5–12 y were recruited for participation in the study. Complete data (consumption and anthropometric measures) were collected for 262 children. In cohort 1, 38 children were dropped from the analyses because of an inability to finish the preload (n = 22), absence from school on consumption days (n = 10), or absence during days of anthropometric assessments (n = 6). In cohort 2, ie, our Hispanic cohort, 53 children did not complete the protocol because of absence from school on consumption days (n = 23) or on the days of anthropometric assessment (n = 8), an inability to consume the entire preload (n = 7), or because their families moved during the school year (n = 15). The reasons for the higher rate of attrition in the second cohort are related to the socioeconomic status (SES) of their families: some families were economically unstable and of indeterminant immigration status and children disappeared from school enrollment midyear. The children who did not complete the protocol did not differ significantly (within their cohort) with respect to family characteristics from those who completed both consumption trials. Because anthropometric data were not collected until after the consumption trials were completed (to avoid confounding children’s eating behavior with the measurement of their weight), we do not know whether the anthropometric measures of the children who did not complete the protocol differed significantly from those who did complete the protocol.
Cohort 1
Families were recruited from the Stanley British Primary School in Denver. A total of 121 children [53 boys and 68 girls; mean (±SEM) age: 99.7 ± 1.8 mo], completed compensation trials and anthropometric assessments. The parents were in their mid-40’s and most were non-Hispanic white (2 Asian, 5 Hispanic, and 13 African-American children participated; see Table 1 for demographic data). Most of the children came from dual parent households (88%).
View this table:
TABLE 1. Parent demographics1
Cohort 2
Families were recruited from the Columbian Elementary School in Denver. A total of 141 children (73 boys and 68 girls; mean age: 101.7 ± 1.3 mo) completed compensation trials and anthropometric assessments. The parents were in their mid-30’s and most were Hispanic (1 African-American and 7 non-Hispanic white children participated; see Table 1). Fifteen percent of mothers and 35% of fathers reported that they did not speak English very well or did not speak it at all. Again, most of the households were dual parent households (89%).
Materials for compensation trials
The children consumed 2 juice preloads, which differed in carbohydrate content and energy density but not in flavor or appearance. The preloads were prepared as previously described (5). One serving of the high-energy preload (163 g) contained 625 kJ (150 kcal)/serving, whereas the low-energy preload contained 13 kJ (3 kcal)/serving. The drinks were placed in clear plastic cups and chilled to 6 °C.
We consulted both the parents and the school cafeteria menus to choose the foods that were offered for the ad libitum lunch. Foods needed to be familiar, easy to prepare and transport, and well-liked by children from both schools. The menu for all ad libitum lunches included macaroni and cheese (100g; Kraft-General Foods, Waukegan, IL), steamed broccoli (62g), carrot sticks (31g), seedless grapes (113g), cheese sticks (31g Frigo Cheese Heads; Stella Foods Inc, Green Bay, WI), graham crackers (25g Cinnamon Crisp; Keebler Company, Elmhurst, IL), and 2% milk (256g; Robinson Dairy, Denver, CO). The total energy of the foods initially served to each child was 2646 kJ (633 kcal), as calculated from the manufacturer’s information. At their request, additional servings of any menu item were provided to the children. This menu was used throughout the study after both the high- and low-energy preloads.
Compensation Index trials
Each child participated in one pair of consumption trials where a trial pair consisted of 2 test days. On the first day of the trial, the children were asked to drink the high- or low-calorie juice as a preload. After 30 min, the children were offered lunch foods from which they could self-select and eat ad libitum. That is, the children were free to choose to eat any or all of the foods in the order that they desired. After 2 d, the children received the other version of the drink preload and again consumed a self-selected lunch. During the meal, the children were asked whether they had consumed breakfast (yes or no) and to tell the researcher what they had consumed to validate their answer. No attempt was made to quantify the amounts consumed. Rather, our purpose was to account for breakfast consumption (yes or no) in our analyses. The children were assigned by sex to the order of presentation of the preload, and investigators and classroom personnel were blinded to the experimental condition.
Food intake was measured by pre- and postweighing all foods offered. Energy intake of foods consumed was calculated by using the manufacturer’s information. A measure of the children’s responsiveness to the caloric content of the juice preloads was derived from the extent to which they adjusted their ad libitum lunch intake. The COMPX score, which indicates the percentage of compensation, was used to characterize the children on a continuum from negative compensation (ie, consuming more at lunch after drinking the high-energy drink) to overcompensation (ie, consuming less at lunch after the high-energy drink.) This COMPX score was calculated for each child by using the following formula:
RESULTS
Subject characteristics
Parents
Demographic information, including parents’ education, SES, and parent weight status is shown in Table 1. In general, parents from cohort 1 were predominantly non-Hispanic white, highly educated, of upper SES, and lean (only 9% of mothers and 6% of fathers had a BMI > 30). Mothers and fathers in cohort 2 were Hispanic had not, on average, completed high school, and were of a lower SES level. Parents from cohort 2 were overweight; 31% of mothers and 20% of fathers were obese (BMI > 30). The mean (±SEM) acculturation score (possible range from 13 to 39 with a high score indicating lower acculturation) of mothers and fathers was 26.8 ± 0.7 and 25.6 ± 0.8, respectively. No significant differences in acculturation levels of mothers and fathers were observed. Some statistically significant differences were observed between mothers and fathers within the cohorts, but the differences were small (Table 1).
Children
The children’s age, anthropometric measures, and consumption characteristics are reported in Table 2. The children of the entire sample were aged 5–11 y and were equally represented across 2 ethnicities (non-Hispanic white and Hispanic) and sex. The children in cohort 1 (predominantly non-Hispanic white) were mostly normal weight; 11% of boys (n = 6) and 4% of girls (n = 3) were above the 95th percentile for age- and sex-adjusted BMIs. Similar to the differences shown by their parents, more children in cohort 2 (predominantly Hispanic) were overweight. Twenty percent of Hispanic boys (n = 17) and 17% of Hispanic girls (n = 12) were above the 95th percentile for age- and sex-adjusted BMI.
View this table:
TABLE 2. Children’s body-composition measures and food intake measures1
Children’s anthropometric data
Non-Hispanic white children in cohort 1 were significantly leaner than were Hispanic children in cohort 2. The BMIs for all children were transformed into BMI z scores so that age- and sex-adjusted comparisons could be made. When examined by ANOVA, the BMI z scores of all children from cohort 2 were significantly higher than those of the cohort 1 children (P < 0.05), and the boys’ (cohorts 1 and 2) BMI z scores were significantly higher than those of the girls (P < 0.05). When ethnicity was added into the model, cohort was no longer a significant contributor to the model, which suggests that ethnicity was a mediator of the cohort differences in BMI z scores.
Maternal BMI and education level were significantly correlated with the children’s BMI z scores (R = 0.30 and R = –0.14, respectively; P < 0.05 for both). No other parent variables related significantly to the children’s BMI z scores.
Children’s eating and compensation patterns
The amount of energy consumed at lunch, the energy per kilogram of body weight, and the children’s mean COMPX scores are reported in Table 2. The children showed limited evidence of self-regulation, with a mean (±SEM) COMPX score (for children across all groups) of 48.6 ± 6.4%. No significant differences in the children’s COMPX scores that were related to children’s cohort, sex, breakfast consumption, parent education, occupation, acculturation, or parent BMI were observed. The individual variability in children’s COMPX scores was large, with some children showing no evidence of self-regulation (COMPX < 0) and others showing a hyperresponsiveness to postingestive cues associated with the drink preloads (COMPX > 100%).
Modest, but significant, effects of age were observed on children’s COMPX scores (R2 = 2%, P < 0.05; Figure 1); older children had lower mean COMPX scores. No significant cohort effects were noted in this relation, and no significant interaction effects were noted.
FIGURE 1.. Mean (±SEM) Compensation Index (COMPX) scores for girls () and boys (). Data were analyzed by a 2-factor ANOVA with main effects for sex and age and a sex x age interaction. A significant age effect (P < 0.05) was noted. Sample sizes (n) are noted beneath each bar.
When the total average energy intake for lunch (meal intake only) was compared across cohorts, sexes, and ages of the children, significant main effects for age, as well as a cohort-by-sex interaction, were found. Boys in cohort 1 consumed significantly more total energy than any other group (P < 0.05). As would be expected, older children consumed more energy at lunch than did younger children (P < 0.001).
When the data for children’s energy intake were adjusted for body weight in kg (see Table 2), the energy intake per kg declined significantly with age (P < 0.001). The boys in cohort 1 consumed more energy, adjusted for body weight, than did any other group (P < 0.001 for cohort-by-sex effect). When ethnicity was added into the model, the cohort effect was no longer significant, suggesting a mediating effect of ethnicity. Parent variables were again not significantly related to the children’s energy intake at the meal.
Relations between the children’s COMPX scores, age, and adiposity
Multiple linear regression was used to test our hypotheses regarding the relations between the children’s COMPX scores and their anthropometric measures. We tested models in which the children’s COMPX scores were used to predict their BMI z scores. These models also included age, cohort, sex, ethnicity, maternal education, and maternal BMI as predictors. Children’s COMPX scores were not predictive of their BMI z scores. The model that best predicted the children’s BMI z scores was based on the children’s age (R2 = 20%), the children’s average energy intake at lunch (R2 = 30%), the children’s energy intake adjusted for bodyweight (in kcal/kg; R2 = 7%), and the mothers’ BMIs (R2 = 4%),
DISCUSSION
The present study provides an important opportunity to characterize the developmental aspects of the responsiveness of elementary school children to internal signals of hunger and satiety. As shown by the results of the present single-meal protocol, 5–11-y-old children continued to show some evidence of caloric compensation. That is, when the energy density of a drink preload was changed, the children responded by altering food intake at a subsequent meal (eg, by consuming more energy at lunch after a low-energy preload). The mean (±SEM) compensation score of these children (48.6 ± 6.4%) was comparable to the COMPX score of preschoolers, which was reported in a similarly designed study (32). Under controlled laboratory conditions, elementary school-aged children are capable of self-regulating energy intake (at least in part) when the setting promotes autonomy and responsiveness to internal cues of hunger and satiety. One factor that may have influenced the degree to which children compensated was the macronutrient composition of the preload. The preloads used in the present study were composed of high-glycemic carbohydrate. The responsiveness of children to a preload of a different nutrient composition (eg, mixed with protein, fat, or both) could produce a difference in the degree to which they compensate. The preload in our study was designed to approximate the common occurrence of children being offered a sweet beverage (juice, juice drinks, or carbonated beverages) shortly before a meal. The results from our study are also consistent with those from a recent study conducted in 6–9-y-old children in Scotland (33). In that study, the authors reported that children (n = 74) made partial adjustments for energy consumed from low- and high-energy preloads and that younger children compensated better than did older children (an average COMPX score of 40% and 32%, respectively).
The present study provides important information about food intake regulation in Hispanic children and is, to our knowledge, the first study (in children or adults) to examine self-regulation in Hispanic persons. Contrary to our expectations, no significant differences in mean COMPX scores were observed between Hispanic (cohort 2) and non-Hispanic white (cohort 1) children, despite the fact that Hispanic children had greater BMI z scores. The children’s COMPX scores were not significantly related to any parent measure, which suggests that SES (as measured by education and occupation) and acculturation are not mediators of the children’s potential to self-regulate energy intake.
The children’s COMPX scores did not predict adiposity in the present sample of children, and the responsiveness to internal cues was not significantly different between non-Hispanic white and Hispanic children. Therefore, it appears that the capability to self-regulate energy intake is present in 5–11-y-old non-Hispanic white and Hispanic children and that other aspects of the environment (eg, food availability and selection, reduced opportunities for physical activity, or both) either diminish or counteract a child’s ability to maintain a healthy weight. Again, our findings are consistent with recent data reported by other investigators that suggested children’s COMPX scores were not significantly related to children’s BMIs (33, 34).
Additionally, no significant sex or sex-by-ethnicity differences in children’s mean COMPX scores were observed. Previously, we reported that preschool-aged boys are more accurate in responding to physiologic cues than are preschool-aged girls (5). It is unknown whether the age-related decline reported here for COMPX scores is a normal developmental trend, or whether it has its origins in the environment (eg, media, dieting cultures, familial environments with respect to food and eating, etc.). Because some lean adults have shown evidence of caloric compensation in similar single-meal protocols, the ability to respond to physiologic feedback from food could become "unlearned" in certain obesigenic environments that constantly promote overeating (35). We cannot determine the origins of the age-related decline in compensation from the present study.
In contrast to previously reported findings, the children’s COMPX scores did not predict adiposity, as measured by BMI z scores. That COMPX was not predictive of adiposity may relate to the importance and power of external environmental cues in overriding responsiveness to internal physiologic cues. Significant between-group differences in the amount of energy consumed at lunch were observed: non-Hispanic white boys consumed more energy and more energy/kg of bodyweight than did any other subgroup. The finding that non-Hispanic white boys consumed more energy, yet were leaner than Hispanic boys, may be accounted for by individual differences in physical activity, energy metabolism, or both. Again, these factors were not measured in our study.
Maternal BMI was a significant predictor of the children’s BMI z scores, and this finding is consistent with a growing amount of literature that links maternal weight status with children’s weight outcome (21, 36-39). When maternal BMI was added into the model, cohort and ethnicity effects were no longer significant, which indicates that maternal BMI is a mediating factor for the cohort effects that were observed in our sample. Other maternal factors did not significantly relate to the children’s adiposity, and paternal factors, including paternal BMI, were not significantly associated with the children’s weight status. Therefore, the differences that we noted in the children’s weight outcome between the cohorts may be related to maternal factors that affect the mothers’ weight outcome, such as food choice, activity patterns, child-feeding patterns, or other environmental factors. The finding that paternal indicators did not significantly relate to the children’s self-regulation or weight outcome is consistent with the recent findings of a longitudinal study of children at risk for obesity (40).
Possible limitations of our study include that children from the predominantly Hispanic cohort 2 were from families that varied greatly in their degree of acculturation and economic stability. Although we asked children if they had consumed breakfast and analyzed for the effects of missing a breakfast, we were not always entirely confident that the children told us when they had not eaten. The children who often experienced enduring hunger may have reacted differently to an eating environment in which no food restrictions were imposed.
Another limitation of the study relates to lack of data regarding the children’s pubertal development. The schools that participated in the research would not agree to physical examinations and, from past efforts, gathering self-reported data from children on pubertal status is extremely difficult and prone to inaccuracies. Therefore, we could not determine whether any of the age effects were related to differences in Tanner stages. Given that most adults do not show evidence of caloric compensation and that children in the present sample showed an age-related decline in compensation, an important question is whether the ability to self-regulate energy intake changes during puberty when hormonal shifts support accelerated growth and fat deposition. Future studies should be directed at older children and should include these variables.
In conclusion, our findings suggest that school-aged children, regardless of ethnicity or weight outcome, can be partially responsive to energy density cues, and make incomplete adjustments, albeit with age-related decreases, for acute changes in energy density of diets. Previously published findings from our laboratory suggest that interventions that focus on helping preschool-aged children attend to hunger and satiety improved the children’s responsiveness to these cues (32). We suggest that future interventions to prevent excess weight gain should include messages related to responsiveness to internal cues.
ACKNOWLEDGMENTS
We gratefully acknowledge the assistance of the teachers, principals, children and families of The Stanley British Primary School and Columbian Elementary School in Denver. We also thank the student interns who helped with data collection and management, most specifically Rebecca Sawyer, Cindy Cummings, and Teresa Guistino. Last we thank Gary Grunwald, our GCRC statistician, for his expert consultation.
SLJ was the Primary Investigator and received the grants that supported the present work. LAT-H coordinated the study, collected data, and was responsible for data management. Both authors participated in data interpretation and manuscript preparation. Neither of the authors had any personal or financial conflicts of interest.
REFERENCES