Parents are of high importance in the development of their children’s dietary preferences that eventually lead to their dietary patterns. Parents directly determine the child’s physical and social environment by deciding which foods are available and in what quantity(Reference Nicklas, Baranowski and Baranowski1–Reference Hearn, Baranowski and Baranowski3). Parents’ own food-related behaviours serve as a role model and thereby affect the dietary habits of their children(Reference Birch and Fisher4, Reference Cullen, Baranowski and Rittenberry5). Indirectly, parents influence their children’s behaviour and habits through socialization. They can apply different food management practices that control, encourage or restrict the intake of certain foods. Research of the recent past has drawn attention to these food-related parenting practices as having an important influence on children’s diet(Reference Faith, Scanlon and Birch6).
With the transition from childhood to adolescence, at the age of 12–13 years, adolescents become more independent, and therefore it might well be that the impact of parenting practices diminishes(Reference De Bourdeaudhuij7). Few studies have, however, investigated the influences of the home food environment (parental intake, availability and parenting practices) on adolescents’ diets.
In a study by Young et al.(Reference Young, Fors and Hayes8), perceived parental modelling and home availability were significantly associated with their children’s fruit and vegetable (F&V) consumption. In a cross-sectional study of project EAT (Eating among Teens)(Reference Hanson, Neumark-Sztainer and Eisenberg9), parental intake was positively associated with dairy intake for boys and with dairy, F&V intake for girls; significant positive associations were found for F&V intake by home availability among girls and for dairy intake by serving milk at meals for male adolescents. In a longitudinal study of project(Reference Arcan, Neumark-Sztainer and Hannan10) EAT, investigating a long-term role for parent modelling and availability above and beyond any short-term impact, parental intake predicted the dietary intake of young adults but not of high-school students; serving of vegetables at dinner (a measure of availability and accessibility) was a significant predictor of adolescents’ and young adults’ intake.
Focus groups suggest that food rules(Reference Neumark-Sztainer, Story and Perry11) and prior food rules(Reference De Bourdeaudhuij7) continue to exert their influence on adolescents’ food choices. In studies of van der Horst et al.(Reference van der Horst, Kremers and Ferreira12) and de Bruijn et al.(Reference de Bruijn, Kremers and de Vries13), more restrictive parenting practices were found to be associated with less soft drink consumption in adolescents. Congruent herewith are the retrospective studies of De Bourdeaudhuij(Reference De Bourdeaudhuij7), in which more fat and sweet foods were consumed by adolescents who reported more permissiveness (fewer restrictions and obligations) in their family at the age of 10 years. However, they found no evidence of a relationship of prior food rules with the consumption of healthy foods in adolescence. In a study by Haerens et al.(Reference Haerens, Craeynest and Deforche14), less restrictive food rules were associated with a higher fat intake in boys and lower fruit consumption in girls; however, no association was found with soft drink consumption.
The present study will build on the previous studies of De Bourdeaudhuij(Reference De Bourdeaudhuij7) in which a cross-sectional design was used to study influences on healthy and less healthy foods and in which the results are based on perceptions of family food rules in the past. In the present study, a longitudinal design was used: the mothers’ reports on parenting practices at the age of 10 years are used to predict intake of healthy and less healthy foods during childhood and adolescence as well as changes in intake during this transition. In addition, inclusion of measures of parental intake and availability will allow one to investigate the hypothesis that each of these factors is independently related to children’s and adolescents’ dietary intake.
Methodology
Procedure
The results presented here are based on the first and fourth measurements of the Longitudinal Eating and Activity study in which children have been followed from 2002 to 2005 (Fig. 1). In 2002, 100 elementary schools from two Flemish regions (East and West Flanders) were randomly selected from the official list of the Flemish government. The principals were sent a recruitment letter and afterwards contacted by phone. There were fifty-nine principals who agreed to cooperate in the study. The main reason for non-participation was lack of time. All children in the fifth grade (10-year-olds) were invited to participate in the study (n 1957). Informed consent to participate in the longitudinal study was received by 1725 parents (88 % of eligible children). In 2002 (T1), the children completed a self-administered questionnaire on eating habits and physical activity, demographic variables and possible psychosocial determinants in the classroom under the supervision of one researcher and their classroom teacher. The same procedure was followed for T2 (2003). In 2004, children changed, however, from primary to secondary schools, making classroom administration not feasible. Therefore, for T3 and T4, the children were contacted at home by postal mail. The envelope contained a letter addressed to the parents asking them to encourage their child to participate in the study, and a letter addressed to the child asking them to login with a personal code to a website and complete an online questionnaire. As the response rate was very low (about 30 %), non-respondents were sent a reminder including a paper–pencil questionnaire and a pre-stamped envelope, 8 weeks later.
Fig. 1 Response at school and individual levels – Longitudinal Eating and Activity study, East and West Flanders, Belgium, 2002–2005
In total, 874 adolescents completed the questionnaire at T4. Of these, 764 could be matched to T1. The main reason for not participating at T3 and T4 was inability to be reached (moved) or unwillingness to fill in the questionnaire. Only the respondents for whom the parental questionnaire was completed by the mother/stepmother were included in the current analyses (n 639).
Ethical approval was obtained from the Ethics Commission of the Ghent University Hospital.
Measures
Dietary patterns
The children/adolescents were asked how many times a week they usually consumed fruit, vegetables, sugared soft drinks, sweets and crisps (FFQ). The response options were: ‘never ( = 0)’, ‘<1 d/week ( = 0·25)’, 1 d/week ( = 1)’, ‘2–4 d/week ( = 3)’, ‘5–6 d/week ( = 5·5)’, ‘once a day’, ‘every day ( = 7)’, and ‘every day more than once ( = 14)’. The consumption of F&V is important in reducing the risk of cancer and CVD(Reference Johnson15–Reference Ness and Powles17). The consumption of soft drinks and sweet and savoury snacks can impede the intake of more nutritious foods by reducing appetite control(Reference Ludwig, Peterson and Gortmaker18).
The two F&V items were added together to form an F&V score; the consumption of regular soft drinks, sweets and crisps was combined to form an excess score(Reference Vereecken, Rossi and Giacchi19).
Parents were asked to report their own intake with parallel questions at T1.
Food-related parenting practices
Items measuring food-related parenting practices were based on a pilot study(Reference Vereecken, Keukelier and Maes20). The items addressed the use of pressure (six items), encouragement through material reward (three items), encouragement through negotiation (five items), catering on children’s demand (four items), permissiveness (seven items), avoiding negative modelling (two items) and verbal praise (two items). Parents were asked to respond on a 5-point scale: 1 = never, 2 = mostly not, 3 = sometimes, 4 = most of the time and 5 = always. Internal consistencies of the subscales are reported in Table 1. For each scale, the average of the group of the respective practices was computed, if more than 50 % of the scale items were answered.
Table 1 Descriptives of dependent and independent variables: LEA study, East and West Flanders, Belgium, 2002–2005 (n 609)
LEA, Longitudinal Eating and Activity study; F&V, fruit and vegetables.
The availability of fruit, soft drinks, biscuits, sweets and crisps was questioned on a 3-point scale (2 = always/mostly available, 1 = sometimes and 0 = rarely/never). The four unhealthy items were combined to form an unhealthy availability scale.
Statistics
Three sets of models were conducted. To examine associations between potential baseline correlates and the food scores, separate multi-level regression analyses were conducted for each parenting practice with baseline and follow-up food scores, controlled for sociodemographic characteristics (gender and mother’s educational level; model 1). To examine the multivariate association of all parenting practices simultaneously, all parenting practices were entered simultaneously in a second set of models (model 2). To examine the associations between parenting practices and change in food scores between baseline and follow-up, the previous model was additionally adjusted for baseline food scores (model 3).
All variables (except sociodemographics) were standardized to allow for relative comparisons of strength between the observed associations. The β-coefficients can be interpreted as the amount of sd change in food score associated with a 1 sd change in the respective parenting practice. Analyses were conducted using MLwiN version 2.02(Reference Rasbash, Charlton and Browne21), with respondents nested within schools at T1 (two-level random intercept model). To estimate the proportion of the explained variance of the home food environment, the proportion of unexplained variance of the full model is compared with a model including only a constant and sociodemographics for baseline measurement and including a constant, sociodemographics and baseline measurement for follow-up data. P values at <0·05 are considered significant.
Results
Of the 609 respondents included in the analyses (600 for the F&V score; 530 for the excess score), 50·6 % were boys. Mother’s educational level was distributed as follows: 24·5 % low (lower technical, higher vocational or less), 23·2 % medium (technical or general higher secondary education) and 52·4 % high (bachelor’s or master’s degree). There were 95% of mothers who reported that fruit was always or mostly available; therefore, this variable is not likely to be able to distinguish and was excluded for further analyses. Descriptives of the scale variables are presented in Table 1. Correlations between the dependent and independent variables are presented in Table 2.
Table 2 Correlations between standardized dependent and independent variables – LEA study, East and West Flanders, Belgium, 2002–2005 (n 609)
LEA, Longitudinal Eating and Activity study; F&V, fruit and vegetables.
*P < 0·05; **P < 0·01; ***P < 0·001.
Table 3 shows that lesser permissiveness, a lower unhealthy availability score and a higher F&V score for mothers were associated with a higher F&V score, both at baseline and follow-up (model 1). More negotiation was only significantly associated with a higher F&V score at baseline, although a significant association with follow-up data became visible when all variables were included in the analyses (model 2). However, the positive association between avoiding negative modelling and follow-up was washed out in the latter model. The only significant predictors for changes in F&V score were mother’s F&V score and availability of unhealthy foods.
Table 3 Associations of baseline factors with the F&V score at baseline and follow-up, 4 years later – LEA study, East and West Flanders, Belgium, 2002–2005 (n 600)
LEA, Longitudinal Eating and Activity study; F&V, fruit and vegetables.
Model 1: Separate regression analyses for each individual factor, controlling for gender and mother’s educational level; for the random part of model 1 only sociodemographics are included for baseline data and sociodemographics and child’s baseline F&V score for follow-up data.
Model 2: Multivariate regression model adjusted for all other variables.
Model 3: Model 2 additionally adjusted for baseline data.
Significant parameters are given in bold.
Reward, catering on demand, permissiveness, availability of unhealthy foods and mother’s excess score were all positively associated with children’s excess score at baseline, whereas avoiding negative modelling behaviour was negatively associated (Table 4, model 1). However, when all variables were entered in the same model (model 2), only availability and mother’s excess score remained significant. Availability of unhealthy foods, mother’s excess score and permissiveness were also positively related to the follow-up score (models 1 and 2) and changes in excess score from baseline to follow-up (model 3). However, the negative association of avoiding negative modelling disappeared in model 2.
Table 4 Associations of baseline factors with the excess score at baseline and follow-up, 4 years later – LEA study, East and West Flanders, Belgium, 2002–2005 (n 530)
LEA, Longitudinal Eating and Activity study; F&V, fruit and vegetables.
Sociodemographics are categorical and all other variables are scale variables.
Model 1: Separate regression analyses for each individual factor, controlling for gender and mother’s educational level; for the random part only sociodemographics are included for baseline data and sociodemographics and child’s baseline excess score for follow-up data.
Model 2: Multivariate regression model adjusted for all other variables.
Model 3: Model 2 additionally adjusted for baseline data.
Significant parameters are given in bold.
Finally, higher dietary scores at baseline predicted higher consumption at follow-up for both dietary scores.
The proportions of variance explained by parenting practices for the baseline measurements are 8 % for the F&V score and 12 % for the excess score. The proportions of variance explained for the follow-up measurements are 5 % and 10 %, respectively.
Attrition analyses
Significant differences were found between the participants included in the present analyses and the baseline only or excluded participants for six of the twelve variables. Those who did not participate at follow-up or were excluded were less likely to have a mother with higher education and had a higher excess score. Their mothers not only had a higher excess and a lower F&V score, but also reported less negotiation and more permissiveness. No gender difference was found.
Discussion
In the present study, a longitudinal design is used to identify parenting practices during childhood that predict intake of healthy and less healthy patterns during childhood and, in future, during adolescence.
The results indicate a consistent association between mothers’ and their children’s intake for both the F&V score and the excess score indicating that children and adolescents’ diets are associated with the foods eaten by their parents.
The present study also explored the role of availability. Availability of fruit seemed not to be a problem: 95 % had mostly or always fruit available at home. Congruent with a study by Larson et al., the availability of less healthy foods and beverages was not only positively associated with the excess score but also negatively with the F&V score(Reference Larson, Neumark-Sztainer and Harnack22), suggesting that parents should be encouraged to reduce the availability of less healthy foods. In a study of 12-year-olds by Haerens et al.(Reference Haerens, Craeynest and Deforche14), the availability of unhealthy foods was related to fat intake and soft drink consumption in boys, but not in girls; however, no association was found with the consumption of fruit.
Further, we investigated the role of food parenting practices on children’s and adolescents’ dietary intake. After taking into account all other factors, especially permissiveness, and to a lesser extent, negotiation seemed to be promising factors. Letting children decide on what they eat and when, and allowing them to consume sweets and soft drinks when they like was significantly detrimental to their future intake of less healthy foods. These results are in line with the findings of De Bourdeaudhuij(Reference De Bourdeaudhuij7): they also found a positive relationship between permissiveness in their family at the age of 10 years and the consumption of more fat and sweet foods during adolescence and no association with healthy foods. In contrast, other studies, however, found that strict parental control may have adverse effects such as increasing children’s preference for and intake of restricted foods(Reference Fisher and Birch23, Reference Fisher and Birch24).
Strengths and limitations
The first limitation of the present study is the use of brief scales to report very complex behaviours. In addition, the parenting practice scales differ from those used in the current literature and have not been validated. However, at the start of the present study, little research had been conducted in this area. The second limitation is the rather crude dietary intake assessment. We only asked about consumption frequency in d/week, and therefore no information on portion size is collected. Third, both dietary intake and parenting scales are based on self report and therefore might be responded to in a socially desirable way. The fourth limitation is the selective dropout, which compromises the generalizability of the results. A strength of the present study is the prospective character and the rather large sample despite the considerable attrition. Nonetheless, causality can still not be stated as prospective relationships can, as for cross-sectional studies, be due to a third antecedent.
Conclusion
These results extend previous ones obtained from cross-sectional studies and give further support to the importance of including the family in prevention campaigns aimed at children and adolescents. Parents’ intake and restricting the availability of unhealthy foods not only appeared to have a consistent impact on children’s and adolescents’ diet, but also less permissive food-related parenting practices may improve adolescents’ diet.
Acknowledgements
Sources of funding: The first and second surveys were funded by the Ghent University (Fund no. 011/105/04). Carine Vereecken and Leen Haerens are postdoctoral researchers supported by the FWO Flanders. The study has been facilitated by the EU-funded HOPE project: ‘Health-promotion through Obesity Prevention across Europe (the Commission of the European Communities, SP5A-CT-2006-044128). The study does not necessarily reflect the Commission’s views and in no way anticipates the Commission’s future policy in this area. Conflicts of interest declaration: The authors have no conflict of interest. C.V. performed most of the statistical analyses and the writing of the manuscript. Authorship responsibilities: L.M. and I.D.B. supervised the study. All authors advised on the study design and interpretation of the results and reviewed and approved the manuscript.
