AJCN. First published ahead of print September 3, 2014 as doi: 10.3945/ajcn.114.088922.
Feeding practices and child weight: is the association bidirectional in preschool children?1–4 Pauline W Jansen, Anne Tharner, Jan van der Ende, Melissa Wake, Hein Raat, Albert Hofman, Frank C Verhulst, Marinus H van Ijzendoorn, Vincent WV Jaddoe, and Henning Tiemeier ABSTRACT Background: Parental feeding practices are associated with children’s body mass index (BMI). It has been generally assumed that parental feeding determines children’s eating behaviors and weight gain, but feeding practices could equally be a parent’s response to child weight. Objective: In longitudinal analyses, we assessed the directionality in the relation between selected controlling feeding practices and BMI in early childhood. Design: Participants were 4166 children from the population-based Generation R Study. BMI was measured at ages 2 and 6 y. With the use of the Child Feeding Questionnaire, parents reported on restriction, monitoring, and pressure to eat (child age: 4 y). BMI and feeding-behavior scales were transformed to SD scores. Results: With the use of linear regression analyses, there was an indication that a higher BMI at age 2 y predicted higher levels of parental restriction (adjusted b = 0.07; 95% CI: 0.04, 0.10) and lower levels of pressure to eat (adjusted b = 20.20; 95% CI: 20.23, 20.17) 2 y later. Restriction at age 4 y positively predicted child BMI at 6 y of age, although this association attenuated to statistical nonsignificance after accounting for BMI at age 4 y (b = 0.01; 95% CI: 20.01, 0.03). Pressure to eat predicted lower BMI independently of BMI at age 4 y (b = 20.02; 95% CI: 20.04, 20.01). For both restriction and pressure to eat, the relation from BMI to parenting was stronger than the reverse (Wald’s test for comparison: P = 0.03 and , 0.001, respectively). Monitoring predicted a lower child BMI, but this relation was explained by confounding factors. Conclusions: Although the feeding-BMI relation is bidirectional, the main direction of observed effects suggests that parents tend to adapt their controlling feeding practices in response to their child’s BMI rather than the reverse. Therefore, some components of current programs aimed at preventing or treating unhealthy child weight may need to be carefully scrutinized, especially those targeting parental food-related restriction and pressure to eat. Am J Clin Nutr doi: 10.3945/ajcn.114.088922.
INTRODUCTION
Feeding practices of parents have been implicated in children’s BMI and overweight (1–3). This proposition particularly relates to the use of control in food-related situations such as regulating the type and amount of food eaten by children (restricting) or cautiously keeping track of what children eat (monitoring) (4). Another controlling feeding strategy is parents’ use of pressure generally in an attempt to
improve the quantity or quality of children’s food intakes by using direct prompts and having rules about what a child should eat (4). It has been theorized that controlling feeding hinders children’s ability to self-regulate food consumption and override internal cues of satiety, thereby resulting in excessive food intake (5–8). Thus far, studies have tended to support these theories (5–8), but their laboratory-based designs and small sample sizes have limited generalizability. It is equally plausible that not only overcontrol but also a lack of parental control contribute to children’s weight gain (9, 10) because children are likely tempted to overeat in our current obesogenic environment. Few longitudinal studies showed support for this hypothesis. Poor parental restriction (11–13) and monitoring (12) of children’s food intakes were prospectively associated with a higher child BMI. Campbell et al (11) suggested that this relation depends on children’s age with the protective effect of restrictive feeding in middle childhood fading away as children enter adolescence and become more independent. This process is consistent with that shown in 1 From the Departments of Child and Adolescent Psychiatry/Psychology (PWJ, JvdE, FCV, and HT) and Pediatrics (VWVJ), Erasmus University Medical Center (Erasmus MC)-Sophia, Rotterdam, Netherlands; The Generation R Study Group (PWJ, AT, and VWVJ) and Departments of Epidemiology (AT, AH, VWVJ, and HT), Public Health (HR), and Psychiatry (HT), Erasmus MC-University Medical Center Rotterdam, Rotterdam, Netherlands; the Institute of Psychology (PWJ) and School of Pedagogical and Educational Sciences (MHvI), Erasmus University Rotterdam, Rotterdam, Netherlands; the Murdoch Childrens Research Institute, Melbourne, Australia (MW); the Department of Pediatrics, University of Melbourne, Melbourne, Australia (MW); and the Centre for Community Child Health, Royal Children’s Hospital, Melbourne, Australia (MW). 2 None of the funding entities were involved in the design of the study or analysis and interpretation of data. 3 Supported by the Netherlands Organization for Health Research and Development (ZonMW) (ZonMW “Geestkracht” program; grant 10.000.1003), the Netherlands Organization for Scientific Research (NOW) (NWO– ZonMW, VIDI grant 017.106.370; to HT), the Sophia Foundation for Medical Research SSWO (grant 602; to PWJ), and an Australian National Health and Medical Research Council Senior Research Fellowship (1046518; to MW). The Murdoch Childrens Research Institute is supported by the Victorian Government’s Operational Infrastructure Support Program. 4 Address reprint requests and correspondence to PW Jansen, Department of Child and Adolescent Psychiatry/Psychology, Erasmus University Medical Center-Sophia, PO Box 2060, 3000 CB Rotterdam, Netherlands. E-mail: [email protected]. Received March 26, 2014. Accepted for publication August 11, 2014. doi: 10.3945/ajcn.114.088922.
Am J Clin Nutr doi: 10.3945/ajcn.114.088922. Printed in USA. Ó 2014 American Society for Nutrition
Copyright (C) 2014 by the American Society for Nutrition
1 of 8
2 of 8
JANSEN ET AL
a study in adolescence (14) but contrasts with research in toddlers that showed no relation between restriction or monitoring and child weight at follow up (15, 16). Parents’ pressure on their children to eat more has also been studied longitudinally with more parental pressure to eat prospectively predicting lower child BMI (12, 13). None of these longitudinal studies tested whether there was an optimal effect of controlling feeding on child BMI (curvilinearity of associations). Previous studies mostly focused on parents’ influences on children’s weight development, whereas it is entirely possible that feeding-weight associations are bidirectional. Evidence-based theories of child development (eg, attachment theory and family systems theory) have suggested that parents generally respond and adapt their parenting strategies to children’s behavior and needs (17). Only 2 studies have explored a possible bidirectionality and reported that high BMI prospectively predicted more parental control (18, 19) and less pressure to eat (18), whereas no evidence was shown to support the reverse effect. These results contradict the previously described unidirectional studies that reported an influence of feeding practices on offspring BMI (11–13). The current study’s aim was to examine longitudinal relations between selected feeding practices and early childhood BMI in a large population based cohort. We expected bidirectional associations with weight influencing parental restriction, monitoring, and pressure, and these feeding practices would also affect weight gain. Moreover, we expected curvilinear relations to reflect that moderate levels of controlling feeding may be associated with the healthiest weight in children.
SUBJECTS AND METHODS
Design This study was embedded in Generation R, which is an ongoing population-based cohort from fetal life onwards (21, 22). Pregnant women who were living in Rotterdam, Netherlands, with an expected delivery date between April 2002 and January 2006 were invited to participate during pregnancy and after the birth of their child (participation rate: 61%). Written informed consent was obtained from all participating children and their parents, and the study was approved by the Medical Ethics Committee of the Erasmus University Medical Center, Rotterdam. Information was obtained by postal questionnaires and hands-on measurements at our own research center and Child Health Centers.
Measures Parental feeding practices Parents completed a postal questionnaire around the fourth birthday of their child which included 3 subscales of the Child Feeding Questionnaire (CFQ) (4). The original English questionnaire was translated by using a standard forward-backward translation method (23). Subscales of the CFQ were used to assess parental attitudes and strategies regarding the control of children’s eating as follows: restriction (8 items), monitoring (3 items), and pressure to eat (4 items). Examples of items were “I intentionally keep some foods out of my child’s reach” (restriction), “How much do you keep track of the high fat foods your child eats?” (monitoring), and “If my child says ‘I’m not hungry,’ I’ll try to get him/her to eat anyway” (pressure to eat). Parents, in most cases the mothers (86.5%), answered the items on a 5-point Likert scale from 1 (never) to 5 (always). Scale scores were only calculated if $75% of items were completed. Continuous scale scores were transformed into SD scores to facilitate effect-size comparisons between scales. Research has provided good evidence for the concurrent validity of the CFQ with actual observations of feeding behaviors of mothers (24). The internal consistency of the administered CFQ scales was moderate for pressure to eat (a = 0.66) and restriction (a = 0.73) and high for monitoring (a = 0.92). We also assessed pressuring feeding at child age 2 y. This feeding practice was used in a cross-lagged modeling approach to estimate both directions of the pressure-to-eat–child-BMI association within the same age period. With the use of the following 5 self-report items, parents (mostly mothers) indicated what they would do if their 2-y-old child did not want to eat: 1) try to convince child to eat, 2) allow child to stop eating fairly quickly (reverse coded), 3) deprive child of something he or she likes (eg, pudding or television watching), 4) child may not leave table until plate is empty, and 5) negotiate with child (if . then .). These items (yes = 1, no = 0) were summed with higher sum scores indicating more pressuring feeding. The internal consistency of this scale was low (a = 0.29), which was not surprising because of the small number of items and the fact that the scale has not been validated. However, the concordance correlation coefficient between the 2 measures of pressure to eat at ages 2 and 4 y was rc = 0.21, which indicated that, despite the different assessment methods and 2-y interval between assessments, these 2 measures assessed related constructs. BMI
Study population Full consent for the postnatal phase of the Generation R Study was obtained for 7295 children and their parents. For the current study, we excluded children without BMI assessed at age 6 y (n = 1801; 24.7%) and those without information on parental feeding practices at age 4 y (n = 1328; 18.2%), which yielded a study population of 4166 children (57.1%). A comparison of included (n = 4166) and excluded (n = 3129) children indicated that data were more often missing in children of lower-educated mothers and with a non-Dutch background (both P , 0.001). No sex differences were shown between children with and without missing data (P = 0.13).
Children’s growth characteristics were obtained at ages 1, 2, 3, and 4 y at the municipal Child Health Centers as part of a routine health care program and at 6 y at the Generation R research center. At both centers, trained staff measured children’s weights and heights by using standardized procedures. Weight was measured by using a mechanical personal scale (SECA) while children were wearing underwear only. Height was measured in standing position by using a Harpenden stadiometer (Holtain Ltd). BMI was calculated as weight divided by height squared. Age- and sex-specific BMI SD scores were calculated by using the Dutch reference population (25) in the Growth Analyzer program.
BIDIRECTIONALITY BETWEEN FEEDING AND CHILD BMI
Covariates Several demographic variables were considered possible confounding factors. Child factors included sex, age at CFQ assessment, and national origin, which was based on the country of birth of both parents. Maternal variables included BMI measured at study intake in early pregnancy and self-reported highest attained educational level (elementary or secondary, higher vocational, and academic), which was considered as a proxy for socioeconomic status. Statistical analyses In all analyses, child BMI and feeding scales were expressed in SD scores. Two sets of linear regression analyses were conducted; each set comprised 3 separate regressions, one for each CFQ scale (ie, restriction, monitoring, and pressure to eat). In the first set, we examined the relation between child BMI at age 2 y and feeding practices at 4 y of age. In the second set, we examined whether feeding practices at ages 2 and 4 y of age predicted child BMI at 4 and 6 y of age, respectively. For both sets of regressions, we present 3 models as follows: 1) unadjusted, 2) adjusted for sociodemographic covariates, and 3) further adjusted for the measure of pressuring feeding practices at age 2 y (first set of regressions) or for baseline BMI at age 2 or 4 y (second set of regressions). Covariates were only included in adjusted models if they changed effect estimates of the unadjusted CFQ-BMI association .5%. With the use of this strategy, all previously mentioned covariates were included in the adjusted model, except for child sex which did not change the effect estimate .5%. In additional analyses, we tested whether feeding-BMI relations were curvilinear by adding a quadratic term to models. Following preceding research (26), weconductedpathanalyses to estimate which direction of CFQ-BMI associations was strongest. We tested 3 path models, again for each feeding scale separately. Each path model included multiple linear regression jointly estimated CFQ-BMI associations in both directions while accounting for continuityinBMI over time. With the inclusion of bothdirections of theCFQ-BMIassociation inone model,pathways were accounted for eachother and coulddirectlybe compared instrength. Wald’s test was used to compare whether the BMI-CFQ association differed statistically from the CFQ-BMI association. Finally, we applied a cross-lagged modeling approach to estimate both directions of the pressure-to-eat–child BMI association within the same age period (from ages 2 to 4 y). We could only test such a model for the pressure to eat because parental restriction and monitoring were not assessed at age 2 y. The model included multiple linear regressions that reflected pressure-to-eat– BMI associations (from ages 2 to 4 and 4 to 6 y), BMI at 2 y of age to pressure to eat at 4 y of age, and continuity in BMI and pressure to eat over time. The model also included cross-sectional correlations between pressure to eat and BMI and was adjusted for the previously mentioned covariates. An acceptable-to-good fit was determined by using a comparative fit index .0.90, and a root mean squared error of approximation ,0.08. Multiple imputation techniques (full conditional specification) were used to account for missing values in baseline BMI (at 2 y of age) and covariates. Imputations were based on available information on all variables included in the study as well as BMI assessments at 1 and 3 y of age (27). Analyses were performed on imputed data, and reported effect estimates were pooled results of 20 imputed data sets. Analyses were performed with STATA 12.0
3 of 8
software (Stata Corp) except for path analyses, which were conducted with Mplus version 7.11 software (Muthe´n & Muthe´n). RESULTS
Characteristics of children and their parents are presented in Table 1. The majority of children included in the study population (68%) were of Dutch origin. Mothers were relatively highly educated with 65% of them having a higher vocational or academic training. Mean (6SD) BMI (in kg/m2) of mothers was 24.5 6 4.1. Child BMI influencing parental feeding In the first set of linear regression analyses, we examined whether child BMI at 2 y preceded feeding behavior of parents at child age 4 y (Table 2). A higher BMI at age 2 y was associated with higher levels of parental restriction and lower levels of pressure to eat 2 y later. These associations were not explained by possible confounding factors nor by pressuring feeding practices at age 2 y [restriction: adjusted b = 0.07 (95% CI: 0.04, 0.10); pressure to eat: adjusted b = 20.20 (95% CI: 20.23, 20.17)]. The final model suggested that the relation from BMI to restriction and pressure to eat was not attributable to any feeding-BMI association at age 2 y. Child BMI at age 2 y did not predict later parental monitoring, although the addition of a quadratic term of monitoring to the model pointed at a significant curvilinear association whereby both a low and high BMI at child age 2 y predicted lower levels of monitoring at age 4 (b-BMI quadratic term = 20.024; 95% CI: 20.04, 20.01). However, this association was attenuated by accounting for confounding factors. Parental feeding influencing child BMI In the second set of linear regression analyses, we examined the reverse association of whether parents’ feeding behaviors predicted child BMI 2 y later (Table 3). Higher levels of pressure to eat at ages 2 and 4 y predicted lower child BMI 2 y later even after accounting for potential confounders. Additional adjustment for child BMI at baseline resulted in strong attenuation of effect estimates with only the relation from ages 4 to 6 y remaining significant (b = 20.02; 95% CI: 20.04, 20.01). This result suggested that pressure to eat at age 4 y predicted child BMI at age 6 y independently of BMI at 4 y of age. Restriction at age 4 y predicted higher child BMI at age 6 y independently of confounding factors, but additional adjustment for child BMI at 4 y of age strongly attenuated effect estimates. An exploration of nonlinear effects showed a significant quadratic term of restriction on child BMI, which indicated that both low and high levels of restriction at age 4 y were related with high BMI at age 6 y (b-restriction quadratic term = 0.02; 95% CI: 0.01, 0.04). Again, the quadratic association was no longer observed after adjustment for confounders. With regard to parental monitoring, no effect on child BMI was observed in the adjusted model. Bidirectional associations between feeding practices and child BMI We examined path models to unravel whether each direction of the BMI-feeding association was independent and whether
4 of 8
JANSEN ET AL TABLE 1 Characteristics of the study population (n = 4166)1 Child characteristics Sex (M) (%) Age at CFQ assessment (y) Ethnicity Dutch (%) Other Western (%) Non-Western (%) BMI SD score at age 2 y3 BMI SD score at age 6 y Maternal characteristics Educational level ,3 y of secondary school (%) Secondary (%) Higher vocational (%) Academic (%) BMI (kg/m2) Pressuring feeding if 2-y-old child would not eat (score) CFQ Restriction scale (score) Monitoring scale (score) Pressure-to-eat scale (score)
n
Values
2071 4166
49.7 4.1 6 0.12
2757 404 983 3462 4166
66.5 9.8 23.7 0.22 6 1.0 0.20 6 0.87
320 1048 1159 1377 3731 3843
8.2 26.8 29.7 35.3 24.4 6 4.1 1.6 6 1.0
4166 4166 4166
23.8 6 6.2 13.2 6 2.4 12.5 6 3.9
1
Some data were missing for child ethnicity (n = 22), maternal educational level (n = 262), BMI (n = 450), and pressuring feeding if child would not eat (n = 323). CFQ, Child Feeding Questionnaire. 2 Mean 6 SD (all such values). 3 Mean BMI SD score at age 2 y with missing values imputed: 0.23 61.0.
directions differed in the strength of the effect. Results of path models are shown in Table 4 and graphically depicted in Figure 1. Child BMI at 2 y of age preceded higher levels of restriction (b = 0.08; 95% CI: 0.04, 0.11), which, in turn, predicted higher BMI at age 6 y (b = 0.03; 95% CI: 0.01, 0.06). This result suggested that the restriction–child BMI association had a bidirectional character, although the relation was strongest from BMI to restriction (Wald’s statistic for comparison = 5.0; P = 0.03). Moreover, in line with results shown in Table 3, additional adjustment of the relation of restriction to BMI for BMI at age 4 y attenuated the relation. For pressure to eat, a bidirectional association was shown whereby child BMI preceded lower levels of pressure (b = 20.21; 95% CI: 20.23, 20.17), and pressure to eat was negatively but less strongly associated with BMI at age 6 y (b = 20.07; 95% CI: 20.09, 20.05; Wald’s -statistic for comparison = 53.3, P , 0.001). Although significant in the unadjusted path model (data not presented in tables; b-monitoring to BMI = 20.08; 95% CI: 20.10, 20.05), associations between monitoring and child BMI attenuated to nonsignificance in the adjusted path model.
Finally, we estimated both directions of the pressure-to-eat– child BMI association within the same age period (from ages 2 to 4 y) in a cross-lagged model. The model presented in Figure 2 had a satisfactory model fit (comparative fit index: 0.96; root mean squared error of approximation: 0.042). There was considerable consistency between the 2 measures of pressure to eat (b = 0.17; 95% CI: 0.14, 0.20). The model showed that the strongest direction was from child BMI to parents’ use of pressure to eat (b = 20.20; 95% CI: 20.23, 20.17) even when we accounted for an association in the opposite direction at the same time (b = 20.01; 95% CI: 20.03, 0.02). DISCUSSION
Findings from this longitudinal population-based study indicate that, between ages 2 and 6 y, the main direction of the feeding–child-BMI association is from child BMI impacting on parenting. Although we showed some evidence that parents and children mutually influenced each other’s behavior in the context of eating, parents primarily adapted their controlling
TABLE 2 Child BMI at age 2 y and parental feeding behavior 2 y later (n = 4166)1 Feeding scales at age 4 y (expressed in SD scores) Child BMI at age 2 y (/SD) Unadjusted Confounder adjusted Further adjusted for pressuring feeding at 2 y of age
Restriction
P
Monitoring2
P
Pressure to eat
P
0.08 (0.04, 0.11) 0.07 (0.04, 0.10) 0.07 (0.04, 0.10)
,0.001 ,0.001 ,0.001
0.01 (20.03, 0.03) — —
0.97 — —
20.21 (20.24, 20.18) 20.21 (20.24, 20.18) 20.20 (20.23, 20.17)
,0.001 ,0.001 ,0.001
All values are bs; 95% CIs in parentheses. Values were derived from linear regression analyses. Confounders adjusted for were child age, child ethnicity, maternal education, and maternal BMI. 2 Adjusted models were not tested because the unadjusted BMI-monitoring association was not statistically significant. 1
5 of 8
BIDIRECTIONALITY BETWEEN FEEDING AND CHILD BMI TABLE 3 Parental feeding behavior and later child BMI (n = 4166)1 Feeding scales (/SD) BMI at age 4 y Feeding at age 2 y Pressuring feeding Unadjusted Confounder adjusted Further adjusted for child BMI at age 6 y Feeding at age 4 y Restriction Unadjusted Confounder adjusted Further adjusted for child Monitoring Unadjusted Confounder adjusted Further adjusted for child Pressure to eat Unadjusted Confounder adjusted Further adjusted for child
Child BMI SD score
P
20.04 (20.07, 20.01) 20.04 (20.08, 20.01) 20.01 (20.04, 0.01)
0.04 0.01 0.32
BMI at age 4 y
0.08 (0.05, 0.10) 0.07 (0.04, 0.09) 0.01 (20.01, 0.03)
,0.001 ,0.001 0.25 ,0.001 0.40
BMI at age 4 y2
20.07 (20.09, 20.04) 20.01 (20.04, 0.02) —
BMI at 4 y
20.13 (20.16, 20.10) 20.17 (20.19, 20.14) 20.02 (20.04, 20.01)
,0.001 ,0.001 0.01
BMI at 2 y
All values are bs; 95% CIs in parentheses. Values were derived from linear regression analyses. Confounders included child age, child ethnicity, maternal education, and maternal BMI. 2 Not tested because previous model was not statistically significant. 1
feeding behaviors to deviations in child weight rather than causing them.
understanding of the complex relation between parents’ feeding and child BMI.
Child BMI influencing parental feeding
Parental feeding influencing child BMI
Our different analytic approaches dovetailed in the indication that parental feeding practices were mostly a response of parents to their child’s BMI. Thus, parents’ pressure to eat was commonly used in the case of a low child weight, whereas parents tended to restrict children’s food intakes in response to relatively high BMI. To our knowledge, only 2 previous small prospective studies examined the possibility that child weight affects parental feeding strategies such as the pressure to eat or restriction (13, 18). Our findings for pressure to eat were consistent with the study of Webber et al (18) in late childhood and related work of Farrow and Blissett (13) that showed that a lower birth weight of infants elicited pressuring feeding of parents. Together, these findings suggest that the response of parents on children’s low weight is not age dependent but may reflect a general effect across the childhood years. Previous studies shoed no effect of child weight on parental restriction (13, 18), possibly because of limited statistical power to detect small differences as we did in our larger sample. The current study suggests that parents are sensitive to their child’s weight and try to normalize offspring weight development by regulating eating behaviors and food intake. However, findings may also reflect parents’ sensitivity to certain eating behaviors accompanying deviations in child weight. Because of the small number of studies that have examined this direction of the association, more research is necessary to confirm our findings from early childhood. Large longitudinal studies that cover birth to adolescence with frequent assessment waves will enable the examination of short- and long-term effects across different ages and can provide a comprehensive
In accordance with some previous studies on this topic (11–13), we showed that controlling feeding practices affected child weight 2 y later. The prospective restriction to BMI association disappeared when we accounted for children’s BMI at the same time that restriction was assessed. This result suggested that, within the particular age range from 4 to 6 y, the reported longitudinal association reflected a preexisting association at age 4 y from which we could not infer conclusions about directionality. We also showed no evidence for an optimal level of restrictive feeding. Thus, although scholars regularly posit that both a lack of restriction and overrestriction contribute to child obesity (5, 28), the current study did not support this hypothesis. In contrast to our findings, several earlier studies reported that parental pressure to eat was not associated with child BMI (14–16, 18). This inconsistency may be explained by our larger sample size, which allowed us to detect small effects of child BMI on parental pressure. Differences in findings may also be explained by dissimilar age ranges and follow-up periods. For instance, Rodgers et al (18) reported that parents’ pressure to eat in 2-y-olds was not related to child BMI at age 3 y, although it did predict more fussiness and less enjoyment of food. These results suggest that parental feeding impacts children’s eating behaviors, but the effect on BMI may not appear until later. The effect of pressure to eat on child weight may be explained in several ways. Intuitively, the reported negative association might suggest that the use of pressure to eat did not have the effect parents probably anticipated; parents mostly used pressure to eat to increase their child’s food intake and enhance weight gain. However, we could not rule out that parents prevented
0.08 (0.04, 0.11) 0.08 (0.04, 0.11)
0.01 (20.03, 0.03) —
20.21 (20.23, 20.17) 20.20 (20.23, 20.17)
,0.001
,0.001
0.60 (0.58, 0.62) NA
0.59 (0.56, 0.61) NA
For CFQ SD score
,0.001
P
0.60 (0.58, 0.62) NA
For BMI SD score
Child BMI at age 2 y to feeding at age 4 y2
,0.001 ,0.001
0.66
,0.001 ,0.001
P
20.07 (20.09, 20.05) 20.03 (20.05, 20.01)
20.02 (20.04, 0.01) —
0.03 (0.01, 0.06) 0.01 (20.01, 0.03)
For BMI SD score
0.28
0.011 0.32
P
,0.001 0.011
Feeding at age 4 y to child BMI at age 6 y2
53.3 98.8
1.0 —
5.0 13.1
Wald’s statistic
,0.001 ,0.001
0.32
0.03 ,0.001
P
Comparison of paths BMI to feeding and feeding to child BMI
1 All values are bs; 95% CIs in parentheses. Values (except from Wald’s statistic) were derived from linear regression analyses. Analyses are also presented in a more intuitive way in Figure 1. Covariates (child age, child ethnicity, maternal education, and maternal BMI) were included in the child-BMI to feeding regressions and feeding to child-BMI regressions. CFQ, Child Feeding Questionnaire; NA, not applicable. 2 Associations were simultaneously included in the same model. 3 Feeding strategies at child age 2 y and child BMI at age 4 y were further included as covariates.
Restriction Confounder adjusted Further adjusted for baseline levels of outcome3 Monitoring Confounder adjusted Further adjusted for baseline levels of outcome3 Pressure to eat Confounder adjusted Further adjusted for baseline levels of outcome3
Results per feeding scale (CFQ)
Stability effect: child BMI from ages 2 to 6 y2
TABLE 4 Path models including associations between parental feeding behavior and child BMI in both directions (n = 4166)1
6 of 8 JANSEN ET AL
BIDIRECTIONALITY BETWEEN FEEDING AND CHILD BMI
FIGURE 1. Path models including associations between parental feeding behavior and child BMI in both directions (n = 4166). Values represent bs derived from linear regression analyses as presented in Table 4. Paths were adjusted for covariates (child age, child ethnicity, maternal education, and maternal BMI) and feeding strategies at child age 2 y (only child-BMI to feeding regressions) and child BMI at age 4 y (only feeding to child-BMI regressions). *P , 0.05; ***P , 0.001.
a worsening of weight problems in their children. If parents had not pressured their children to eat, offspring’s BMI at follow-up may well have been lower than now observed. Another possible explanation is that controlling feeding practices have a counterproductive effect through the compromising of children’s development of self-regulation and inner responses to hunger and satiety cues (5, 6). Pressure to eat may also increase children’s adversity to eating (8). Evidence that eating behaviors and the regulation of food intake are largely shaped in preschool years and remain stable thereafter (20) may explain that we showed an effect of parental control on child BMI in early childhood, whereas some studies in older children reported no association (14, 18). Parental monitoring preceded lower child BMI at follow-up, but this relation was explained by sociodemographic variables and maternal BMI. Indeed, lower-educated, overweight mothers of immigrant background use different feeding practices than do mothers from more-affluent backgrounds (29), whereas their children also have highest risk of overweight because of a range of environmental risk factors (30, 31). Thus, the lack of accounting for confounding factors in some studies may explain mixed reports on the effect of parental monitoring (12–16, 18). Although the sociodemographic variables included in our study should be considered confounders, it was also possible that disadvantaged children’s overweight risk partly resulted from parents’ lack of monitoring. This possibility suggests that socially disadvantaged mothers may benefit from health education about the positive effect of guiding children’s food intakes combined with teaching these mothers how to better monitor offspring food intake. The current study was strengthened by its population-based design and a thorough examination of both directions of the feeding–
7 of 8
child-BMI association by using different approaches. Ideally, the CFQ would have been assessed repeatedly to enable an examination of both directions of the association within the same age window by using the same measure. Furthermore, the measurement of a broader spectrum of feeding beyond controlling feeding would have provided a more-comprehensive picture of parental influences in the context of eating. Also, the CFQ assesses only “overt” restrictive feeding (32) (ie, directly limiting food intake). Hence, we could not be certain that the findings can be generalized to more “covert” restriction (32) such as the whole family avoiding fast food restaurants or no unhealthy snacks being stocked in the home. Another limitation was that feeding practices were assessed by maternal report only. Finally, our nonresponse analyses indicated an underrepresentation of socially disadvantaged children who are at increased risk of overweight (30).This finding may have restricted our study’s external validity if the feeding-BMI associations differed between responding and nonresponding families. The current study indicated complex bidirectional associations between parental feeding and BMI in early childhood with a significantly stronger effect of child BMI on controlling feeding behaviors than vice versa. This result is in contrast with a recent study on general parenting in which we showed that parenting inconsistency predicted subsequent higher BMI but not the reverse in middle childhood (33). Thus, it seems that parents may adapt their specific feeding practices to children’s weight problems, whereas the adaptation of their more-general parenting behaviors is less likely. Although these studies indicated a modest impact of general and specific parenting behaviors, the influence of parents on child BMI is likely broader and involves different mechanisms including role modeling, family routines, and the provision of food and opportunities for physical activity. Several current pediatric overweight interventions include components that focus on reducing the level of control parents exert in food-related situations, such as discouraging the use of restrictive or coercive feeding practices (34, 35). These approaches need to be carefully scrutinized because our results suggest that such strategies may not have the desired effect because parents tend to adapt their controlling feeding practices in response to their child’s BMI rather than the reverse. More broadly, these results indicate that it is important to establish not only the size but temporal directions of all behaviors that predict child BMI before intervention research to change these behaviors is planned. Finally, we recommend intervention research into the more covert mealtime habits of families that have been
FIGURE 2. Cross-lagged model of associations between pressure to eat and child BMI (n = 4166). Values represent bs derived from linear regression analyses. Model-fit indexes indicated a good model fit (comparative fit index; 0.96, root mean squared error of approximation: 0.042. Covariates (child age, child ethnicity, maternal education, and maternal BMI) were included in child-BMI to pressure regressions and in pressure to child-BMI regressions. *P , 0.05; ***P , 0.001. CFQ, Child Feeding Questionnaire.
8 of 8
JANSEN ET AL
associated with healthy weight (36–38), which, at this point in time, look the most promising. The Generation R Study is conducted by the Erasmus MC–University Medical Centre Rotterdam in close collaboration with the Erasmus University Rotterdam, School of Law and Faculty of Social Sciences; the Municipal Health Service Rotterdam area, Rotterdam; the Rotterdam Homecare Foundation, Rotterdam; and the Stichting Trombosedienst & Artsenlaboratorium Rijnmond, Rotterdam. We gratefully acknowledge the contribution of the participating pregnant women and their partners, general practitioners, hospitals, midwives, and pharmacies in Rotterdam. The authors’ responsibilities were as follows—PWJ, FCV, MHvI, VWVJ, and HT: designed the current study; HR, AH, FCV, VWVJ, and HT: made important contributions to the conceptualization and design of the overall Generation R Study; PWJ and AT: conducted the study; VWVJ: provided essential materials by supervising the data collection within the Generation R Study; PWJ: analyzed data or performed statistical analyses and wrote the manuscript; JvdE: supervised the analysis of data and performance of statistical analyses; AT, JvdE, and MW: contributed to the interpretation of results; HT: supervised the writing of the manuscript; PWJ and HT: had primary responsibility for final content of the manuscript; and all authors: reviewed the manuscript for important intellectual content and approved the final manuscript as submitted. AT works at ErasmusAge, which is a research centre funded by Nestle´ Nutrition (Nestec Ltd), Metagenics Inc, and AXA. PWJ, JvdE, MW, HR, AH, FCV, MHvI, VWVJ, and HT had no conflicts of interest.
REFERENCES 1. Hurley KM, Cross MB, Hughes SO. A systematic review of responsive feeding and child obesity in high-income countries. J Nutr 2011;141: 495–501. 2. Jansen PW, Roza SJ, Jaddoe VW, Mackenbach JD, Raat H, Hofman A, Verhulst FC, Tiemeier H. Children’s eating behavior, feeding practices of parents and weight problems in early childhood: results from the population-based Generation R Study. Int J Behav Nutr Phys Act 2012;9:130. 3. Wardle J, Carnell S. Parental feeding practices and children’s weight. Acta Paediatr Suppl 2007;96:5–11. 4. Birch LL, Fisher JO, Grimm-Thomas K, Markey CN, Sawyer R, Johnson SL. Confirmatory factor analysis of the Child Feeding Questionnaire: a measure of parental attitudes, beliefs and practices about child feeding and obesity proneness. Appetite 2001;36:201–10. 5. Golan M, Bachner-Melman R. Self-regulation and the management of childhood obesity. J Comm Med Health Educ 2011;1:107. 6. Patrick H, Hennessy E, McSpadden K, Oh A. Parenting styles and practices in children’s obesogenic behaviors: scientific gaps and future research directions. Child Obes 2013;9(suppl):S73–86. 7. Fisher JO, Birch LL. Restricting access to palatable foods affects children’s behavioral response, food selection, and intake. Am J Clin Nutr 1999;69:1264–72. 8. Galloway AT, Fiorito LM, Francis LA, Birch LL. ’Finish your soup’: counterproductive effects of pressuring children to eat on intake and affect. Appetite 2006;46:318–23. 9. Jain A, Sherman SN, Chamberlin LA, Carter Y, Powers SW, Whitaker RC. Why don’t low-income mothers worry about their preschoolers being overweight? Pediatrics 2001;107:1138–46. 10. Wardle J, Sanderson S, Guthrie CA, Rapoport L, Plomin R. Parental feeding style and the inter-generational transmission of obesity risk. Obes Res 2002;10:453–62. 11. Campbell K, Andrianopoulos N, Hesketh K, Ball K, Crawford D, Brennan L, Corsini N, Timperio A. Parental use of restrictive feeding practices and child BMI z-score. A 3-year prospective cohort study. Appetite 2010;55:84–8. 12. Faith MS, Berkowitz RI, Stallings VA, Kerns J, Storey M, Stunkard AJ. Parental feeding attitudes and styles and child body mass index: prospective analysis of a gene-environment interaction. Pediatrics 2004; 114:e429–36. 13. Farrow CV, Blissett J. Controlling feeding practices: cause or consequence of early child weight? Pediatrics 2008;121:e164–9. 14. Spruijt-Metz D, Li C, Cohen E, Birch L, Goran M. Longitudinal influence of mother’s child-feeding practices on adiposity in children. J Pediatr 2006;148:314–20.
15. Gregory JE, Paxton SJ, Brozovic AM. Maternal feeding practices, child eating behaviour and body mass index in preschool-aged children: a prospective analysis. Int J Behav Nutr Phys Act 2010;7:55–65. 16. Rodgers RF, Paxton SJ, Massey R, Campbell KJ, Wertheim EH, Skouteris H, Gibbons K. Maternal feeding practices predict weight gain and obesogenic eating behaviors in young children: a prospective study. Int J Behav Nutr Phys Act 2013;10:24. 17. Black MM, Aboud FE. Responsive feeding is embedded in a theoretical framework of responsive parenting. J Nutr 2011;141:490–4. 18. Webber L, Cooke L, Hill C, Wardle J. Child adiposity and maternal feeding practices: a longitudinal analysis. Am J Clin Nutr 2010;92:1423–8. 19. Rhee KE, Coleman SM, Appugliese DP, Kaciroti NA, Corwyn RF, Davidson NS, Bradley RH, Lumeng JC. Maternal feeding practices become more controlling after and not before excessive rates of weight gain. Obesity (Silver Spring) 2009;17:1724–9. 20. Ashcroft J, Semmler C, Carnell S, van Jaarsveld CH, Wardle J. Continuity and stability of eating behaviour traits in children. Eur J Clin Nutr 2008;62:985–90. 21. Jaddoe VW, van Duijn CM, Franco OH, van der Heijden AJ, van Ijzendoorn MH, de Jongst JC, van der Lugt A, Mackenbach JP, Moll HA, Raat H, et al. The Generation R Study: design and cohort update 2012. Eur J Epidemiol 2012;27:739–56. 22. Tiemeier H, Velders FP, Szekely E, Roza SJ, Dieleman G, Jaddoe VWV, Uitterlinden AG, White TJ, Bakermans-Kranenburg MJ, Hofman A, et al. The Generation R Study: a review of design, findings to date, and a study of the 5-HTTLPR by environmental interaction from fetal life onward. J Am Acad Child Adolesc Psychiatry 2012;51:1119–35.e7. 23. Behling O, Law KS. Translating questionnaires and other research instruments: problems and solutions. Thousand Oaks, CA: Sage, 2000. 24. Farrow CV, Blissett JM. Is maternal psychopathology related to obesigenic feeding practices at 1 year? Obes Res 2005;13:1999–2005. 25. Fredriks AM, van Buuren S, Burgmeijer RJ, Meulmeester JF, Beuker RJ, Brugman E, Roede MJ, Verloove-Vanhorick SP, Wit JM. Continuing positive secular growth change in The Netherlands 1955-1997. Pediatr Res 2000;47:316–23. 26. van Jaarsveld CH, Llewellyn CH, Johnson L, Wardle J. Prospective associations between appetitive traits and weight gain in infancy. Am J Clin Nutr 2011;94:1562–7. 27. Spratt M, Carpenter J, Sterne JA, Carlin JB, Heron J, Henderson J, Tilling K. Strategies for multiple imputation in longitudinal studies. Am J Epidemiol 2010;172:478–87. 28. Kitzmann KM, Dalton WT, Buscemi J. Beyond parenting practices: family context and the treatment of pediatric obesity. Fam Relat 2008;57:13–23. 29. Berge JM. A review of familial correlates of child and adolescent obesity: what has the 21st century taught us so far? Int J Adolesc Med Health 2009;21:457–83. 30. Jansen PW, Mensah FK, Nicholson JM, Wake M. Family and neighbourhood socioeconomic inequalities in childhood trajectories of BMI and overweight: Longitudinal Study of Australian Children. PLoS ONE 2013;8:e69676. 31. Shrewsbury V, Wardle J. Socioeconomic status and adiposity in childhood: a systematic review of cross-sectional studies 1990-2005. Obesity (Silver Spring) 2008;16:275–84. 32. Ogden J, Reynolds R, Smith A. Expanding the concept of parental control: A role for overt and covert control in children’s snacking behaviour? Appetite 2006;47:100–6. 33. Jansen PW, Giallo R, Westrupp EM, Wake M, Nicholson JM: Bidirectional associations between mothers’ and fathers’ parenting consistency and child BMI. Pediatrics 2013;132:e1513–20. 34. Birch LL, Ventura AK. Preventing childhood obesity: what works? Int J Obes (Lond) 2009;33(suppl 1):S74–81. 35. Eneli IU, Crum PA, Tylka TL. The trust model: a different feeding paradigm for managing childhood obesity. Obesity (Silver Spring) 2008;16:2197–204. 36. Brown K, Ogden J, Gibson L, Vogele C. The role of parental control practices in explaining children’s diet and BMI. Appetite 2008;50:252–9. 37. Niemeier HM, Raynor HA, Lloyd-Richardson EE, Rogers ML, Wing RR. Fast food consumption and breakfast skipping: predictors of weight gain from adolescence to adulthood in a nationally representative sample. J Adolesc Health 2006;39:842–9. 38. Wansink B, van Kleef E. Dinner rituals that correlate with child and adult BMI. Obesity (Silver Spring 2014;22:E91–5.
Feeding practices and child weight: is the association bidirectional in preschool children?1–4 Pauline W Jansen, Anne Tharner, Jan van der Ende, Melissa Wake, Hein Raat, Albert Hofman, Frank C Verhulst, Marinus H van Ijzendoorn, Vincent WV Jaddoe, and Henning Tiemeier ABSTRACT Background: Parental feeding practices are associated with children’s body mass index (BMI). It has been generally assumed that parental feeding determines children’s eating behaviors and weight gain, but feeding practices could equally be a parent’s response to child weight. Objective: In longitudinal analyses, we assessed the directionality in the relation between selected controlling feeding practices and BMI in early childhood. Design: Participants were 4166 children from the population-based Generation R Study. BMI was measured at ages 2 and 6 y. With the use of the Child Feeding Questionnaire, parents reported on restriction, monitoring, and pressure to eat (child age: 4 y). BMI and feeding-behavior scales were transformed to SD scores. Results: With the use of linear regression analyses, there was an indication that a higher BMI at age 2 y predicted higher levels of parental restriction (adjusted b = 0.07; 95% CI: 0.04, 0.10) and lower levels of pressure to eat (adjusted b = 20.20; 95% CI: 20.23, 20.17) 2 y later. Restriction at age 4 y positively predicted child BMI at 6 y of age, although this association attenuated to statistical nonsignificance after accounting for BMI at age 4 y (b = 0.01; 95% CI: 20.01, 0.03). Pressure to eat predicted lower BMI independently of BMI at age 4 y (b = 20.02; 95% CI: 20.04, 20.01). For both restriction and pressure to eat, the relation from BMI to parenting was stronger than the reverse (Wald’s test for comparison: P = 0.03 and , 0.001, respectively). Monitoring predicted a lower child BMI, but this relation was explained by confounding factors. Conclusions: Although the feeding-BMI relation is bidirectional, the main direction of observed effects suggests that parents tend to adapt their controlling feeding practices in response to their child’s BMI rather than the reverse. Therefore, some components of current programs aimed at preventing or treating unhealthy child weight may need to be carefully scrutinized, especially those targeting parental food-related restriction and pressure to eat. Am J Clin Nutr doi: 10.3945/ajcn.114.088922.
INTRODUCTION
Feeding practices of parents have been implicated in children’s BMI and overweight (1–3). This proposition particularly relates to the use of control in food-related situations such as regulating the type and amount of food eaten by children (restricting) or cautiously keeping track of what children eat (monitoring) (4). Another controlling feeding strategy is parents’ use of pressure generally in an attempt to
improve the quantity or quality of children’s food intakes by using direct prompts and having rules about what a child should eat (4). It has been theorized that controlling feeding hinders children’s ability to self-regulate food consumption and override internal cues of satiety, thereby resulting in excessive food intake (5–8). Thus far, studies have tended to support these theories (5–8), but their laboratory-based designs and small sample sizes have limited generalizability. It is equally plausible that not only overcontrol but also a lack of parental control contribute to children’s weight gain (9, 10) because children are likely tempted to overeat in our current obesogenic environment. Few longitudinal studies showed support for this hypothesis. Poor parental restriction (11–13) and monitoring (12) of children’s food intakes were prospectively associated with a higher child BMI. Campbell et al (11) suggested that this relation depends on children’s age with the protective effect of restrictive feeding in middle childhood fading away as children enter adolescence and become more independent. This process is consistent with that shown in 1 From the Departments of Child and Adolescent Psychiatry/Psychology (PWJ, JvdE, FCV, and HT) and Pediatrics (VWVJ), Erasmus University Medical Center (Erasmus MC)-Sophia, Rotterdam, Netherlands; The Generation R Study Group (PWJ, AT, and VWVJ) and Departments of Epidemiology (AT, AH, VWVJ, and HT), Public Health (HR), and Psychiatry (HT), Erasmus MC-University Medical Center Rotterdam, Rotterdam, Netherlands; the Institute of Psychology (PWJ) and School of Pedagogical and Educational Sciences (MHvI), Erasmus University Rotterdam, Rotterdam, Netherlands; the Murdoch Childrens Research Institute, Melbourne, Australia (MW); the Department of Pediatrics, University of Melbourne, Melbourne, Australia (MW); and the Centre for Community Child Health, Royal Children’s Hospital, Melbourne, Australia (MW). 2 None of the funding entities were involved in the design of the study or analysis and interpretation of data. 3 Supported by the Netherlands Organization for Health Research and Development (ZonMW) (ZonMW “Geestkracht” program; grant 10.000.1003), the Netherlands Organization for Scientific Research (NOW) (NWO– ZonMW, VIDI grant 017.106.370; to HT), the Sophia Foundation for Medical Research SSWO (grant 602; to PWJ), and an Australian National Health and Medical Research Council Senior Research Fellowship (1046518; to MW). The Murdoch Childrens Research Institute is supported by the Victorian Government’s Operational Infrastructure Support Program. 4 Address reprint requests and correspondence to PW Jansen, Department of Child and Adolescent Psychiatry/Psychology, Erasmus University Medical Center-Sophia, PO Box 2060, 3000 CB Rotterdam, Netherlands. E-mail: [email protected]. Received March 26, 2014. Accepted for publication August 11, 2014. doi: 10.3945/ajcn.114.088922.
Am J Clin Nutr doi: 10.3945/ajcn.114.088922. Printed in USA. Ó 2014 American Society for Nutrition
Copyright (C) 2014 by the American Society for Nutrition
1 of 8
2 of 8
JANSEN ET AL
a study in adolescence (14) but contrasts with research in toddlers that showed no relation between restriction or monitoring and child weight at follow up (15, 16). Parents’ pressure on their children to eat more has also been studied longitudinally with more parental pressure to eat prospectively predicting lower child BMI (12, 13). None of these longitudinal studies tested whether there was an optimal effect of controlling feeding on child BMI (curvilinearity of associations). Previous studies mostly focused on parents’ influences on children’s weight development, whereas it is entirely possible that feeding-weight associations are bidirectional. Evidence-based theories of child development (eg, attachment theory and family systems theory) have suggested that parents generally respond and adapt their parenting strategies to children’s behavior and needs (17). Only 2 studies have explored a possible bidirectionality and reported that high BMI prospectively predicted more parental control (18, 19) and less pressure to eat (18), whereas no evidence was shown to support the reverse effect. These results contradict the previously described unidirectional studies that reported an influence of feeding practices on offspring BMI (11–13). The current study’s aim was to examine longitudinal relations between selected feeding practices and early childhood BMI in a large population based cohort. We expected bidirectional associations with weight influencing parental restriction, monitoring, and pressure, and these feeding practices would also affect weight gain. Moreover, we expected curvilinear relations to reflect that moderate levels of controlling feeding may be associated with the healthiest weight in children.
SUBJECTS AND METHODS
Design This study was embedded in Generation R, which is an ongoing population-based cohort from fetal life onwards (21, 22). Pregnant women who were living in Rotterdam, Netherlands, with an expected delivery date between April 2002 and January 2006 were invited to participate during pregnancy and after the birth of their child (participation rate: 61%). Written informed consent was obtained from all participating children and their parents, and the study was approved by the Medical Ethics Committee of the Erasmus University Medical Center, Rotterdam. Information was obtained by postal questionnaires and hands-on measurements at our own research center and Child Health Centers.
Measures Parental feeding practices Parents completed a postal questionnaire around the fourth birthday of their child which included 3 subscales of the Child Feeding Questionnaire (CFQ) (4). The original English questionnaire was translated by using a standard forward-backward translation method (23). Subscales of the CFQ were used to assess parental attitudes and strategies regarding the control of children’s eating as follows: restriction (8 items), monitoring (3 items), and pressure to eat (4 items). Examples of items were “I intentionally keep some foods out of my child’s reach” (restriction), “How much do you keep track of the high fat foods your child eats?” (monitoring), and “If my child says ‘I’m not hungry,’ I’ll try to get him/her to eat anyway” (pressure to eat). Parents, in most cases the mothers (86.5%), answered the items on a 5-point Likert scale from 1 (never) to 5 (always). Scale scores were only calculated if $75% of items were completed. Continuous scale scores were transformed into SD scores to facilitate effect-size comparisons between scales. Research has provided good evidence for the concurrent validity of the CFQ with actual observations of feeding behaviors of mothers (24). The internal consistency of the administered CFQ scales was moderate for pressure to eat (a = 0.66) and restriction (a = 0.73) and high for monitoring (a = 0.92). We also assessed pressuring feeding at child age 2 y. This feeding practice was used in a cross-lagged modeling approach to estimate both directions of the pressure-to-eat–child-BMI association within the same age period. With the use of the following 5 self-report items, parents (mostly mothers) indicated what they would do if their 2-y-old child did not want to eat: 1) try to convince child to eat, 2) allow child to stop eating fairly quickly (reverse coded), 3) deprive child of something he or she likes (eg, pudding or television watching), 4) child may not leave table until plate is empty, and 5) negotiate with child (if . then .). These items (yes = 1, no = 0) were summed with higher sum scores indicating more pressuring feeding. The internal consistency of this scale was low (a = 0.29), which was not surprising because of the small number of items and the fact that the scale has not been validated. However, the concordance correlation coefficient between the 2 measures of pressure to eat at ages 2 and 4 y was rc = 0.21, which indicated that, despite the different assessment methods and 2-y interval between assessments, these 2 measures assessed related constructs. BMI
Study population Full consent for the postnatal phase of the Generation R Study was obtained for 7295 children and their parents. For the current study, we excluded children without BMI assessed at age 6 y (n = 1801; 24.7%) and those without information on parental feeding practices at age 4 y (n = 1328; 18.2%), which yielded a study population of 4166 children (57.1%). A comparison of included (n = 4166) and excluded (n = 3129) children indicated that data were more often missing in children of lower-educated mothers and with a non-Dutch background (both P , 0.001). No sex differences were shown between children with and without missing data (P = 0.13).
Children’s growth characteristics were obtained at ages 1, 2, 3, and 4 y at the municipal Child Health Centers as part of a routine health care program and at 6 y at the Generation R research center. At both centers, trained staff measured children’s weights and heights by using standardized procedures. Weight was measured by using a mechanical personal scale (SECA) while children were wearing underwear only. Height was measured in standing position by using a Harpenden stadiometer (Holtain Ltd). BMI was calculated as weight divided by height squared. Age- and sex-specific BMI SD scores were calculated by using the Dutch reference population (25) in the Growth Analyzer program.
BIDIRECTIONALITY BETWEEN FEEDING AND CHILD BMI
Covariates Several demographic variables were considered possible confounding factors. Child factors included sex, age at CFQ assessment, and national origin, which was based on the country of birth of both parents. Maternal variables included BMI measured at study intake in early pregnancy and self-reported highest attained educational level (elementary or secondary, higher vocational, and academic), which was considered as a proxy for socioeconomic status. Statistical analyses In all analyses, child BMI and feeding scales were expressed in SD scores. Two sets of linear regression analyses were conducted; each set comprised 3 separate regressions, one for each CFQ scale (ie, restriction, monitoring, and pressure to eat). In the first set, we examined the relation between child BMI at age 2 y and feeding practices at 4 y of age. In the second set, we examined whether feeding practices at ages 2 and 4 y of age predicted child BMI at 4 and 6 y of age, respectively. For both sets of regressions, we present 3 models as follows: 1) unadjusted, 2) adjusted for sociodemographic covariates, and 3) further adjusted for the measure of pressuring feeding practices at age 2 y (first set of regressions) or for baseline BMI at age 2 or 4 y (second set of regressions). Covariates were only included in adjusted models if they changed effect estimates of the unadjusted CFQ-BMI association .5%. With the use of this strategy, all previously mentioned covariates were included in the adjusted model, except for child sex which did not change the effect estimate .5%. In additional analyses, we tested whether feeding-BMI relations were curvilinear by adding a quadratic term to models. Following preceding research (26), weconductedpathanalyses to estimate which direction of CFQ-BMI associations was strongest. We tested 3 path models, again for each feeding scale separately. Each path model included multiple linear regression jointly estimated CFQ-BMI associations in both directions while accounting for continuityinBMI over time. With the inclusion of bothdirections of theCFQ-BMIassociation inone model,pathways were accounted for eachother and coulddirectlybe compared instrength. Wald’s test was used to compare whether the BMI-CFQ association differed statistically from the CFQ-BMI association. Finally, we applied a cross-lagged modeling approach to estimate both directions of the pressure-to-eat–child BMI association within the same age period (from ages 2 to 4 y). We could only test such a model for the pressure to eat because parental restriction and monitoring were not assessed at age 2 y. The model included multiple linear regressions that reflected pressure-to-eat– BMI associations (from ages 2 to 4 and 4 to 6 y), BMI at 2 y of age to pressure to eat at 4 y of age, and continuity in BMI and pressure to eat over time. The model also included cross-sectional correlations between pressure to eat and BMI and was adjusted for the previously mentioned covariates. An acceptable-to-good fit was determined by using a comparative fit index .0.90, and a root mean squared error of approximation ,0.08. Multiple imputation techniques (full conditional specification) were used to account for missing values in baseline BMI (at 2 y of age) and covariates. Imputations were based on available information on all variables included in the study as well as BMI assessments at 1 and 3 y of age (27). Analyses were performed on imputed data, and reported effect estimates were pooled results of 20 imputed data sets. Analyses were performed with STATA 12.0
3 of 8
software (Stata Corp) except for path analyses, which were conducted with Mplus version 7.11 software (Muthe´n & Muthe´n). RESULTS
Characteristics of children and their parents are presented in Table 1. The majority of children included in the study population (68%) were of Dutch origin. Mothers were relatively highly educated with 65% of them having a higher vocational or academic training. Mean (6SD) BMI (in kg/m2) of mothers was 24.5 6 4.1. Child BMI influencing parental feeding In the first set of linear regression analyses, we examined whether child BMI at 2 y preceded feeding behavior of parents at child age 4 y (Table 2). A higher BMI at age 2 y was associated with higher levels of parental restriction and lower levels of pressure to eat 2 y later. These associations were not explained by possible confounding factors nor by pressuring feeding practices at age 2 y [restriction: adjusted b = 0.07 (95% CI: 0.04, 0.10); pressure to eat: adjusted b = 20.20 (95% CI: 20.23, 20.17)]. The final model suggested that the relation from BMI to restriction and pressure to eat was not attributable to any feeding-BMI association at age 2 y. Child BMI at age 2 y did not predict later parental monitoring, although the addition of a quadratic term of monitoring to the model pointed at a significant curvilinear association whereby both a low and high BMI at child age 2 y predicted lower levels of monitoring at age 4 (b-BMI quadratic term = 20.024; 95% CI: 20.04, 20.01). However, this association was attenuated by accounting for confounding factors. Parental feeding influencing child BMI In the second set of linear regression analyses, we examined the reverse association of whether parents’ feeding behaviors predicted child BMI 2 y later (Table 3). Higher levels of pressure to eat at ages 2 and 4 y predicted lower child BMI 2 y later even after accounting for potential confounders. Additional adjustment for child BMI at baseline resulted in strong attenuation of effect estimates with only the relation from ages 4 to 6 y remaining significant (b = 20.02; 95% CI: 20.04, 20.01). This result suggested that pressure to eat at age 4 y predicted child BMI at age 6 y independently of BMI at 4 y of age. Restriction at age 4 y predicted higher child BMI at age 6 y independently of confounding factors, but additional adjustment for child BMI at 4 y of age strongly attenuated effect estimates. An exploration of nonlinear effects showed a significant quadratic term of restriction on child BMI, which indicated that both low and high levels of restriction at age 4 y were related with high BMI at age 6 y (b-restriction quadratic term = 0.02; 95% CI: 0.01, 0.04). Again, the quadratic association was no longer observed after adjustment for confounders. With regard to parental monitoring, no effect on child BMI was observed in the adjusted model. Bidirectional associations between feeding practices and child BMI We examined path models to unravel whether each direction of the BMI-feeding association was independent and whether
4 of 8
JANSEN ET AL TABLE 1 Characteristics of the study population (n = 4166)1 Child characteristics Sex (M) (%) Age at CFQ assessment (y) Ethnicity Dutch (%) Other Western (%) Non-Western (%) BMI SD score at age 2 y3 BMI SD score at age 6 y Maternal characteristics Educational level ,3 y of secondary school (%) Secondary (%) Higher vocational (%) Academic (%) BMI (kg/m2) Pressuring feeding if 2-y-old child would not eat (score) CFQ Restriction scale (score) Monitoring scale (score) Pressure-to-eat scale (score)
n
Values
2071 4166
49.7 4.1 6 0.12
2757 404 983 3462 4166
66.5 9.8 23.7 0.22 6 1.0 0.20 6 0.87
320 1048 1159 1377 3731 3843
8.2 26.8 29.7 35.3 24.4 6 4.1 1.6 6 1.0
4166 4166 4166
23.8 6 6.2 13.2 6 2.4 12.5 6 3.9
1
Some data were missing for child ethnicity (n = 22), maternal educational level (n = 262), BMI (n = 450), and pressuring feeding if child would not eat (n = 323). CFQ, Child Feeding Questionnaire. 2 Mean 6 SD (all such values). 3 Mean BMI SD score at age 2 y with missing values imputed: 0.23 61.0.
directions differed in the strength of the effect. Results of path models are shown in Table 4 and graphically depicted in Figure 1. Child BMI at 2 y of age preceded higher levels of restriction (b = 0.08; 95% CI: 0.04, 0.11), which, in turn, predicted higher BMI at age 6 y (b = 0.03; 95% CI: 0.01, 0.06). This result suggested that the restriction–child BMI association had a bidirectional character, although the relation was strongest from BMI to restriction (Wald’s statistic for comparison = 5.0; P = 0.03). Moreover, in line with results shown in Table 3, additional adjustment of the relation of restriction to BMI for BMI at age 4 y attenuated the relation. For pressure to eat, a bidirectional association was shown whereby child BMI preceded lower levels of pressure (b = 20.21; 95% CI: 20.23, 20.17), and pressure to eat was negatively but less strongly associated with BMI at age 6 y (b = 20.07; 95% CI: 20.09, 20.05; Wald’s -statistic for comparison = 53.3, P , 0.001). Although significant in the unadjusted path model (data not presented in tables; b-monitoring to BMI = 20.08; 95% CI: 20.10, 20.05), associations between monitoring and child BMI attenuated to nonsignificance in the adjusted path model.
Finally, we estimated both directions of the pressure-to-eat– child BMI association within the same age period (from ages 2 to 4 y) in a cross-lagged model. The model presented in Figure 2 had a satisfactory model fit (comparative fit index: 0.96; root mean squared error of approximation: 0.042). There was considerable consistency between the 2 measures of pressure to eat (b = 0.17; 95% CI: 0.14, 0.20). The model showed that the strongest direction was from child BMI to parents’ use of pressure to eat (b = 20.20; 95% CI: 20.23, 20.17) even when we accounted for an association in the opposite direction at the same time (b = 20.01; 95% CI: 20.03, 0.02). DISCUSSION
Findings from this longitudinal population-based study indicate that, between ages 2 and 6 y, the main direction of the feeding–child-BMI association is from child BMI impacting on parenting. Although we showed some evidence that parents and children mutually influenced each other’s behavior in the context of eating, parents primarily adapted their controlling
TABLE 2 Child BMI at age 2 y and parental feeding behavior 2 y later (n = 4166)1 Feeding scales at age 4 y (expressed in SD scores) Child BMI at age 2 y (/SD) Unadjusted Confounder adjusted Further adjusted for pressuring feeding at 2 y of age
Restriction
P
Monitoring2
P
Pressure to eat
P
0.08 (0.04, 0.11) 0.07 (0.04, 0.10) 0.07 (0.04, 0.10)
,0.001 ,0.001 ,0.001
0.01 (20.03, 0.03) — —
0.97 — —
20.21 (20.24, 20.18) 20.21 (20.24, 20.18) 20.20 (20.23, 20.17)
,0.001 ,0.001 ,0.001
All values are bs; 95% CIs in parentheses. Values were derived from linear regression analyses. Confounders adjusted for were child age, child ethnicity, maternal education, and maternal BMI. 2 Adjusted models were not tested because the unadjusted BMI-monitoring association was not statistically significant. 1
5 of 8
BIDIRECTIONALITY BETWEEN FEEDING AND CHILD BMI TABLE 3 Parental feeding behavior and later child BMI (n = 4166)1 Feeding scales (/SD) BMI at age 4 y Feeding at age 2 y Pressuring feeding Unadjusted Confounder adjusted Further adjusted for child BMI at age 6 y Feeding at age 4 y Restriction Unadjusted Confounder adjusted Further adjusted for child Monitoring Unadjusted Confounder adjusted Further adjusted for child Pressure to eat Unadjusted Confounder adjusted Further adjusted for child
Child BMI SD score
P
20.04 (20.07, 20.01) 20.04 (20.08, 20.01) 20.01 (20.04, 0.01)
0.04 0.01 0.32
BMI at age 4 y
0.08 (0.05, 0.10) 0.07 (0.04, 0.09) 0.01 (20.01, 0.03)
,0.001 ,0.001 0.25 ,0.001 0.40
BMI at age 4 y2
20.07 (20.09, 20.04) 20.01 (20.04, 0.02) —
BMI at 4 y
20.13 (20.16, 20.10) 20.17 (20.19, 20.14) 20.02 (20.04, 20.01)
,0.001 ,0.001 0.01
BMI at 2 y
All values are bs; 95% CIs in parentheses. Values were derived from linear regression analyses. Confounders included child age, child ethnicity, maternal education, and maternal BMI. 2 Not tested because previous model was not statistically significant. 1
feeding behaviors to deviations in child weight rather than causing them.
understanding of the complex relation between parents’ feeding and child BMI.
Child BMI influencing parental feeding
Parental feeding influencing child BMI
Our different analytic approaches dovetailed in the indication that parental feeding practices were mostly a response of parents to their child’s BMI. Thus, parents’ pressure to eat was commonly used in the case of a low child weight, whereas parents tended to restrict children’s food intakes in response to relatively high BMI. To our knowledge, only 2 previous small prospective studies examined the possibility that child weight affects parental feeding strategies such as the pressure to eat or restriction (13, 18). Our findings for pressure to eat were consistent with the study of Webber et al (18) in late childhood and related work of Farrow and Blissett (13) that showed that a lower birth weight of infants elicited pressuring feeding of parents. Together, these findings suggest that the response of parents on children’s low weight is not age dependent but may reflect a general effect across the childhood years. Previous studies shoed no effect of child weight on parental restriction (13, 18), possibly because of limited statistical power to detect small differences as we did in our larger sample. The current study suggests that parents are sensitive to their child’s weight and try to normalize offspring weight development by regulating eating behaviors and food intake. However, findings may also reflect parents’ sensitivity to certain eating behaviors accompanying deviations in child weight. Because of the small number of studies that have examined this direction of the association, more research is necessary to confirm our findings from early childhood. Large longitudinal studies that cover birth to adolescence with frequent assessment waves will enable the examination of short- and long-term effects across different ages and can provide a comprehensive
In accordance with some previous studies on this topic (11–13), we showed that controlling feeding practices affected child weight 2 y later. The prospective restriction to BMI association disappeared when we accounted for children’s BMI at the same time that restriction was assessed. This result suggested that, within the particular age range from 4 to 6 y, the reported longitudinal association reflected a preexisting association at age 4 y from which we could not infer conclusions about directionality. We also showed no evidence for an optimal level of restrictive feeding. Thus, although scholars regularly posit that both a lack of restriction and overrestriction contribute to child obesity (5, 28), the current study did not support this hypothesis. In contrast to our findings, several earlier studies reported that parental pressure to eat was not associated with child BMI (14–16, 18). This inconsistency may be explained by our larger sample size, which allowed us to detect small effects of child BMI on parental pressure. Differences in findings may also be explained by dissimilar age ranges and follow-up periods. For instance, Rodgers et al (18) reported that parents’ pressure to eat in 2-y-olds was not related to child BMI at age 3 y, although it did predict more fussiness and less enjoyment of food. These results suggest that parental feeding impacts children’s eating behaviors, but the effect on BMI may not appear until later. The effect of pressure to eat on child weight may be explained in several ways. Intuitively, the reported negative association might suggest that the use of pressure to eat did not have the effect parents probably anticipated; parents mostly used pressure to eat to increase their child’s food intake and enhance weight gain. However, we could not rule out that parents prevented
0.08 (0.04, 0.11) 0.08 (0.04, 0.11)
0.01 (20.03, 0.03) —
20.21 (20.23, 20.17) 20.20 (20.23, 20.17)
,0.001
,0.001
0.60 (0.58, 0.62) NA
0.59 (0.56, 0.61) NA
For CFQ SD score
,0.001
P
0.60 (0.58, 0.62) NA
For BMI SD score
Child BMI at age 2 y to feeding at age 4 y2
,0.001 ,0.001
0.66
,0.001 ,0.001
P
20.07 (20.09, 20.05) 20.03 (20.05, 20.01)
20.02 (20.04, 0.01) —
0.03 (0.01, 0.06) 0.01 (20.01, 0.03)
For BMI SD score
0.28
0.011 0.32
P
,0.001 0.011
Feeding at age 4 y to child BMI at age 6 y2
53.3 98.8
1.0 —
5.0 13.1
Wald’s statistic
,0.001 ,0.001
0.32
0.03 ,0.001
P
Comparison of paths BMI to feeding and feeding to child BMI
1 All values are bs; 95% CIs in parentheses. Values (except from Wald’s statistic) were derived from linear regression analyses. Analyses are also presented in a more intuitive way in Figure 1. Covariates (child age, child ethnicity, maternal education, and maternal BMI) were included in the child-BMI to feeding regressions and feeding to child-BMI regressions. CFQ, Child Feeding Questionnaire; NA, not applicable. 2 Associations were simultaneously included in the same model. 3 Feeding strategies at child age 2 y and child BMI at age 4 y were further included as covariates.
Restriction Confounder adjusted Further adjusted for baseline levels of outcome3 Monitoring Confounder adjusted Further adjusted for baseline levels of outcome3 Pressure to eat Confounder adjusted Further adjusted for baseline levels of outcome3
Results per feeding scale (CFQ)
Stability effect: child BMI from ages 2 to 6 y2
TABLE 4 Path models including associations between parental feeding behavior and child BMI in both directions (n = 4166)1
6 of 8 JANSEN ET AL
BIDIRECTIONALITY BETWEEN FEEDING AND CHILD BMI
FIGURE 1. Path models including associations between parental feeding behavior and child BMI in both directions (n = 4166). Values represent bs derived from linear regression analyses as presented in Table 4. Paths were adjusted for covariates (child age, child ethnicity, maternal education, and maternal BMI) and feeding strategies at child age 2 y (only child-BMI to feeding regressions) and child BMI at age 4 y (only feeding to child-BMI regressions). *P , 0.05; ***P , 0.001.
a worsening of weight problems in their children. If parents had not pressured their children to eat, offspring’s BMI at follow-up may well have been lower than now observed. Another possible explanation is that controlling feeding practices have a counterproductive effect through the compromising of children’s development of self-regulation and inner responses to hunger and satiety cues (5, 6). Pressure to eat may also increase children’s adversity to eating (8). Evidence that eating behaviors and the regulation of food intake are largely shaped in preschool years and remain stable thereafter (20) may explain that we showed an effect of parental control on child BMI in early childhood, whereas some studies in older children reported no association (14, 18). Parental monitoring preceded lower child BMI at follow-up, but this relation was explained by sociodemographic variables and maternal BMI. Indeed, lower-educated, overweight mothers of immigrant background use different feeding practices than do mothers from more-affluent backgrounds (29), whereas their children also have highest risk of overweight because of a range of environmental risk factors (30, 31). Thus, the lack of accounting for confounding factors in some studies may explain mixed reports on the effect of parental monitoring (12–16, 18). Although the sociodemographic variables included in our study should be considered confounders, it was also possible that disadvantaged children’s overweight risk partly resulted from parents’ lack of monitoring. This possibility suggests that socially disadvantaged mothers may benefit from health education about the positive effect of guiding children’s food intakes combined with teaching these mothers how to better monitor offspring food intake. The current study was strengthened by its population-based design and a thorough examination of both directions of the feeding–
7 of 8
child-BMI association by using different approaches. Ideally, the CFQ would have been assessed repeatedly to enable an examination of both directions of the association within the same age window by using the same measure. Furthermore, the measurement of a broader spectrum of feeding beyond controlling feeding would have provided a more-comprehensive picture of parental influences in the context of eating. Also, the CFQ assesses only “overt” restrictive feeding (32) (ie, directly limiting food intake). Hence, we could not be certain that the findings can be generalized to more “covert” restriction (32) such as the whole family avoiding fast food restaurants or no unhealthy snacks being stocked in the home. Another limitation was that feeding practices were assessed by maternal report only. Finally, our nonresponse analyses indicated an underrepresentation of socially disadvantaged children who are at increased risk of overweight (30).This finding may have restricted our study’s external validity if the feeding-BMI associations differed between responding and nonresponding families. The current study indicated complex bidirectional associations between parental feeding and BMI in early childhood with a significantly stronger effect of child BMI on controlling feeding behaviors than vice versa. This result is in contrast with a recent study on general parenting in which we showed that parenting inconsistency predicted subsequent higher BMI but not the reverse in middle childhood (33). Thus, it seems that parents may adapt their specific feeding practices to children’s weight problems, whereas the adaptation of their more-general parenting behaviors is less likely. Although these studies indicated a modest impact of general and specific parenting behaviors, the influence of parents on child BMI is likely broader and involves different mechanisms including role modeling, family routines, and the provision of food and opportunities for physical activity. Several current pediatric overweight interventions include components that focus on reducing the level of control parents exert in food-related situations, such as discouraging the use of restrictive or coercive feeding practices (34, 35). These approaches need to be carefully scrutinized because our results suggest that such strategies may not have the desired effect because parents tend to adapt their controlling feeding practices in response to their child’s BMI rather than the reverse. More broadly, these results indicate that it is important to establish not only the size but temporal directions of all behaviors that predict child BMI before intervention research to change these behaviors is planned. Finally, we recommend intervention research into the more covert mealtime habits of families that have been
FIGURE 2. Cross-lagged model of associations between pressure to eat and child BMI (n = 4166). Values represent bs derived from linear regression analyses. Model-fit indexes indicated a good model fit (comparative fit index; 0.96, root mean squared error of approximation: 0.042. Covariates (child age, child ethnicity, maternal education, and maternal BMI) were included in child-BMI to pressure regressions and in pressure to child-BMI regressions. *P , 0.05; ***P , 0.001. CFQ, Child Feeding Questionnaire.
8 of 8
JANSEN ET AL
associated with healthy weight (36–38), which, at this point in time, look the most promising. The Generation R Study is conducted by the Erasmus MC–University Medical Centre Rotterdam in close collaboration with the Erasmus University Rotterdam, School of Law and Faculty of Social Sciences; the Municipal Health Service Rotterdam area, Rotterdam; the Rotterdam Homecare Foundation, Rotterdam; and the Stichting Trombosedienst & Artsenlaboratorium Rijnmond, Rotterdam. We gratefully acknowledge the contribution of the participating pregnant women and their partners, general practitioners, hospitals, midwives, and pharmacies in Rotterdam. The authors’ responsibilities were as follows—PWJ, FCV, MHvI, VWVJ, and HT: designed the current study; HR, AH, FCV, VWVJ, and HT: made important contributions to the conceptualization and design of the overall Generation R Study; PWJ and AT: conducted the study; VWVJ: provided essential materials by supervising the data collection within the Generation R Study; PWJ: analyzed data or performed statistical analyses and wrote the manuscript; JvdE: supervised the analysis of data and performance of statistical analyses; AT, JvdE, and MW: contributed to the interpretation of results; HT: supervised the writing of the manuscript; PWJ and HT: had primary responsibility for final content of the manuscript; and all authors: reviewed the manuscript for important intellectual content and approved the final manuscript as submitted. AT works at ErasmusAge, which is a research centre funded by Nestle´ Nutrition (Nestec Ltd), Metagenics Inc, and AXA. PWJ, JvdE, MW, HR, AH, FCV, MHvI, VWVJ, and HT had no conflicts of interest.
REFERENCES 1. Hurley KM, Cross MB, Hughes SO. A systematic review of responsive feeding and child obesity in high-income countries. J Nutr 2011;141: 495–501. 2. Jansen PW, Roza SJ, Jaddoe VW, Mackenbach JD, Raat H, Hofman A, Verhulst FC, Tiemeier H. Children’s eating behavior, feeding practices of parents and weight problems in early childhood: results from the population-based Generation R Study. Int J Behav Nutr Phys Act 2012;9:130. 3. Wardle J, Carnell S. Parental feeding practices and children’s weight. Acta Paediatr Suppl 2007;96:5–11. 4. Birch LL, Fisher JO, Grimm-Thomas K, Markey CN, Sawyer R, Johnson SL. Confirmatory factor analysis of the Child Feeding Questionnaire: a measure of parental attitudes, beliefs and practices about child feeding and obesity proneness. Appetite 2001;36:201–10. 5. Golan M, Bachner-Melman R. Self-regulation and the management of childhood obesity. J Comm Med Health Educ 2011;1:107. 6. Patrick H, Hennessy E, McSpadden K, Oh A. Parenting styles and practices in children’s obesogenic behaviors: scientific gaps and future research directions. Child Obes 2013;9(suppl):S73–86. 7. Fisher JO, Birch LL. Restricting access to palatable foods affects children’s behavioral response, food selection, and intake. Am J Clin Nutr 1999;69:1264–72. 8. Galloway AT, Fiorito LM, Francis LA, Birch LL. ’Finish your soup’: counterproductive effects of pressuring children to eat on intake and affect. Appetite 2006;46:318–23. 9. Jain A, Sherman SN, Chamberlin LA, Carter Y, Powers SW, Whitaker RC. Why don’t low-income mothers worry about their preschoolers being overweight? Pediatrics 2001;107:1138–46. 10. Wardle J, Sanderson S, Guthrie CA, Rapoport L, Plomin R. Parental feeding style and the inter-generational transmission of obesity risk. Obes Res 2002;10:453–62. 11. Campbell K, Andrianopoulos N, Hesketh K, Ball K, Crawford D, Brennan L, Corsini N, Timperio A. Parental use of restrictive feeding practices and child BMI z-score. A 3-year prospective cohort study. Appetite 2010;55:84–8. 12. Faith MS, Berkowitz RI, Stallings VA, Kerns J, Storey M, Stunkard AJ. Parental feeding attitudes and styles and child body mass index: prospective analysis of a gene-environment interaction. Pediatrics 2004; 114:e429–36. 13. Farrow CV, Blissett J. Controlling feeding practices: cause or consequence of early child weight? Pediatrics 2008;121:e164–9. 14. Spruijt-Metz D, Li C, Cohen E, Birch L, Goran M. Longitudinal influence of mother’s child-feeding practices on adiposity in children. J Pediatr 2006;148:314–20.
15. Gregory JE, Paxton SJ, Brozovic AM. Maternal feeding practices, child eating behaviour and body mass index in preschool-aged children: a prospective analysis. Int J Behav Nutr Phys Act 2010;7:55–65. 16. Rodgers RF, Paxton SJ, Massey R, Campbell KJ, Wertheim EH, Skouteris H, Gibbons K. Maternal feeding practices predict weight gain and obesogenic eating behaviors in young children: a prospective study. Int J Behav Nutr Phys Act 2013;10:24. 17. Black MM, Aboud FE. Responsive feeding is embedded in a theoretical framework of responsive parenting. J Nutr 2011;141:490–4. 18. Webber L, Cooke L, Hill C, Wardle J. Child adiposity and maternal feeding practices: a longitudinal analysis. Am J Clin Nutr 2010;92:1423–8. 19. Rhee KE, Coleman SM, Appugliese DP, Kaciroti NA, Corwyn RF, Davidson NS, Bradley RH, Lumeng JC. Maternal feeding practices become more controlling after and not before excessive rates of weight gain. Obesity (Silver Spring) 2009;17:1724–9. 20. Ashcroft J, Semmler C, Carnell S, van Jaarsveld CH, Wardle J. Continuity and stability of eating behaviour traits in children. Eur J Clin Nutr 2008;62:985–90. 21. Jaddoe VW, van Duijn CM, Franco OH, van der Heijden AJ, van Ijzendoorn MH, de Jongst JC, van der Lugt A, Mackenbach JP, Moll HA, Raat H, et al. The Generation R Study: design and cohort update 2012. Eur J Epidemiol 2012;27:739–56. 22. Tiemeier H, Velders FP, Szekely E, Roza SJ, Dieleman G, Jaddoe VWV, Uitterlinden AG, White TJ, Bakermans-Kranenburg MJ, Hofman A, et al. The Generation R Study: a review of design, findings to date, and a study of the 5-HTTLPR by environmental interaction from fetal life onward. J Am Acad Child Adolesc Psychiatry 2012;51:1119–35.e7. 23. Behling O, Law KS. Translating questionnaires and other research instruments: problems and solutions. Thousand Oaks, CA: Sage, 2000. 24. Farrow CV, Blissett JM. Is maternal psychopathology related to obesigenic feeding practices at 1 year? Obes Res 2005;13:1999–2005. 25. Fredriks AM, van Buuren S, Burgmeijer RJ, Meulmeester JF, Beuker RJ, Brugman E, Roede MJ, Verloove-Vanhorick SP, Wit JM. Continuing positive secular growth change in The Netherlands 1955-1997. Pediatr Res 2000;47:316–23. 26. van Jaarsveld CH, Llewellyn CH, Johnson L, Wardle J. Prospective associations between appetitive traits and weight gain in infancy. Am J Clin Nutr 2011;94:1562–7. 27. Spratt M, Carpenter J, Sterne JA, Carlin JB, Heron J, Henderson J, Tilling K. Strategies for multiple imputation in longitudinal studies. Am J Epidemiol 2010;172:478–87. 28. Kitzmann KM, Dalton WT, Buscemi J. Beyond parenting practices: family context and the treatment of pediatric obesity. Fam Relat 2008;57:13–23. 29. Berge JM. A review of familial correlates of child and adolescent obesity: what has the 21st century taught us so far? Int J Adolesc Med Health 2009;21:457–83. 30. Jansen PW, Mensah FK, Nicholson JM, Wake M. Family and neighbourhood socioeconomic inequalities in childhood trajectories of BMI and overweight: Longitudinal Study of Australian Children. PLoS ONE 2013;8:e69676. 31. Shrewsbury V, Wardle J. Socioeconomic status and adiposity in childhood: a systematic review of cross-sectional studies 1990-2005. Obesity (Silver Spring) 2008;16:275–84. 32. Ogden J, Reynolds R, Smith A. Expanding the concept of parental control: A role for overt and covert control in children’s snacking behaviour? Appetite 2006;47:100–6. 33. Jansen PW, Giallo R, Westrupp EM, Wake M, Nicholson JM: Bidirectional associations between mothers’ and fathers’ parenting consistency and child BMI. Pediatrics 2013;132:e1513–20. 34. Birch LL, Ventura AK. Preventing childhood obesity: what works? Int J Obes (Lond) 2009;33(suppl 1):S74–81. 35. Eneli IU, Crum PA, Tylka TL. The trust model: a different feeding paradigm for managing childhood obesity. Obesity (Silver Spring) 2008;16:2197–204. 36. Brown K, Ogden J, Gibson L, Vogele C. The role of parental control practices in explaining children’s diet and BMI. Appetite 2008;50:252–9. 37. Niemeier HM, Raynor HA, Lloyd-Richardson EE, Rogers ML, Wing RR. Fast food consumption and breakfast skipping: predictors of weight gain from adolescence to adulthood in a nationally representative sample. J Adolesc Health 2006;39:842–9. 38. Wansink B, van Kleef E. Dinner rituals that correlate with child and adult BMI. Obesity (Silver Spring 2014;22:E91–5.
