To discover whether eating behaviour traits show continuity and stability over childhood.
Mothers of 428 twin children from the Twins Early Development Study participated in a study of eating and weight in 1999 when the children were 4 years old. Families were contacted again in 2006 when the children were aged 10 years, with complete data on 322 children; a response rate of 75%. At both times, mothers completed the Child Eating Behaviour Questionnaire (CEBQ) for each child. Continuity was assessed with correlations between scores at the two time points, and stability by changes in mean scores over time.
For all CEBQ subscales, correlations between the two time points were highly significant (P-values <0.001). For satiety responsiveness, slowness in eating, food responsiveness, enjoyment of food, emotional overeating and food fussiness, correlations ranged from r=0.44 to 0.55, with lower continuity for emotional undereating (r=0.29). Over time, satiety responsiveness, slowness in eating, food fussiness, and emotional undereating decreased, while food responsiveness, enjoyment of food and emotional overeating increased.
Eating behaviours, including those associated with a tendency to overeat, emerge early in the developmental pathway and show levels of individual continuity comparable to stable personality traits. Appetitive traits related to higher satiety tended to decrease with maturation, while those associated with food responsiveness tended to increase. This pattern is consistent with strong tracking of body mass index alongside a progressive increase in the risk of obesity.
There is increasing evidence that adiposity ‘tracks’ across the life course, that is, fatter children mature into fatter adolescents and fatter adults, while thinner children tend to stay thinner adults (Serdula et al., 1993; Whitaker et al., 1997; Wang et al., 2000; Steinberger et al., 2001). Energy balance is part of the process that mediates adiposity and therefore, energy intake and expenditure might also be expected to show some degree of individual continuity (tracking). Despite most measures of diet having only moderate reliability, longitudinal studies of dietary intake indicate some degree of continuity in food choices over childhood (Skinner et al., 2002; Zive et al., 2002), and even from adolescence to adulthood (Lien and Klepp, 1999; Lake et al., 2006). The Framingham Children's Study reported correlations of 0.74 and 0.66 for fruit/vegetable and dairy intake respectively, in children from ages 3 to 12 years (Moore et al., 2005).
Another way to look at dietary continuity is to investigate tracking of eating behaviour traits. Hunger, external responsiveness, satiety sensitivity and emotional eating are aspects of appetite that have been shown to be associated with adiposity (Van Strien et al., 1985; Barkeling et al., 1992; Epstein et al., 1996; Braet and Van Strien, 1997; Jansen et al., 2003; Wardle, 2006; Wardle, 2007). Modest continuity in ‘eating in the absence of hunger’ (Fisher and Birch, 2002) and dietary restraint (Marchi and Cohen, 1990; Shunk and Birch, 2004) have been reported in childhood. Variations in appetite have also been observed in infancy, with infants who show greater enthusiasm for feeding gaining more weight (Stunkard et al., 1999; Wright and Birks, 2000).
While research in adults has typically used psychometric measures (for example, Three Factor Eating Questionnaire, Stunkard and Messick, 1985; Dutch Eating Behaviour Questionnaire (DEBQ), Van Strien et al., 1986), most research in children has used behavioural measures (for example, reduction of intake after a preload and increase in intake with highly palatable foods). Behavioural measures have the advantage of avoiding subjectivity but they are difficult to use in larger scale, longitudinal studies because of the practicalities of carrying out repeated behavioural assessments. Two psychometric measures of appetite have been developed recently for use with children. A paediatric, parent-completed version of the DEBQ assesses restrained, external and emotional eating and has been valuable in the clinical context (Braet and Van Strien, 1997), showing good test–retest reliability and internal consistency (Caccialanza et al., 2004). The Child Eating Behaviour Questionnaire (CEBQ) is a parent-completed measure designed to assess normal variation in eating behaviour, with scales measuring a larger number of constructs, including satiety responsiveness (SR), slowness in eating (SE), food responsiveness (FR), enjoyment of food (EF), food fussiness (FF), emotional overeating (EOE) and emotional undereating (EUE; Wardle et al., 2001b). Short-term reliability is good and significant validity has been demonstrated against measured food intake (Wardle et al., 2001a; Carnell and Wardle, 2007).
The present study examined continuity and change in CEBQ scores from ages 4 to 11 in a sample of 322 twin children. There is no hard and fast rule about the magnitude of correlation required to support a claim for ‘tracking’, so we based our predictions on longitudinal correlations for other traits that are generally accepted to be stable aspects of temperament or personality. A meta-analysis of longitudinal studies of personality (with a mean time interval between testing of 6.7 years) showed correlations that varied with age, rising from 0.41 within childhood to 0.55 at age 30 years and 0.70 after age 50 years (Fraley and Roberts, 2005). Because the children in the present study were young, correlations in the range 0.3–0.5 would be consistent with the hypothesis of trait continuity. We also assessed changes in mean CEBQ scores, because alongside individual continuity, there can still be changes in the average level of the trait. Given evidence of better caloric compensation in younger children (Birch and Deysher, 1986), we hypothesized that appetitive traits related to satiety would decrease with age, while traits associated with higher consumption would increase with age.
Participants and methods
Participants were drawn from the Twins Early Development Study (TEDS), a population-based sample of over 15 000 twin pairs born in the United Kingdom between 1994 and 1996 (Oliver and Plomin, 2007). The families described in the present study were taking part in an investigation of genetic and environmental influences on appetite and growth, and had been selected on the basis of parental body mass index (BMI). Half the families had obese or overweight parents and half had normal-weight parents. Sampling children from obese and normal-weight parents allowed us to represent children across a range of weights and, potentially, eating behaviour characteristics. The groups were matched for geography and paternal occupation (Wardle et al., 2001a). Both groups contributed data to the present study, with results combined after testing for group differences in continuity and stability.
The sample in 1999 consisted of 428 children (214 families), 200 from obese or overweight families and 228 from normal-weight families, from all round England and Wales, of whom we had complete CEBQ data for 400 children. The families were contacted again in 2006, and we obtained complete data from 322 children; a response rate of 75%. In the final sample, 182 children were from families with obese or overweight parents and 140 from normal-weight. Families were visited at home where mothers completed the CEBQ.
Ethical approval was granted by the University College London Committee for the Ethics of non-NHS Human Research. Verbal consent was obtained from parents on the telephone before the home visit. Consent forms were completed by parents on behalf of themselves and their children at the start of the home visit.
Mothers completed the CEBQ for each of the twins. The CEBQ scales included in this study were satiety responsiveness (five items), slowness in eating (four items), responsiveness to food (five items), enjoyment of food (four items), food fussiness (six items), emotional overeating (four items) and emotional undereating (four items). Responses were scored from 1–5 (‘never’, ‘seldom’, ‘sometimes’, ‘often’ and ‘always’). The following Cronbach's α-values using the 2006 CEBQ data showed good internal reliability: SR: α=0.79; SE: α=0.70, FR: α=0.83, EF: α=0.86, FF: α=0.91, EOE: α=0.77, and EUE: α=0.72. Results from 1999 were very similar and are not reported here.
Mothers also gave information on their own educational status, the fathers' occupation and the family's material socioeconomic status (MSES). Maternal education was categorized into two groups: lower (no qualifications, GCSE, CSE or O-level) and higher (A-level, S-level, HNC, HDD, undergraduate degree and postgraduate degree) educational qualifications. Fathers' occupation was divided into three categories (unskilled/partly skilled, skilled manual/nonmanual and professional/managerial/technical). MSES was assessed with four simple questionnaire items, namely: housing tenure (0—rented and 1—owned), car ownership (0—no car, 1—1 car and 2—>1 car), dishwasher ownership (0—no and 1—yes), and computer ownership (0—no and 1—yes). An index was created by summing the scores on these four questions to create a scale from zero (lower MSES) to five (higher MSES, indicated by owning two or more cars, a dishwasher, a computer and owning a house/flat). MSES was then divided into low (0–2) and high (3–5) categories. The index has been used in other research studies (Wardle et al., 2004).
Statistical analyses were performed using SPSS 14.0. CEBQ subscale scores were generated by calculating the mean of the items. To check that the CEBQ structure was the same at both ages, we compared the pattern of intercorrelations. As all scales were approximately and normally distributed, Pearson's correlations were used to assess individual continuity from ages 4 to 11 years. Partial correlations were also calculated controlling for the precise time interval between visits. Correlations were calculated separately for children of obese and normal-weight parents, for monozygotic (MZ) and dizygotic (DZ) twins, for first- and second-born twins, for boys and girls and for children age 4 years above or below the 85th percentile of BMI for age and sex relative to 1990 British norms (Cole, 2002), to test for group differences in continuity. Paired samples t-tests were used to compare means of the subscale scores between ages 4 and 11 years. Effect size for each of the subscales was calculated by dividing the mean change score (between ages 4 and 11 years) by the s.d. of the initial mean score (Cohen, 1992). Again, results were calculated for each subgroup to check that there were no differences.
Characteristics of the sample with complete data from both years (follow-up sample) and all 1999 participants (baseline sample) are outlined in Table 1. Participants with complete data did not differ significantly from the baseline sample on any of the variables, although there was a trend for fewer boys than girls to take part in the follow-up. Children had a mean age of 4.4 (s.d. 0.4) in 1999 and 11.2 (0.5) in 2006. Approximately, half the children (51%) were from DZ pairs and half from MZ pairs (49%). In the follow-up sample, 38.5% were boys and 61.5% girls.
Individual continuity of CEBQ subscales from ages 4 to 11 years
There were no significant differences in correlations over time by parental weight category, sex, twin birth order, zygosity or child weight status at 4 years. Correlations are therefore reported for the full sample only. Table 2 presents correlations over time for each of the CEBQ subscales. All correlations were highly significant, and ranged from 0.29 for EUE to 0.55 for FF (P<0.001), with most values falling between 0.44 and 0.46. Partial correlations controlling for time differences between the visits produced almost identical results (not tabulated).
Stability of CEBQ subscales
There were no differences in changes in mean CEBQ scores by gender, zygosity, birth order, parent group or child weight status at 4 years, so data from the whole sample were used in the final analyses. Table 3 gives the mean CEBQ subscale scores in children at ages 4 and 11 years, along with the results of paired t-tests. SR, SE, FF and EUE significantly decreased over time, whereas FR, EF and EOE significantly increased over time. To investigate the magnitude of these changes over time, effect size for each subscale was calculated. The effect size of the changes indicate very small increases in FR (ES<0.20), small changes in EF, FF and EUE (0.20<ES<0.50), and moderate changes in SR, SE and EOE (0.50<ES<0.80).
To test for effects of the twin structure of the data on relationships, correlations and t-tests were repeated separately in first-born and second-born twins. However, as results for each subgroup were very similar, these results are not presented.
The results of this study demonstrate that aspects of eating behaviour that can be measured reliably at age 4 years show striking continuity throughout childhood. Significant correlations were found for all seven CEBQ subscales between ages 4 and 11 years. This indicates that children who were relatively high on satiety responsiveness at age 4 years still had relatively high scores at 11 years, and, similarly, those who enjoyed food the most at age 4 years, still enjoyed it more than their peers at 11 years. The majority of correlations fell between 0.44 and 0.46, with a slightly lower correlation in emotional undereating (0.29) and a higher correlation for food fussiness (r=0.55). Results were robust across parental weight category, child weight category at 4 years, child gender and zygosity.
The magnitude of the continuity in eating behaviours seen here is comparable to the continuity observed in longitudinal studies of temperamental characteristics such as shyness, activity or emotionality (Asendorpf, 1992; Roberts and DelVecchio, 2000), which are widely viewed as stable traits (Fraley and Roberts, 2005). This supports the idea that children have characteristic ways of interacting with their food environments and these persist over time. Moderate rather than high correlations are to be expected in view of the developmental changes between ages 4 and 11 years and the variation in rate of development between children.
Individual continuity can coexist with changes in mean levels of a trait, and the present sample showed significant changes from ages 4 and 11 years. As predicted, scores on scales related to stronger satiety responsiveness, such as eating slowly, being fussy about eating, getting full easily or eating less in conditions of emotional arousal, reduced over time. In contrast, traits related to food responsiveness and upregulation of eating, such as enjoyment of food, or overeating in response to emotional arousal, increased over time. These changes indicate that although children retain their position relative to others, small-to-moderate changes in the magnitude of appetitive traits occur through childhood. This is similar to findings in relation to food choice in adolescents, where there is evidence both for strong tracking and mean changes over time (Kelder et al., 1994). Developmental changes could reflect changes in the food environment. For example, enjoyment of food may increase and food fussiness decrease as the child is exposed to a wide variety of foods in the environment, while speed of eating may increase as the child becomes more proficient at feeding.
The pattern of changes in eating behaviour suggests an increasing likelihood of overeating as children get older. One hypothesis is that as aspects of the food environment change—for example, as a result of a wider variety of choices and more ‘child-friendly’ food—the manifest expression of these traits increases. Satiety is also known to have a strong ‘sensory-specific’ element (Rolls, 1986), so as food variety and children's control over what they eat increase, children may choose foods that minimize sensory-specific satiety and, thus, the satiety responsiveness levels overall.
One limitation of this study is that the CEBQ was completed by the parent on both occasions. However, it is unlikely that parents recalled their previous responses over the 6-year interval, and the patterning of differential changes over time supports this. The use of psychometric rather than behavioural measures could be seen as a limitation, but it may also be a strength. Mothers regularly observe their children's eating behaviours, putting them in a strong position to appraise habitual behaviour, as opposed to one-off behaviours. Using a parent-report instrument also avoids difficulties in comprehension and self-awareness that would be encountered while administering a similar questionnaire to children. It is true that CEBQ responses may be affected by the mother's emotional attitude towards food and by forms of social desirability bias, but the observed correlations between the CEBQ scores and objective measures of eating behaviour alleviate some of these concerns (Carnell and Wardle, 2007). The current estimate of individual continuity may be inflated by a shared method and a shared observer, but it is clear that continuity for eating behaviour traits measured first at age 4 years and then 7 years later is similar to the continuity of personality traits over a similar time period.
Future research should explore the origins of these appetitive characteristics, which could be genetic or environmental. Characteristic patterns of eating may have been learned through continual interaction between parent and child; alternatively, genetic factors may play an important role. Nevertheless, given this evidence for tracking in eating behaviour in children, the CEBQ could still be used to identify young children with problematic appetitive styles irrespective of the origins of the problem. Parents might find it helpful to know that children are not very likely to simply ‘grow out of’ any unhealthy eating traits. Instead, as children grow older and have more influence over food choices, it will become increasingly important to make the home environment a ‘healthy food zone’ that minimizes opportunities for the expression of appetitive traits that could facilitate excess weight gain.
Asendorpf JB (1992). A Brunswikean approach to trait continuity: application to shyness. J Pers 60, 53–77.
Barkeling B, Ekman S, Rossner S (1992). Eating behaviour in obese and normal weight 11-year-old children. Int J Obes Relat Metab Disord 16, 355–360.
Birch LL, Deysher M (1986). Caloric compensation and sensory specific satiety: evidence for self regulation of food intake by young children. Appetite 7, 323–331.
Braet C, Van Strien T (1997). Assessment of emotional, externally induced and restrained eating behaviour in nine to twelve-year-old obese and non-obese children. Behav Res Ther 35, 863–873.
Caccialanza R, Nicholls D, Cena H, Maccarini L, Rezzani C, Antonioli L et al. (2004). Validation of the Dutch Eating Behaviour Questionnaire parent version (DEBQ-P) in the Italian population: a screening tool to detect differences in eating behaviour among obese, overweight and normal-weight preadolescents. Eur J Clin Nutr 58, 1217–1222.
Carnell S, Wardle J (2007). Measuring behavioural susceptibility to obesity: validation of the child eating behaviour questionnaire. Appetite 48, 104–113.
Cohen J (1992). A power primer. Psychol Bull 112, 155–159.
Cole TJ (2002). A chart to link child centiles of body mass index, weight and height. Eur J Clin Nutr 56, 1194–1199.
Epstein LH, Paluch R, Coleman KJ (1996). Differences in salivation to repeated food cues in obese and nonobese women. Psychosom Med 58, 160–164.
Fisher JO, Birch LL (2002). Eating in the absence of hunger and overweight in girls from 5 to 7 y of age. Am J Clin Nutr 76, 226–231.
Fraley RC, Roberts BW (2005). Patterns of continuity: a dynamic model for conceptualizing the stability of individual differences in psychological constructs across the life course. Psychol Rev 112, 60–74.
Jansen A, Theunissen N, Slechten K, Nederkoorn C, Boon B, Mulkens S et al. (2003). Overweight children overeat after exposure to food cues. Eat Behav 4, 197–209.
Kelder SH, Perry CL, Klepp KI, Lytle LL (1994). Longitudinal tracking of adolescent smoking, physical activity, and food choice behaviors. Am J Public Health 84, 1121–1126.
Lake AA, Mathers JC, Rugg-Gunn AJ, Adamson AJ (2006). Longitudinal change in food habits between adolescence (11–12 years) and adulthood (32–33 years): the ASH30 Study. J Public Health (Oxf) 28, 10–16.
Lien N, Klepp KI (1999). Consumption of fruit and vegetables in a teenage cohort—observed changes. Tidsskr Nor Laegeforen 119, 2327–2330.
Marchi M, Cohen P (1990). Early childhood eating behaviors and adolescent eating disorders. J Am Acad Child Adolesc Psychiatry 29, 112–117.
Moore LL, Singer MR, Bradlee ML, Djousse L, Proctor MH, Cupples LA et al. (2005). Intake of fruits, vegetables, and dairy products in early childhood and subsequent blood pressure change. Epidemiology 16, 4–11.
Oliver BR, Plomin R (2007). Twins Early Development Study (TEDS): a multivariate, longitudinal genetic investigation of language, cognition and behaviour problems from childhood through adolescence. Twin Res Hum Genet 10, 96–105.
Roberts BW, DelVecchio WF (2000). The rank-order consistency of personality traits from childhood to old age: a quantitative review of longitudinal studies. Psychol Bull 126, 3–25.
Rolls BJ (1986). Sensory-specific satiety. Nutr Rev 44, 93–101.
Serdula MK, Ivery D, Coates RJ, Freedman DS, Williamson DF, Byers T (1993). Do obese children become obese adults? A review of the literature. Prev Med 22, 167–177.
Shunk JA, Birch LL (2004). Validity of dietary restraint among 5- to 9-year old girls. Appetite 42, 241–247.
Skinner JD, Carruth BR, Wendy B, Ziegler PJ (2002). Children's food preferences: a longitudinal analysis. J Am Diet Assoc 102, 1638–1647.
Steinberger J, Moran A, Hong CP, Jacobs Jr DR, Sinaiko AR (2001). Adiposity in childhood predicts obesity and insulin resistance in young adulthood. J Pediatr 138, 469–473.
Stunkard AJ, Berkowitz RI, Stallings VA, Schoeller DA (1999). Energy intake, not energy output, is a determinant of body size in infants. Am J Clin Nutr 69, 524–530.
Stunkard AJ, Messick S (1985). The three-factor eating questionnaire to measure dietary restraint, disinhibition and hunger. J Psychosom Res 29, 71–83.
Van Strien T, Frijters JE, Roosen RG, Knuiman-Hijl WJ, Defares PB (1985). Eating behavior, personality traits and body mass in women. Addict Behav 10, 333–343.
Van Strien T, Frijters JER, Bergers GPA, Defares PB (1986). Dutch eating behaviour questionnaire for assessment of restrained, emotional and external eating behaviour. Int J Eat Disord 5, 295–315.
Wang Y, Ge K, Popkin BM (2000). Tracking of body mass index from childhood to adolescence: a 6-y follow-up study in China. Am J Clin Nutr 72, 1018–1024.
Wardle J (2006). Eating behaviour in obesity. In: Shepherd R and Raats M (eds). The Psychology of Food Choices. CABI: UK, pp. 375–388.
Wardle J (2007). Eating behaviour and obesity. Obes Rev 8 (Suppl 1), S73–S75.
Wardle J, Guthrie C, Sanderson S, Birch L, Plomin R (2001a). Food and activity preferences in children of lean and obese parents. Int J Obes Relat Metab Disord 25, 971–977.
Wardle J, Guthrie CA, Sanderson S, Rapoport L (2001b). Development of the Children's Eating Behaviour Questionnaire. J Child Psychol Psychiatry 42, 963–970.
Wardle J, McCaffrey K, Nadel M, Atkin W (2004). Socioeconomic differences in cancer screening participation: comparing cognitive and psychosocial explanations. Soc Sci Med 59, 249–261.
Whitaker RC, Wright JA, Pepe MS, Seidel KD, Dietz WH (1997). Predicting obesity in young adulthood from childhood and parental obesity. N Engl J Med 337, 869–873.
Wright C, Birks E (2000). Risk factors for failure to thrive: a population-based survey. Child Care Health Dev 26, 5–16.
Zive MM, Berry CC, Sallis JF, Frank GC, Nader PR (2002). Tracking dietary intake in white and Mexican-American children from age 4 to 12 years. J Am Diet Assoc 102, 683–689.
This research was funded by the Biotechnology and Biological Sciences Research Council.
Contributors: JW had the idea for the study and obtained funding. JA and CS carried out the home visits. JA, CHMvJ and CS participated in analysis of the data. All authors took part in the interpretation of results and drafting the manuscript, and approved the final version of the manuscript.
About this article
Cite this article
Ashcroft, J., Semmler, C., Carnell, S. et al. Continuity and stability of eating behaviour traits in children. Eur J Clin Nutr 62, 985–990 (2008). https://doi.org/10.1038/sj.ejcn.1602855
Does social capital moderate the association between children's emotional overeating and parental stress? A cross-sectional study of the stress-buffering hypothesis in a sample of mother-child dyads
Social Science & Medicine (2020)
Bidirectional relationships between appetitive behaviours and body mass index in childhood: a cross-lagged analysis in the Generation XXI birth cohort
European Journal of Nutrition (2020)
Obesity risk in Hispanic children: Bidirectional associations between child eating behavior and child weight status over time
Eating Behaviors (2020)
Cognitive Development (2020)