Influence of Parents’ Eating Behaviors and Child Feeding Practices on Children's Weight Status
See Appendix for list of study centers.
The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Bonny Specker, Ethel Austin Martin Program in Human Nutrition, Box 2204, EAM Building, South Dakota State University, Brookings, SD 57007. E-mail: Bonny.Specker@sdstate.edu
Objective: To investigate the effects of mothers’ and fathers’ eating behaviors, child feeding practices, and BMI on percentage body fat and BMI in their children.
Research Methods and Procedures: Four hundred fifty-eight parents (239 mothers, 219 fathers) were asked to complete two questionnaires: the Three-Factor Eating Questionnaire and the Child Feeding Questionnaire, which measure dimensions of parent eating behavior and child feeding practices, respectively. Parent BMI was calculated from self-reported height and weight; children's measures included BMI and percentage fat assessed by DXA. Regression analyses were used to analyze relationships between parents’ BMI and questionnaire scores and children's weight status.
Results: One hundred forty-three mothers and 68 fathers returned questionnaires, representing parents of 148 children 3 to 5 years old (78 boys). Children's weight was related to mothers’ BMI, but not fathers’. Girls had a greater BMI if either parent reported being overweight as a child, and both girls and boys were likely to be overweight if their mothers believed they had risky eating habits (fussiness, eating too much, etc.). Girls with fathers who were more controlling had a higher percentage fat; these fathers were also more concerned about their daughters’ future health.
Discussion: Mothers exert a strong influence over their children's weight and seem to be more concerned about their children's eating behaviors; however, fathers play a role in imposing child feeding practices. Gender bias may be present in child feeding, as suggested by dissimilar effects of parent practices on the weight status of girls vs. boys. Fathers should be included in future studies analyzing parent feeding practices and children's weight outcome.
There are many variables within the family setting that can affect children's eating behavior and, ultimately, their weight outcome. Included among these are parents’ eating behaviors, foods made available to children, and child feeding strategies utilized (1, 2, 3, 4). Parents play a pivotal role in the development of their child's food preferences and energy intake (4), with research indicating that certain child feeding practices, such as exerting excessive control over what and how much children eat, may contribute to childhood overweight (1).
There are two primary aspects of control: restriction, which involves restricting children's access to junk foods and restricting the total amount of food, and pressure, which involves pressuring children to eat healthy foods and pressuring to eat more in general. Parents may use a combination of these methods to obtain a desired result; for example, pressuring a child to eat healthy foods by using bribes or rewards consisting of sugary snacks that are otherwise restricted (5). This tactic may lead to increased desire for and overconsumption of restricted foods when children are allowed free access to them (2) and may result in poorer eating regulation due to focusing children's attention on external cues (food portion size, rewards, and cleaning the plate) instead of allowing internal cues (hunger and satiety) to regulate their intake (1). Over time, this may result in increased risk for childhood overweight and/or obesity. A recent review by Faith et al. (6) suggests that the parent practice of restriction, in particular, is more likely to be associated with increased child eating and weight status vs. other aspects of parent control and feeding dimensions.
Much of the available research has focused on mothers’ eating behaviors and feeding practices and the effects on young girls. Few studies have included boys in their analysis, and, to our knowledge, no studies have examined how fathers’ feeding practices affect children's weight outcome (separate from mothers). In addition, few studies have related child feeding practices to children's percentage body fat as measured by DXA (7, 8). The aim of the current study was to relate daughters’ and sons’ BMI and percentage fat to mothers’ and fathers’ child feeding practices, eating behaviors, and BMI. Based on previous research (1, 3, 9), it was hypothesized that mothers’ dietary restraint and/or disinhibition, control in child feeding, and BMI would be positively related to daughters’ weight status but weakly associated, if at all, to sons’ weight status. Although the father's role in the family eating environment is currently unknown, we hypothesized a priori that we likely would not see significant relationships with children's weight outcomes.
Research Methods and Procedures
This cross-sectional study was approved by the South Dakota State University Human Subjects Review Board. Parents (239 mothers, 219 fathers) of children (n = 239) who were at one time participating in the South Dakota Children's Health Study (SDCHS)1 were asked to participate in the current study by completing two questionnaires. The SDCHS was a randomized trial that investigated the effects of physical activity and calcium supplementation on total bone mineral content and bone size parameters in 3- to 5-year-old children. Subject recruitment, research methods, and results of the SDCHS are described in detail elsewhere (10).
The first questionnaire, Stunkard and Messick's Three-Factor Eating Questionnaire (TFEQ) (11), is a 51-item instrument that measures three characteristics of parent eating behavior: cognitive restraint of eating (restricting calories to control body weight), disinhibited eating (difficulty in stopping or inability to resist emotional or social cues to eat even when not hungry), and perceived hunger (self-rating of feelings of hunger and how it affects eating behavior). This survey has been used in previous research examining parent eating behavior and how it relates to parent feeding practices, children's eating, and children's weight, and its psychometric properties have been well studied (12). The second questionnaire, the Child Feeding Questionnaire (CFQ), was adapted from Costanzo and Woody's Parent Interview (13) by Johnson and Birch (1). This instrument was designed to assess parents’ perceptions and concerns about child obesity, in addition to child feeding attitudes and practices. It consists of 24 forced-choice items evaluating parents’ use of control in feeding their children, concerns about their children's weight status and eating behaviors, and perceptions of their own weight when they were children. Examples of questions include “How concerned are you that your child will become overweight?” and “A child who doesn't finish all of their dinner should not get dessert” and “Parents have to be especially careful to make sure their children eat enough.” This survey has been used in previous studies examining parent feeding practices and children's eating behavior and weight status outcomes (1, 9, 14, 15, 16). All items were measured using a seven-point Likert-type scale, and responses ranged from disagree strongly to strongly agree or unconcerned to extremely concerned. Additional parent information used in the analyses were years of education and BMI calculated from self-reported height and weight as documented at their child's baseline SDCHS visit. The validity of self-reported heights and weights in adults has been established previously (17, 18, 19).
Children's Body Composition and Anthropometric Measures
Baseline measurements from the SDCHS were collected for the children whose parent or parents returned a questionnaire. These included height and weight for the calculation of BMI (kilograms per meter squared) and results from a whole-body DXA scan (Hologic QDR 4500A; Hologic Inc., Bedford, MA) for the calculation of total body percentage fat. Height was determined using a fixed measuring tape and recorded to the nearest 0.5 cm, and weight was recorded to the nearest 0.1 kg using a digital scale (SECA GMBH & Co., Hamburg, Germany). The validity of DXA for determining body composition in children has been previously established (20, 21).
Each parent eating factor produced the following range of scores: restraint (0 to 21), disinhibition (0 to 16), and hunger (0 to 14). Each child feeding factor produced the following range of scores: history of parents’ weight (2 to 14), control over child's eating (10 to 70), perceived child's eating risk (6 to 43), and concern for child's future health (4 to 28). For both questionnaires, a higher score indicated a greater presence of that factor. An adjusted score was calculated if not all questions pertaining to a factor were answered. The adjusted score was calculated as the average of the remaining scores multiplied by the total number of questions for that factor. At least 80% of the questions must have been answered for an adjusted score to be calculated and included in the data analysis.
All statistics were analyzed using the JMP 5.1 statistical software package (SAS Institute Inc., Cary, NC). Bivariate correlations were run between child weight status variables and potential parent confounding factors (parent BMI, years of education, and eating factor scores) to establish covariates to be included in multiple regression models evaluating associations between child weight status and parent questionnaire scores. Data are presented as mean ± standard deviation unless otherwise indicated, and significance is determined by p < 0.05 and a trend at p < 0.10.
One hundred forty-three mothers (60%) and 68 fathers (31%) returned at least one questionnaire, representing parents of 148 children (78 boys, 70 girls) who had a mean age of 4.0 years (3.0 to 5.3) and were primarily non-Hispanic white (96.6%).
Parents who returned questionnaires were generally well-educated; 99.3% of mothers completed the 12th grade, and 48% completed 16 years of school or more. Ninety-nine percent of fathers completed the 12th grade, and 42% completed 16 years or more. Seventeen percent of mothers and 24% of fathers were obese (BMI ≥ 30), and 30% of mothers and 46% of fathers were overweight (25 ≥ BMI < 30) (Table 1A). Overall, girls had a slightly lower BMI than boys with a higher percentage fat (Table 1B). In both girls and boys, BMI and percentage fat were closely related (r = 0.65 and 0.57, respectively, p < 0.001).
Table 1A. . Mean (range) of parent characteristics
|BMI||25.7 (17.3 to 47.9)||139||27.3 (17.7 to 41.9)||68|
| BMI < 18.5|| ||2 (1%)|| ||1 (1%)|
| 18.5 ≥ BMI < 25|| ||72 (52%)|| ||20 (29%)|
| 25 ≥ BMI < 30|| ||42 (30%)|| ||31 (46%)|
| BMI ≥ 30|| ||23 (17%)|| ||16 (24%)|
|Years of education||15.1 (11.0 to 24.0)||142||15.1 (12.0 to 22.0)||65|
Table 1B. . Mean (range) of child characteristics.
|Age||3.9 (3.0 to 5.1)||70||4.0 (3.0 to 5.3)||78||0.62|
|BMI||15.7 (12.1 to 18.8)||70||16.1 (13.7 to 20.6)||78||0.06|
| BMI < 25†|| ||59 (84%)|| ||68 (87%)|| |
| 25 ≥ BMI < 30|| ||11 (16%)|| ||8 (10%)|| |
| BMI ≥ 30|| ||0 (0%)|| ||2 (3%)|| |
|Fat (%)||28.1 (20.6 to 39.3)||70||23.3 (15.0 to 38.0)||78||<0.001|
A subsample of the parents who replied was used to analyze similarities in eating behaviors and feeding practices between parents of the same family. This was done to examine whether there might be potential for a shared parent influence on child's weight status, which may indicate that an aspect of the family environment, rather than an individual parent's practice, affects the child's weight outcome. In this study, parents scored similarly on level of control exerted over children's eating (r = 0.31, p < 0.05) and how they perceived their child's eating risk (r = 0.50, p < 0.001); parents also scored alike regarding their own restrictive eating behavior (r = 0.31, p < 0.05) (Table 2).
Table 2. . Pearson correlations (r) between CFQ and TFEQ scores within mother-father dyads
|CFQ|| || || |
| Weight history (n = 44)§,*||8.1 ± 1.5||7.5 ± 1.2||0.25|
| Control (n = 48)¶||38.1 ± 8.4||42.9 ± 7.1||0.31‡|
| Child's eating risk (n = 47)‖||22.9 ± 2.9||22.6 ± 2.6||0.50†|
| Future health concerns (n = 47)**||12.3 ± 5.2||12.4 ± 5.5||0.23|
|TFEQ|| || || |
| Restraint (n = 63)||9.4 ± 4.8||6.5 ± 3.7||0.31‡|
| Disinhibition (n = 63)||6.4 ± 3.8||5.2 ± 3.6||0.06|
| Hunger (n = 63)||4.8 ± 3.2||4.6 ± 3.0||0.20|
Mothers who reported having greater concern for their child's future health had a higher BMI (p < 0.001), and mothers with a higher BMI indicated being more overweight when they were children (p < 0.001) (Table 3). Overweight mothers reported significantly more disinhibited eating behavior and a greater amount of perceived hunger. Fathers’ BMI and disinhibited eating were also strongly associated, with a weaker relationship present between BMI and level of restraint. Fathers who were overweight as children had a higher BMI currently (p < 0.01).
Table 3. . Spearman rank correlations (r) between parent BMI and scores from the CFQ and TFEQ; median (range, n) scores are also given
|CFQ|| || || || |
| Weight history||0.48*||8 (4 to 14, 136)||0.35†||8 (5 to 10, 51)|
| Control||0.06||39 (13 to 55, 140)||−0.02||42 (27 to 60, 54)|
| Child's eating risk||−0.04||23 (14 to 31, 139)||−0.08||23 (15 to 27, 54)|
| Future health concerns||0.33*||12 (4 to 28, 138)||0.13||11 (4 to 24, 54)|
|TFEQ|| || || || |
| Restraint||−0.05||9 (0 to 20, 143)||0.22§||7 (0 to 17, 68)|
| Disinhibition||0.44*||5 (0 to 16, 143)||0.42*||4 (1 to 14, 68)|
| Hunger||0.18‡||4 (0 to 14, 143)||0.02||4 (1 to 12, 68)|
Mothers’ BMI was related to daughters’ BMI as well as sons’ BMI and percentage fat (r = 0.24, p < 0.05; r = 0.4, p < 0.001; r = 0.29, p < 0.01, respectively). No relationships were found between fathers’ BMI and daughters’ and sons’ weight status. An inverse relationship was discovered between daughters’ BMI and fathers’ years of education (r = −0.28, p < 0.05) and remained significant after controlling for mothers’ BMI (r = −0.34, p < 0.05), which was found to independently predict daughters’ BMI.
No relationships were found between parent TFEQ scores and child's BMI or percentage fat. There was a positive trend between daughters’ BMI and mothers’ disinhibited eating using bivariate analysis (r = 0.21, p < 0.10); however, this did not persist under multivariate testing controlling for established covariates (mothers’ BMI and fathers’ education). This suggested that confounding variables (e.g., mothers’ BMI) may have mediated this relationship or explained the same variance. Indeed, when mothers’ disinhibition was added to a model containing mothers’ BMI and fathers’ education to determine whether it also was a covariate, both disinhibition (p = 0.32) and mothers’ BMI (p = 0.24) fell out of the model, suggesting that these variables were explaining the same variance in daughters’ BMI.
Using multivariate analyses and controlling for daughters’ covariates (mothers’ BMI and fathers’ education) and sons’ covariates (mothers’ BMI), the following significant associations were found between child's weight status and parent feeding scores (Table 4): Daughters’ BMI was related to both mothers’ and fathers’ weight history as well as mothers’ perceived child's eating risk, and daughters’ percentage fat was associated with fathers’ level of control and future health concerns. Sons’ BMI and percentage fat were also related to mothers’ perceived child's eating risk as well as fathers’ weight history. Several relationships between parent feeding scores and child's weight status that were found using bivariate analysis became insignificant after controlling for established covariates (Table 4). This clearly indicates that mothers’ BMI, in particular, is a very strong, independent predictor of children's BMI and percentage fat and explains more of the variance in child's weight status outcomes than do the parent feeding factors in question.
Table 4. . Pearson correlation coefficients and multivariate model adjustments between daughters’ and sons’ BMI and percentage body fat and mothers’ and fathers’ scores from the CFQ
|Sample size|| || || || || || || || |
| Parent weight history||63|| ||25|| ||73|| ||26|| |
| Control||67|| ||26|| ||73|| ||28|| |
| Child's eating risk||66|| ||26|| ||73|| ||28|| |
| Future health concerns||67|| ||26|| ||71|| ||28|| |
|BMI|| || || || || || || || |
| Parent weight history||0.38†||0.03||0.50†||0.04||0.25‡||0.43||0.36§||0.06|
| Control||−0.09|| ||0.10|| ||−0.03|| ||−0.16|| |
| Child's eating risk||0.41*||<0.001||0.30|| ||0.47*||<0.001||0.27||0.02‖|
| Future health concerns||0.12|| ||0.02|| ||0.23§||0.27||0.36§||0.60|
|Body fat (%)|| || || || || || || || |
| Parent weight history||0.25‡||0.22||−0.22|| ||0.03|| ||0.43‡||0.04|
| Control||−0.06|| ||0.30||0.02‖||−0.001|| ||0.06|| |
| Child's eating risk||0.34†||0.004||0.12|| ||0.36†||0.008||0.07|| |
| Future health concerns||0.07|| ||0.35§||0.05||0.23§||0.48||0.03|| |
This study investigated the effects of mothers’ and fathers’ eating behaviors, child feeding practices, and BMI on measures of BMI and body fat in preschool sons and daughters. No significant relationships were found between parents’ eating behaviors and children's weight status after controlling for independent predictor variables; however, certain associations were noted between parents’ feeding practices and children's weight outcomes. For example, parents who reported being more overweight as a child tended to have heavier children, and mothers who thought their children had risky eating behaviors had heavier sons and daughters.
Parent Eating Behavior and Child Weight
Similar to our study, Whitaker et al. (8) found no significant relationships between parent eating behaviors as measured by the TFEQ and children's weight-for-height percentile, BMI, or percentage fat as measured by DXA. However, there is growing evidence that such relationships do exist. Cutting et al. (22) found that mothers who reported more disinhibited eating behavior had 3- to 6-year-old daughters who were more overweight, even after controlling for mothers’ BMI. Mothers in our study who were disinhibited also had daughters who were overweight, but the relationship appeared to be mediated by mothers’ BMI because it became insignificant when controlling for this factor. However, this may be largely explained by the fact that heavier mothers in our study also tended to report higher levels of disinhibited eating, thus masking the possibility of detecting a distinct relationship between mothers’ eating behavior and daughters’ weight. It certainly is possible that parents’ eating behaviors (including disinhibition) may mediate parent-child similarities in weight status by influencing the development of these eating behaviors in children.
Other research has shown, indirectly, that mothers’ restrictive eating behavior affects daughters’ weight, with the effect mediated by mothers’ restrictive feeding practices (2, 23), that is, restrictive eating predicts restriction in feeding, which in turn predicts daughters’ eating behavior and weight status. Studies demonstrating these findings have primarily been conducted within a laboratory setting; thus, although these relationships may have certainly been present in our subjects, they were not demonstrated due to the observational (non-intervention) nature of the study design. Very few studies have shown a direct relationship between parent TFEQ scores and measures of fatness in their children; thus, it may be that parents’ eating behaviors affect their children's weight through alternative mechanisms (i.e., feeding practices).
Child Feeding Practices and Child Adiposity
Previous research has suggested that the parent feeding practice of control (especially restriction) is related to overweight primarily in daughters (1, 2, 3). Again, these associations mostly have been found indirectly. We did not find parent control to be a significant indicator of children's overweight in our study. The only significant finding to this end was between fathers and daughters, with more controlling fathers having daughters with a higher percentage body fat, and these fathers also reported more concern for their children's future health. Contrary to our hypothesis, we did not find that mothers’ control over children's eating was associated with either daughters’ or sons’ weight measures. The small sample size of fathers makes it impossible to present a clear relationship between control and children's weight; however, our findings suggest that fathers may play a role in imposing feeding practices on their children and that these practices may have consequences. Robinson et al. (14) utilized the CFQ to interview mothers and fathers of 392 third-graders and, unlike our results, found an inverse relationship between control over children's eating and overweight in daughters (parent responses were combined; however, over 80% were mothers or female guardians). No such relationship was found in sons, and, in fact, the findings in daughters were only marginally significant when controlled for other factors.
In 2001, Birch et al. published a new version of the CFQ, which was not available for use in the year 2000, when the data for this analysis were collected. However, before 2001, several notable studies using the original CFQ were published and provided valuable data on the relationship between parents’ feeding practices for their children and children's eating behavior and weight status outcomes (1, 9, 14, 15, 16). Thus, we feel that although the most current version of the CFQ was not used in our study, the data nonetheless add to the growing body of literature on parent feeding strategies and implications for their children. The major difference between the original and current versions of the CFQ is the separation of the control factor into two separate constructs, restriction and pressure. These constructs measure distinct aspects of parent feeding, and this may help to explain why we did not see relationships between parent control and children's weight measures. The original CFQ may not have been sensitive enough to these constructs to detect significant relationships. Lending support to this theory, the recent review by Faith et al. (6) points out that only parent restriction, and not pressure to eat or other feeding dimensions, is related to child overweight and obesity.
Other limitations include the cross-sectional nature of this study, offering only a snapshot of the current situation, rendering it impossible to determine whether a parent's feeding practices affect the child's weight status, or the child's weight affects how the parent feeds. The sample size of fathers is relatively small for us to make concrete claims regarding relationships with children's weight. Also, a perhaps integral piece of this parent-feeding/child-weight puzzle is missing—a tangible measure of children's eating, including their behaviors, intake control, and food preferences, which are typically examined in a laboratory setting.
Few differences were found in the effects on daughters vs. sons, but one in particular was parents’ weight history. Mothers’ weight history was strongly related to daughters’ BMI, whereas it was not associated at all with sons’ weight measures. Whether this indicates a greater genetic influence between mothers and daughters or a copying of the mother's eating behaviors (i.e., more disinhibited eating) by the daughter is not known. Mothers in our study seemed more aware of their daughters’ and sons’ potentially harmful weight status because only mothers thought that risky eating behaviors were present if the child was overweight. Francis et al. (24) found that mothers were highly influenced by their own weight history as well as concern for their daughters’ weight status when it came to developing the feeding practices imposed on their daughters, thus possibly contributing learned behavior. Even though their study did not include sons, these learned behaviors may be applicable to both. Fathers in our study who were overweight as children tended to have both daughters and sons currently overweight; however, fathers who were overweight currently were no more likely to have overweight children than non-overweight fathers.
In summary, it seems evident that the way parents eat affects the environment in which they feed their children, and this may be gender-biased. Furthermore, based on previous research, parents’ feeding practices clearly seem to affect their children's eating behavior. At ages 3 to 5 years, this behavior might only be evident in a controlled setting (i.e., laboratory) and may not be reflected in their weight status due to limited autonomy over eating at this young age. However, at older ages (i.e., school age), when children have greater independence in food selection and intake, these earlier learned behaviors may begin to be expressed and subsequently affect the child's weight status. For example, Fisher and Birch (25) studied 192 white girls and found that parents’ reports of restricting daughters’ access to foods at age 5 predicted girls’ eating in the absence of hunger at age 7 even after controlling for other variables, and those who ate large amounts of food at 5 and 7 years were 4.6 times as likely to be overweight at both ages.
Future research should elaborate on the role that fathers play in developing feeding practices imposed on children and should examine effects on daughters and sons. Elucidations are needed on how mothers’ and fathers’ eating behaviors affect their feeding practices. Studies examining children's weight status ideally should include a measurement of the child's food intake, preferences, and eating behaviors, whether laboratory-based or through questionnaires. Finally, longitudinal, vs. cross-sectional, study designs are preferable in that the researcher can delineate whether the feeding practices are formed in response to the child's weight, or the child's weight develops in response to the parent's feeding practices. It will also allow examination of changes in the child's weight and eating behaviors over time, as the consequences of parent influences imposed early on begin to be expressed.
This work was supported, in part, by NIH Grant R01-AR45310 and by the Ethel Austin Program in Human Nutrition at South Dakota State University. We acknowledge the families who participated in this project and thank them for their time and commitment.
Nonstandard abbreviations: SDCHS, South Dakota Children's Health Study; TFEQ, Three-Factor Eating Questionnaire; CFQ, Child Feeding Questionnaire.