Parental attachment insecurity predicts child and adult high-caloric food consumption

Abstract

Eating habits are established early and are difficult to change once formed. This study investigated the role of caregiver–child attachment quality and its associations with high-caloric food consumption in a sample of middle socio-economic status children and adults, respectively. Survey data were collected from an online questionnaire administered separately to 213 (143 girls) children and 216 parents (adult sample; 180 women). Two studies showed that an insecure parental attachment, whether actual (Study 1; children) or recalled (Study 2; adults), significantly and positively predicted high-caloric food consumption in both samples. The present findings highlight the importance of parental attachment and its association with unhealthy eating patterns in children and adults.

Keywords

adults attachment children eating food consumption insecurity obesity

Introduction

Dietary habits formed early in life constitute the foundation for a life course of healthy eating habits (Birch, 1999). Unfortunately, healthy eating habits are more and more difficult to establish in the actual context of plenty, where calorie-rich/high-fat foods are widely available and inexpensive (Drewnowski and Almiron-Roig, 2010; Swinburn et al., 2011). Food overconsumption, particularly the consumption of high-caloric foods (HCFs), has been singled out as a core driver of obesity (Swinburn et al., 2011). Overweight and obesity bear important negative physiological and psychological consequences for children, and later on, adults, including cardiovascular diseases, certain types of cancers, depression, and anxiety (Hebebrand and Herpertz-Dahlmann, 2009; Moriarty-Kelsey and Daniels, 2010).

Parents are critical in the formation and modification of eating habits (Tabacchi et al., 2007). Because early dietary habits established in childhood are difficult to change once formed, it is important to understand the role parents play in children’s eating habits and how these influences are internalized and enacted in food choices in both childhood and adulthood.

Parental/caregiver patterns of interaction with and responsiveness to children constitute the basis of future emotion regulation (Mikulincer and Shaver, 2007). Individuals (children and adults) with secure attachments bonds learn that others can be counted on to provide comfort and support, which makes them much more successful in regulating their emotions, whereas the reverse is true for insecurely attached individuals. The role of emotion regulation in unhealthy and disordered eating has been well established in both children and adults (Czaja et al., 2009; Evers et al., 2010; Harrison et al., 2010; Volkow et al., 2008). Consequently, it is important to understand how parental-attachment bonds might foster differential dietary habits in everyday life. Unfortunately, to date, fewer than a handful of studies have examined attachment and everyday food consumption, and none of these included children, a particularly vulnerable population.

Attachment conceptualization and link with emotional regulation

The attachment theory aims to explain how caregiver–child interaction quality becomes gradually internalized to guide a variety of arousal and distress responses throughout the life course (Bowlby, [1969] 1999; Sroufe, 1979). Attachment is the foundation of emotional regulation (Mikulincer and Shaver, 2007; Alexander and Siegel, 2013) or one’s ability to monitor, evaluate, and modify the magnitude/experience of emotion-eliciting events (Aldao et al., 2010; Nolen-Hoeksema, 2012).

The attachment system is universal and, when activated by distress, motivates individuals to seek and maintain proximity with protective others (Sroufe, 1979). Attachment develops from early child–caregiver interactions (generally parents) in which sensitive and available caregivers who cater to their children’s physical and emotional needs provide them with a secure base upon which they learn to organize their distressing experiences and successfully deal with them. As explained above, these attachments are internalized and guide, among others, how a person regulates emotions. For instance, secure parent–child bonds teach the child that proximity or support-seeking appeases negative felt emotions because others can be trusted to provide comfort when in need (Bretherton, 1992; Goossens et al., 2012; Kobak and Sceery, 1988). This confidence is internalized over time and helps secure individuals acknowledge and display emotions as an important step to restoring emotional balance (Cassidy, 1994; Mikulincer and Shaver, 2012). Secure individuals are thus said to rely on primary attachment strategies to regulate distress.

When, on the other hand, the certitude of another’s availability is violated repeatedly, individuals must resort to secondary attachment strategies to help them regulate negative emotions. These strategies are reflected within insecure attachments better conceptualized as anxious and avoidant dimensions (axes) on a bi-dimensional Cartesian plane (Mikulincer and Shaver, 2012). Individuals scoring high on the attachment-anxiety dimension resort to hyperactivating emotional strategies to handle distress. These strategies are characterized by constant vigilance for potential threats and distress, ongoing concerns about the availability of attachment figures, and efforts to re-establish proximity (Gillath et al., 2009; Mikulincer and Shaver, 2007). On the other hand, individuals scoring high on the attachment-avoidant dimension use deactivating emotional strategies to regulate distress. Deactivating strategies are characterized by attempts to regulate distress alone and by minimal reliance on attachment figures (Gillath et al., 2009; Mikulincer and Shaver, 2007). The use of insecure or secondary attachment strategies has been both theoretically and empirically associated with maladaptive emotional regulation, while the reverse has been found for attachment security (for a complete review, see Mikulincer and Shaver, 2007).

Researchers have theorized that emotion regulation failure is a core mechanism through which attachment and maladaptive eating behaviors are related (Goossens et al., 2012). For instance, anxiously attached adults and children worry constantly about being rejected by caregiving figures, whereas avoidant individuals dismiss stress-related thoughts (Gillath et al., 2009; Kobak and Sceery, 1988). Ultimately, both strategies will backfire as insecurely attached individuals’ distress is not optimally addressed, making these individuals more likely to resort to eating to soothe negative affect (Anderson et al., 2012).

Attachment and eating literature review

In children, two studies have investigated the link between attachment and loss of control (LC) over eating, a key aspect of disordered eating. LC over eating is an inherent feature of binge eating, or eating large amounts of foods in a short time, and is highly predictive of eating pathology and general distress (see Goossens et al., 2009). Goossens et al. (2011, 2012) found that children aged 8–11 years who reported LC over eating as opposed to those who did not were less likely to be securely attached to both mothers and fathers.

In adults, women with more insecure parental attachments, whether recalled from childhood or reported in adulthood, were more likely to report bulimic (LC over eating with purging behaviors) and binge (LC over eating with feelings of guilt and shame; no purging behaviors) eating behaviors (Davis, 1998; Han, 2009). These relationships were also uncovered when considering attachment toward mothers and fathers separately, with higher insecure attachment styles predicting higher disordered eating in both women (Tucker and McNamara, 1995; Lehoux and Howe, 2007; Han, 2009) and men (Han, 2009).

Although evidence exists for a relationship between attachment and clinical aspects of eating, these behaviors are not representative of general eating patterns. To date, only three studies have examined the link between attachment and food consumption/choice. In an experimental study, Patton (1992) showed that bulimic women whose maternal attachment was threatened using an insecure attachment prime (“mama is leaving me”) ate more crackers than women whose maternal attachment was not threatened (“mama is loaning it”). Similarly, Wilkinson et al. (2013) found that when women were primed with an anxious versus a secure romantic-attachment prime, they ate more chocolate chip cookies in terms of calories. Conversely, Prichard et al. (2012) found that maternal attachment security did not predict daughters’ HCF or vegetable consumption in a sample of young adult women. Rather, only mothers’ consumption of these foods and daughters’ restrained eating behavior predicted daughters’ HCF and vegetable consumption. A main limitation of this study was that daughters self-reported food consumption for themselves and their mothers, thus potentially biasing maternal food-consumption reports. However, this finding is in accordance with research showing that parent–child total energy and fat intakes were strongly correlated (Beydoun and Wang, 2009).

In sum, the present available evidence linking attachment and eating is heavily biased toward clinical aspects of eating and female samples. None of the above studies examined unhealthy eating in children. Moreover, although laboratory studies found a relationship between attachment and episodic food consumption, these results have not been replicated outside this context. Because of research findings showing that parental HCF consumption predicts child HCF consumption, we will also aim to partial out this effect as a potential confounder of the attachment–child HCF consumption relationship.

This article thus aims to complement the available research on attachment and everyday food consumption by examining the link between parental attachment and HCF consumption in children and adults. We define HCF consumption as consumption of any foods that have been shown to be associated with long-term weight gain and obesity. We propose that parental (primary caregiver) attachments forged in childhood carry a pervasive weight in everyday eating choices, particularly in HCF consumption, a primary driver of obesity rates. More importantly, we propose that due to its pervasiveness, this relationship can be discerned when examining actual child–caregiver attachment (Study 1) as well as when investigating recalled parental attachment in adults (Study 2). Understanding how attachment is linked with HCF consumption in children is of great value in promoting healthier eating habits and averting potential long-term weight gain in children.

Study 1: overview

To investigate the role of attachment in unhealthy food consumption, we measured HCF consumption and attachment in a sample of 213 children aged 8–12 years. We hypothesize that children reporting higher attachment insecurity will eat more HCFs, even after controlling for potentially confounding variables. As parent–child total energy intake were found to be highly correlated (Beydoun and Wang, 2009) and predictive of HCF consumption beyond attachment security, this study further aims to disentangle these confounding effects. In this study, both child and parent HCF consumption were available through separate self-report.

Study 1: method

Procedure

To reach children aged 8–12 years, a research assistant scheduled visits with 34 French schools whose administrators agreed to have our research introduced to children in the third, fourth, fifth, and sixth grades. The research assistant distributed a complete information package inviting children and one of their parents to partake in a larger online study beyond the scope of this article. All schools were pre-selected to be of middle socio-economic status (SES), as a mandatory requirement for the larger online study, to provide a homogeneous sample of children with similar access to healthy and unhealthy foods.

Signed and sealed informed consent/assent forms were returned to school by each child and picked up 2 weeks following distribution by the research assistant. The research assistant then proceeded to activate and distribute individual passwords granting parents and children separate access to an online questionnaire. The questionnaire comprised a variety of measures and took approximately 1 hour to complete. Participants (both children and parents) had 1 week to complete the questionnaire and could return to it as many times as they wished within the allotted time. This flexibility was offered to respect the cognitive developmental capacity of children and potential time pressure faced by parents. All procedures were approved by and, hence, conducted in accordance with the ethical standards of the university’s and the Montreal School Board’s ethics committees. Participants received no compensation for answering this questionnaire.

Participants

Complete data were obtained for 213 children aged 8–12 years. Four children were in Grade 3, 87 were in Grade 4, 81 were in Grade 5, and 41 were in Grade 6; 90 percent of the children spoke French. Matching child–parent data were obtained for only 158 parents (55 parents less than the final total number of children; see Table 1 and Appendix 1 for complete sample description). Upon comparing the 55 children whose parents did not fill out the HCF questionnaire with the 158 children whose parents did, we found that children whose parents did not fill out the HCF questionnaire reported higher parental-attachment insecurity, consumed more HCFs, had slightly higher body mass indices (BMIs), and were older (see Table 1). As stated above, a main goal of this study was to partial out the effects of parental HCF consumption from the relationship between child parental attachment and HCF consumption. Simultaneously, it was impossible to ignore the above group differences. To solve this issue, we thus performed and reported analyses in parallel for both the complete child sample (N = 213) and the child–parent dyad data (N = 158). Differences in results between these two samples are noted in Study 1: Results section.

Table 1.

Means, standard deviations, and correlations of descriptive, independent, and dependent variables for complete child data (N = 213) and complete child–parent dyad data (N = 158; Study 1). Correlations in the upper diagonal show the complete child sample, while the ones in the lower diagonal concern the complete parent–child dyad data.

			Sample difference	InsecD	Age	C-HCF	zBMI	P-HCF
N	213	158
Gender	143 girls	106 girls	χ² = .00
InsecD	1.77 (0.80)	1.66 (0.68)	3.05***	1	.07	.42****	−.08	–
Age	10.17 (0.92)	10.09 (0.93)	2.14*	.00	1	.15*	.10	–
C-HCF	6.90 (5.17)	6.07 (3.69)	3.05**	.32****	.03	1	.10	–
zBMI	0.69 (1.55)	0.08 (0.38)		−.04	.08	.11	1	–
P-HCF	–	4.69 (2.74)	–	.04	−.17*	.16*	.06	1

zBMI: body mass index z-scores transformed using World Health Organization (WHO) criteria; InsecD: insecure attachment dimension; C-HCF: high-caloric food consumption of children; P-HCF: high-caloric food consumption of parents; sample diff.: t-test difference between the complete child sample and the parent–child dyad sample.

T-test significant differences between child–parent dyad data (N = 158) versus child–no-parent sample (N = 55) are indicated for each variable.

p < .05, **p < .01, ***p < .001, and ****p < .0001.

Materials

Attachment

Child attachment toward parents was measured using a simplified French version of the Adult Attachment Questionnaire (AAQ; Simpson et al., 1992) validated in French by Bouthillier et al. (1996) and adapted to child–parent relationships by replacing the word others/people with parents (see Muris et al., 2001; Roelofs et al., 2006: for similar methodology). This questionnaire decomposes Hazan and Shaver’s (1987) three attachment vignettes, based on Ainsworth et al’s (1978) descriptions of secure, avoidant, and anxious attachment in children, into 13 individual propositions mapped on two dimensions, that is, anxiety and avoidance. Although we did not discriminate between maternal and paternal attachment styles within this questionnaire, Goossens et al. (2011, 2012) found that both attachment toward mothers and attachment toward fathers predicted LC over eating in children.

Before administering the simplified French AAQ, we conducted a qualitative validation of it (length, understanding, and sentence complexity) by inviting 10 children aged 8–12 years to our laboratory to provide feedback on the questionnaire. Following this, two items deemed too difficult were dropped; the final administered questionnaire included 11 statements. Children were thus asked to rate the quality of their relationship with their parents using a 1 (strongly disagree) to 5 (strongly agree) Likert scale. An example item was “I find it relatively easy to get close to my parents.” Attachment questionnaires have been conclusively and reliably adapted to children aged 6–14 years (Roelofs et al., 2006; Finzi-Dottan et al., 2006; Muris et al., 2001; Sharpe et al., 1998).

To validate the AAQ’s two-factor structure, we performed a confirmatory factor analysis (CFA) on the complete child data (χ²(26) = 58.94, p = .0002; comparative fit index (CFI) = .96; root mean square error of approximation (RMSEA) = .08; standardized root mean square residual (SRMR) = .05). CFA results led to two other statements being dropped, reducing the items from 11 to 9 (see Bouthillier et al., 1996: for similar procedure). Whereas the avoidant dimension showed good reliability (five items; α_{all children} = .84), the anxious dimension’s internal consistency was at the lower limit of acceptability (four items; α_{all children} = .66). Both attachment dimensions were also highly interrelated (r_{all children} = .62, p < .0001), indicating that they might be tapping into a similar construct. As such, to reach informed conclusions and to minimize bias from a weaker attachment-anxiety index as well as collinearity, we combined the anxious and avoidance dimensions into an insecure attachment dimension (α = .84; Koskina and Giovazolias, 2010: for similar index). Higher scores reflected higher attachment insecurity or inability to depend on and greater worries about the availability of parental figures. To check for self-report biases, we ran an analysis of variance (ANOVA) with age as a proxy for cognitive development. Results showed that reported insecure attachment did not differ by child cognitive development, p > .05.

HCF consumption

A short food-consumption questionnaire was elaborated to reflect general eating habits. Participants were presented with a 32-item list of commonly consumed foods, such as yogurt, carrots, fries, and cookies. They were instructed to think about an ordinary day, visualizing breakfast, lunch, supper, and snacks, and to indicate how many times per day, week, or month they ate each of the 32 food items. Food-item examples were provided particularly to aid children in their estimations, for example, chocolate bar (Mars, Reese’s, Coffee Crisp, Oh Henry!, etc.). This questionnaire emphasized frequency of consumption as opposed to the quantity of food items, as the former is a greater contributor to food variance (Thompson and Subar, 2008). School-age children have been shown to reliably report their food consumption by the fourth grade (National Cancer Institute, 2010).

Daily HCF consumption was determined by summing the number of times children and parents ate any of the 21 potentially obesogenic foods included in the food-consumption questionnaire. These foods were also shown to contribute to energy, fat, and carbohydrate intake by Canadian and Quebecer statistics (Caminard, 2011; Garriguet, 2007; Johnson-Down et al., 2006) and included (1) bacon and sausages, (2) sugared cereals, (3) unsweetened cereals, (4) potatoes, (5) smoked meat or cold cuts, (6) fried chicken, (7) hot dogs, (8) hamburgers, (9) fries or poutine, (10) cheese, (11) pasta, (12) red meat, (13) crackers, (14) candy, (15) cookies, (16) chocolate bars, (17) ice cream, (18) nuts, (19) crisps, (20) waffles, and (21) granola bars (α _{all children} = .75; α _parents = .67). Short dietary-assessment instruments have been shown to be reliable when used for comparative purposes (Thompson and Subar, 2008), as is the case in this study. An additional reliability check showed that reported HCF consumption did not differ by child age, p > .05.

Control variables

HCF consumption might be affected by child BMI, age, and gender (Tabacchi et al., 2007). Children self-reported their height and weight and, when in doubt, were instructed to ask for an adult’s help. To account for reporting biases, in all our analyses, we used age- and gender-adapted BMI percentiles calculated using the World Health Organization (WHO) Anthropo BMI tool (Back, 2011). Based on their BMI percentiles, children were also categorized as very thin, thin, normal, overweight, or obese (BMI category). Additional statistical checks showed that age, used as a proxy for cognitive development, did not predict child BMI percentile or BMI category, p > .05. The above variables were investigated for their confounding effects and included in the final model if significant.

Statistical analyses

First, as children were nested within 34 schools, mixed regression models with random intercepts and slopes were run to allow for student variability within each school as well as school variability (Singer, 1998). Child’s school did not predict HCF consumption, nor did school SES indices (variables provided by the government of Québec), ps > .05. Following the principle of parsimony, we proceeded to re-run our analyses using simple regression models (general linear models (GLMs)). All analyses were done using SAS version 9.3. Hierarchical analyses results are available upon request.

Study 1: results

Correlations

Table 1 shows that in both child samples HCF consumption was significantly correlated with attachment, in that higher attachment insecurity was linked with higher HCF consumption in children. For the child-dyad sample, parent–child HCF consumption was significantly correlated, in that the more HCFs parents consumed, the more HCFs children ate, p < .05. Age was only associated with HCF consumption in the complete child data. No associations were found between standardized BMI scores and child HCF consumption in either sample. However, additional ANOVA tests showed that HCF consumption did vary by BMI category for the complete child sample but not for the child-dyad sample. In other words, for the complete child sample, compared with children who were classified as obese, overweight and normal-weight children reported eating less HCFs (M_obese = 9.32 (7.78) vs M_overweight = 6.15 (4.48) and M_{normal weight} = 6.45 (4.25)). HCF consumption did not vary by gender in either sample, p > .05. We thus proceeded to formal model testing by including only significant relationships.

Predictors of child HCF consumption

GLM models included parental insecurity as predictor and HCF consumption as the outcome variable. Age and BMI category were entered as control variables in the complete child sample, and parental HCF consumption was entered as a control variable in the child–parent dyad model (Table 2). As expected, attachment insecurity was a strong predictor of child HCF consumption (ps < .0001), with higher parental-attachment insecurity being linked with more HCF consumption (Model 1), even after introducing age and BMI category as covariates (Models 2 and 3). Attachment insecurity remained a strong predictor of child HCF consumption, even after partialing out parental HCF consumption (Model 4).

Table 2.

Beta estimates (β), standards errors (SEs), and confidence intervals (CIs) of HCF food-consumption predictors in children (Study 1) and adults (Study 2).

Study 1 (children)
	All children				Child–parent dyads
	F	β (SE)	CI		F	β (SE)	CI
			Low	High			Low	High
Model 1	R² = .18				R² = .11
InsecD	44.89****	2.70 (0.40)	1.91	3.50	18.36****	1.75 (0.41)	0.94	2.56
Model 2	R² = .19				R² = .11
InsecD	43.37****	2.52 (0.40)	1.85	3.44	17.73****	1.74 (0.41)	0.93	2.56
Age	3.70	0.70 (0.35)	−.02	1.36	0.21	0.14 (0.30)	−0.46	0.74
Model 3	R² = .23				R² = .13
InsecD	41.21****	2.61 (0.41)	1.81	3.41	17.93****	1.77 (0.41)	0.95	2.60
Age	3.40	0.66 (0.36)	−0.05	1.36	0.12	0.11 (0.31)	−0.51	0.73
zBMI cat	2.42*				0.86
	Very thin	−3.88 (2.46)	−8.74	0.98	Very thin	−2.32 (1.94)	−6.14	1.51
	Thin	−2.75 (2.48)	−7.63	2.13	Thin	−2.75 (2.62)	−7.91	2.42
	Normal	−2.47 (0.88)	−4.21	−.73	Normal	−1.27 (0.82)	−2.89	0.35
	Overweight	−2.68 (1.01)	−4.67	−.69	Overweight	−1.18 (0.92)	−3.00	0.64
Model 4	–				R² = .13
InsecD	–				18.61****	1.75 (0.41)	0.95	2.55
HCF-P	–				4.60*	0.22 (0.10)	0.02	0.42
Study 2 (adults)
	F	β (SE)	CI		F	β (SE)	CI
			Low	High			Low	High
Model 1	R² = .06
InsecD	13.30***	1.50 (0.41)	0.69	2.32
Model 2	R² = .09
InsecD	12.01**	1.45 (0.42)	0.63	2.28
Education^$	3.04⁺
	HS	2.55 (1.03)	1.06	4.82
	Coll	0.37 (0.75)	−1.11	1.86

zBMI: body mass index z-scores transformed using World Health Organization (WHO) criteria; HS: high school degree; Coll: college degree.

Reference category for education is completed or incomplete university diploma.

p = .05, *p < .05, **p < .01, ***p < .001, and ****p < .0001.

Parental HCF consumption also predicted child HCF consumption. Moreover, compared with children who were categorized as obese, those classified as overweight and normal weight ate fewer HCFs. The lack of significant differences between children classified as obese with those classified as thin and very thin might be explained by their low numbers, total N = 8. Age did not significantly predict child HCF consumption, p > .05.

Discussion

Study 1 examined the effect of parental-attachment quality on child HCF consumption while controlling for important covariates. In line with hypotheses, children with higher insecure attachment—higher worries about parental availability and/or discomfort in being close to parents—consumed higher daily amounts of HCFs compared with less insecurely attached children. These findings are in line with Goossens et al. (2011, 2012), who showed that children with less-secure parental attachments reported more uncontrolled eating episodes, and with laboratory findings in which attachment anxiety and insecurity were associated with higher consumption of a snack item (Patton, 1992; Wilkinson et al., 2013). To date no study has examined children’s general eating habits and how attachment might be linked with unhealthy food consumption. We thus extend available findings to parental-attachment insecurity in children and further generalize the context of these findings by using survey data.

Our findings are not in line with those of Prichard et al. (2012), who found that attachment did not predict daughter unhealthy food consumption in a sample of college women. This inconsistency between findings could be explained by specific attachment patterns playing differential roles in HCF consumption, according to which it is not the presence of secure parental bonds but rather the absence of such relationships that becomes a better predictor of HCF consumption. As expected from the above authors, parental HCF consumption did predict child HCF consumption.

This study is not free of limitations. First, although previous studies have adapted AAQs to children (Roelofs et al., 2006), the AAQ had not been previously used in a child population. The most important limitation of this study was the ability of children to correctly self-report their HCF consumption or parental attachments, among other variables. Moreover, we were not able to find any school effects. This might be due to restrictions of the larger study, as the present child sample pool was recruited from middle-SES French schools, making the sample more homogeneous with respect to ethnicity, SES, and BMI, and even school. As such, an essential second step to reinforce the present results would be to generalize the above findings to another population, such as adults, who are not nested within schools, and who can self-report their behaviors more reliably.

Study 2: overview

This study aims to replicate the above findings in a sample of adults and remediate important limitations of Study 1, notably children’s ability to self-report behaviors and undiscerned school effects. Within this context, we aim to investigate the attachment–eating relationship within a sample of adults to further generalize our findings. In Study 2, data from 216 adults are used to investigate whether parental attachment, as recalled by adults from their youth, predicts HCF consumption. As in Study 1, we hypothesize that higher recalled insecure attachment to parents will predict higher HCF consumption in adults, even after controlling for potentially confounding variables. Investigating how parental attachment is linked with HCF consumption in adults will help assess the importance of first forged bonds and the role they play in present food consumption.

Study 2: method

Procedure

Adults in this study were parents recruited through their child’s school (see Study 1: Procedure for complete description). This sample was treated independently of their child’s participation. Adults provided informed consent to participate in the study, in accordance with the university’s and the Montreal School Board’s ethics committees, and received no compensation for their participation.

Participants

The final sample included 216 adults (180 women), of whom 143 had some or complete university studies. Participants were 90 percent French-speaking, were 41.31 years of age on average (standard deviation (SD) = 7.71), and had a 25.49 (SD = 6.97) BMI (see Appendix 1 for complete sample description).

Materials

Attachment

Recalled parental attachment in adults was measured using the same simplified French version of the AAQ, validated for children (see Study 1: Materials). Similar to our adaptation of the AAQ for children, to assess early caregiver–child relationships in adults, we replaced the words others/people with the word parents. Adults were told that we were interested in learning a bit more about their relationship with their own parents when they were young. Using a Likert scale from 1 (strongly disagree) to 5 (strongly agree), participants then checked the box corresponding to their feelings.

The final attachment scale included nine items (avoidant dimension: five items, α = .82; anxious dimension: four items, α = .64): two were omitted after the scale was qualitatively validated with children and two were removed after running a CFA with two factors/attachment dimensions using SAS PROC FACTOR (CFA; χ² (26) = 101.41, p < .0001; CFI = 0.89; RMSEA = 0.11; SRMR = 0.06; Akaike information criterion (AIC) = 49.41). Note that the two-factor structure of the model showed only an acceptable fit, which could be attributed to small degrees of freedom and a low number of participants. Similar to Study 1, the reliability of the anxious index was rather low, and the correlation between dimensions was high (r = .60, p < .0001). To remedy these weaknesses, we composed an insecurity scale by averaging all nine attachment items (α = .83; mean = 1.93 (.74)), with higher scores signifying higher attachment insecurity toward parents in youth.

HCF consumption

The same 32-item food-consumption questionnaire aiming to capture general eating habits was administered to adults. All items were transformed on a per-day basis, and the same 21 items were used to compute daily HCF consumption scores with good reliability (α = .87; mean = 5.16 (4.81)). These items also captured the outlined Canadian and Quebecer top sources of energy, fat, and carbohydrates in adults (Johnson-Down et al., 2006).

Control variables

Initial control variables shown to influence HCF consumption included gender, BMI (kg/m²), and level of education (Prättälä et al., 2007; McLaren, 2007), assessed by asking participants to indicate their highest level of education between (1) complete/incomplete high school studies (low education level; N = 23), (2) complete/incomplete college studies (middle education level; N = 50), and (3) complete/incomplete university studies (high-education level; N = 143). Only variables significantly associated with HCF consumption were included in our subsequent analyses.

Statistical analyses

In this case, data were analyzed directly using GLM regression models in SAS version 9.3. As in our study of children, HCF consumption was the dependent variable and attachment insecurity the independent variable. Control variables were then introduced to test for potential confounds.

Study 2: results

Correlations

HCF consumption was significantly and positively related to attachment insecurity, r = .25 and p < .001. BMI was not associated with HCF consumption, p > .05. Further ANOVA tests showed that, compared to individuals with a high-education level, those with low education ate more HCF foods, p < .05. HCF consumption did not significantly differ by gender or BMI category, ps < .05.

Predictors of adult HCF consumption

In line with our hypotheses, results showed that attachment insecurity significantly predicted HCF consumption in adults (Table 2; Model 1), in that higher reported parental-attachment insecurity in youth was associated with higher present HCF consumption. These results remained significant even after introducing education into the model as a covariate (Model 2). Level of education was a significant covariate in this model, in that individuals with low education reported higher HCF consumption than their high-education counterparts, p = .05.

Discussion

Study 2 examined how youth parent–child attachment influenced present HCF consumption in adults. In line with our hypotheses, we found that adults who rated their youth parental-attachment patterns as more insecure (higher worry patterns/inability to depend/discomfort with closeness) consumed more HCFs than adults who evaluated their youth parental attachment as being less insecure. These findings provide preliminary evidence that daily eating practices are in part influenced by attachment patterns conveyed from childhood.

Our results are consistent with findings showing that higher insecure parental attachments were associated with a higher likelihood of reporting clinical disordered eating (Han, 2009; Lehoux and Howe, 2007; Tucker and McNamara, 1995). These findings also complement Patton’s (1992) experimental results to show that parental-attachment influences translate to a larger context, that of everyday unhealthy eating behaviors. Results also showed that having less education was positively associated with HCF food consumption, which is in line with lower SES being an important obesity-risk factor (McLaren, 2007).

Although Study 2 addressed important concerns of Study 1—the reliability of child self-report issue and the lack of school effects—a main limitation of Study 2 was that parental attachment was measured as it was recalled from youth and not as it was now. Recall can be biased in multiple ways that could include downplaying of negative memories for avoidant individuals (Kobak and Sceery, 1988). Nevertheless, it is not unusual to measure recalled attachment, as is the case with the AAQ, Adult Attachment Interview (Main et al., 2008), or the Attachment History Questionnaire (Pottharst, 1990). Other limitations of Study 2 included the adult sample being highly educated and lacking diversity, as well as the HCF questionnaire not being formally validated, despite its good reliability.

General discussion

We provide initial evidence of the importance of early child–caregiver socialization and of how insecure attachment patterns may shape a lifetime path of vulnerabilities whereby children and adults who cannot fully depend on or trust caregivers show a higher preference for unhealthy foods in general. This result was expected, as attachment insecurity has been linked with lower ego resiliency, ego under-control, and anxiety in children and late adolescents (Arend et al., 1979; Kobak and Sceery, 1988), and with greater use of ineffective emotion regulation strategies (rumination, self-blame, and wishful thinking) in adults (Mikulincer and Shaver, 2007). Future studies should seek to explicitly investigate how emotional regulation may mediate the relationship between attachment orientation and eating in children and adults alike. Furthermore, studies should adopt a longitudinal view and establish an explicit link with BMI, as important population studies with children showed lack of attachment security, whether measured in infancy or in childhood, to be linked with important BMI increases in subsequent years (Anderson et al., 2012; Anderson and Whitaker, 2011; Goossens et al., 2012). No links with present BMI were found.

Note that both adult and child samples used in this article were related, as 158 children could be matched with their respective parental figures. Nevertheless, although correlation analyses showed that parental attachment to their own parents in their youth was not related to their children’s actual attachment to them, p > .05, we did observe the same pattern of results in both samples. Furthermore, the explicative power of parental attachment in children was much stronger than that of youth attachment in adults (18% vs 6%). This pattern of results is expected, as parental attachment plays a much more direct and essential role in children, whereas other factors—even other attachment figures—may do so in adults. As mentioned previously, both Study 1 and Study 2 samples lacked diversity and were French-speaking. However, attachment theory is universal (Bowlby, [1969] 1999), and, as such, being able to observe significant effects in such homogeneous samples is rather encouraging for finding potential attachment–eating effects in populations with more BMI and SES variations, as well as other nationalities.

This article found preliminary evidence showing that insecure parental attachment forged in infancy is associated with higher obesogenic food consumption in children and adults, even after controlling for confounding variables. These results emphasize the importance of achieving a healthy caregiver–child relationship shaped by parental trust, availability, and sensitivity, as participants who reported or recalled lacking such relationships were more likely to report unhealthy eating patterns. From an intervention standpoint, parents-to-be could receive a short information sheet teaching them that responding to a child’s attention-seeking does not make the child needy but rather teaches him or her to become trusting and independent. From a counseling and clinical standpoint, children and parents of children with eating issues could benefit greatly from developing secure and trusting relationships.

These findings also build on and complement the limited available literature for children and adults on attachment and general eating patterns, showing that parental attachment might play a role in dietary habits during youth and beyond. Our evidence suggests that enhancing and improving child–caregiver relationships through the fostering of parental availability, trust, and warmth could lead to improved dietary habits in children, a gain that might carry into adulthood.

Footnotes

Funding

This work was funded by grants from the Danone Institute [Grant # 206960] and the Social Sciences and Humanities Research Council [SSHRC; Grant # 206760] to Dr. Laurette Dubé as well as by Psychosocial Oncology Research Training (PORT) and Fonds de la recherché en santé du Québec (FRSQ) fellowships to Aida Faber.

References

Ainsworth

MDS

Blehar

Waters

. (1978) Patterns of Attachment: A Psychosocial Study of the Strange Situation. Hillsdale, NJ: Lawrence Erlbaum Associates.

Aldao

Nolen-Hoeksema

Schweizer

(2010) Emotion-regulation strategies across psychopathology: A meta-analytic review. Clinical Psychology Review 30: 217–237.

Alexander

Siegel

(2013) Perceived hunger mediates the relationship between attachment anxiety and emotional eating. Eating Behaviors 14: 374–377.

Anderson

Whitaker

(2011) Attachment security and obesity in US preschool-aged children. Archives of Pediatrics & Adolescent Medicine 165: 235–242.

Anderson

Gooze

Lemeshow

. (2012) Quality of early maternal-child relationship and risk of adolescent obesity. Pediatrics 129: 132–140.

Arend

Gove

Sroufe

(1979) Continuity of individual adaptation from infancy to kindergarten: A predictive study of ego-resiliency and curiosity in preschoolers. Child Development 50: 950.

Back

(2011) Effects of parental relations and upbringing in troubled adolescent eating behaviors. Eating Disorders: The Journal of Treatment & Prevention 19: 403–424.

Beydoun

Wang

(2009) Parent-child dietary intake resemblance in the United States: Evidence from a large representative survey. Social Science & Medicine 68: 2137–2144.

Birch

(1999) Development of food preferences. Annual Review of Nutrition 19: 41–62.

10.

Bouthillier

Tremblay

Hamelin

. (1996) Traduction et validation canadienne-française d’un questionnaire évaluant l’attachement chez l’adulte. [French-Canadian translation of a questionnaire evaluating adult attachment styles]. Revue canadienne des sciences du comportement [Canadian Journal of Behavioural Science] 28: 74–77.

11.

Bowlby

([1969] 1999) Attachment and Loss. New York, NY: Basic Books.

12.

Bretherton

(1992) The origins of attachment theory: John Bowlby and Mary Ainsworth. Developmental Psychology 28: 759–775.

13.

Caminard

(2011) Les défis de l’alimentation des enfants et adolescents québécois (9-18 ans); Enquête sur la santé dans les collectivités canadiennes, cycle 2.2, Nutrition (2004). Institut de la statistique du Québec Available at: http://www.stat.gouv.qc.ca/statistiques/sante/enfants-ados/alimentation/defis-alimentation-enfants-ados.pdf

14.

Cassidy

(1994) Emotion regulation: Influences of attachment relationships. Monographs of the Society for Research in Child Development 59: 228–249.

15.

Czaja

Rief

Hilbert

(2009) Emotion regulation and binge eating in children. International Journal of Eating Disorders 42: 356–362.

16.

Davis

(1998) Predictors of binge eating: The roles of attachment and dissociation. Dissertation Abstracts International: Section B: The Sciences and Engineering, College of Arts and Sciences, Georgia State University, Atlanta, GA, p. 184.

17.

Drewnowski

Almiron-Roig

(2010) Human perceptions and preferences for fat-rich foods. In: Montmayeur

le Coutre

(eds) Fat Detection: Taste, Texture, and Post Ingestive Effects. Boca Raton, FL: CRC Press, pp. 265–294. Available at: http://www.ncbi.nlm.nih.gov/books/NBK53528/

18.

Evers

Marijn Stok

de Ridder

DTD

(2010) Feeding your feelings: Emotion regulation strategies and emotional eating. Personality and Social Psychology Bulletin 36: 792–804.

19.

Finzi-Dottan

Manor

Tyano

(2006) ADHD, temperament, and parental style as predictors of the child’s attachment patterns. Child Psychiatry and Human Development 37: 103–114.

20.

Garriguet

(2007) Plus de fruits, moins de gras: Les habitudes alimentaires des canadiens. Transition Hiver 7–10. Available at: http://www.vanierinstitute.ca/include/get.php?nodeid=1422

21.

Gillath

Giesbrecht

Shaver

(2009) Attachment, attention, and cognitive control: Attachment style and performance on general attention tasks. Journal of Experimental Social Psychology 45: 647–654.

22.

Goossens

Braet

Bosmans

. (2011) Loss of control over eating in pre-adolescent youth: The role of attachment and self-esteem. Eating Behaviors 12: 289–295.

23.

Goossens

Braet

Van Durme

. (2012) The parent-child relationship as predictor of eating pathology and weight gain in preadolescents. Journal of Clinical Child & Adolescent Psychology 41: 445–457.

24.

Goossens

Braet

Van Vlierberghe

. (2009) Loss of control over eating in overweight youngsters: The role of anxiety, depression and emotional eating. European Eating Disorders Review 17: 68–78.

25.

Han

(2009) Emotion Regulation, Coping, and Attachment in Bingeing Behaviors. West Lafayette, IN: Purdue University, p. 171.

26.

Harrison

Sullivan

Tchanturia

. (2010) Emotional functioning in eating disorders: Attentional bias, emotion recognition and emotion regulation. Psychological Medicine 40: 1887–1897.

27.

Hazan

Shaver

(1987) Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology 52: 511–524.

28.

Hebebrand

Herpertz-Dahlmann

(2009) Psychological and psychiatric aspects of pediatric obesity. Child and Adolescent Psychiatric Clinics of North America 18: 49–65.

29.

Johnson-Down

Ritter

Starkey

. (2006) Primary food sources of nutrients in the diet of Canadian adults. Canadian Journal of Dietetic Practice & Research 67: 7–13.

30.

Kobak

Sceery

(1988) Attachment in late adolescence: Working models, Affect regulation, and representations of self and others. Child Development 59: 135–146.

31.

Koskina

Giovazolias

(2010) The effect of attachment insecurity in the development of eating disturbances across gender: The role of body dissatisfaction. Journal of Psychology: Interdisciplinary and Applied 144: 449–471.

32.

Lehoux

Howe

(2007) Perceived non-shared environment, personality traits, family factors and developmental experiences in bulimia nervosa. British Journal of Clinical Psychology 46: 47–66.

33.

McLaren

(2007) Socioeconomic status and obesity. Epidemiologic Reviews 29: 29–48.

34.

Main

Hesse

Goldwyn

(2008) Studying differences in language usage in recounting attachment history: An introduction to the AAI. In: Howard

Steele

(eds) Clinical Applications of the Adult Attachment Interview. New York, NY: Guilford Press, pp. 31–68.

35.

Mikulincer

Shaver

(2007) Attachment processes and emotion regulation. In: Mikulincer

Shaver

(eds) Attachment in Adulthood: Structure, Dynamics, and Change. New York, NY: Guilford Press, pp. 188–218.

36.

Mikulincer

Shaver

(2012) An attachment perspective on psychopathology. World Psychiatry 11: 11–15.

37.

Moriarty-Kelsey

Daniels

(2010) Childhood obesity is the fuel that fires adult metabolic abnormalities and cardiovascular disease. Childhood Obesity 6: 250–256.

38.

Muris

Meesters

van Melick

. (2001) Self-reported attachment style, attachment quality, and symptoms of anxiety and depression in young adolescents. Personality and Individual Differences 30: 809–818.

39.

National Cancer Institute (2010) Risk factor monitoring and methods. Available at: http://riskfactor.cancer.gov/tools/children/review/agegroups/schoolage/ (accessed 6 June 2014).

40.

Nolen-Hoeksema

(2012) Emotion regulation and psychopathology: The role of gender. Annual Review of Clinical Psychology 8: 161–187.

41.

Patton

(1992) Fear of abandonment and binge eating. A subliminal psychodynamic activation investigation. Journal of Nervous and Mental Disease 180: 484–490.

42.

Pottharst

(1990) The search for methods and measures. In: Pottharst

(ed.) Research Explorations in Adult Attachment. New York, NY: Peter Lang Publishing, 9–37.

43.

Prättälä

Paalanen

Grinberga

. (2007) Gender differences in the consumption of meat, fruit and vegetables are similar in Finland and the Baltic countries. European Journal of Public Health 17: 520–525.

44.

Prichard

Hodder

Hutchinson

. (2012) Predictors of mother–daughter resemblance in dietary intake. The role of eating styles, mothers’ consumption, and closeness. Appetite 58: 271–276.

45.

Roelofs

Meesters

ter Huurne

. (2006) On the links between attachment style, parental rearing behaviors, and internalizing and externalizing problems in non-clinical children. Journal of Child and Family Studies 15: 319–332.

46.

Sharpe

Killen

Bryson

. (1998) Attachment style and weight concerns in preadolescent and adolescent girls. International Journal of Eating Disorders 23: 39–44.

47.

Simpson

Rholes

Nelligan

(1992) Support seeking and support giving within couples in an anxiety-provoking situation: The role of attachment styles. Journal of Personality and Social Psychology 62: 434–446.

48.

Singer

(1998) Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. Journal of Educational and Behavioral Statistics 23: 323–355.

49.

Sroufe

(1979) The coherence of individual development: Early care, attachment, and subsequent developmental issues. American Psychologist 34: 834–841.

50.

Swinburn

Sacks

Hall

. (2011) The global obesity pandemic: Shaped by global drivers and local environments. The Lancet 378: 804–814.

51.

Tabacchi

Giammanco

La Guardia

. (2007) A review of the literature and a new classification of the early determinants of childhood obesity: From pregnancy to the first years of life. Nutrition Research 27: 587–604.

52.

Thompson

Subar

(2008) Dietary assessment methodology. In: Coulston

Boushey

(eds) Nutrition in the Prevention and Treatment of Disease. 2nd ed. San Diego, CA: Academic Press, pp. 3–22.

53.

Tucker

McNamara

(1995) Assessing the relationship between parents’ object relations and their daughters’ eating disturbances. Eating Disorders: The Journal of Treatment & Prevention 3: 311–323.

54.

Volkow

Wang

G-J

Fowler

. (2008) Overlapping neuronal circuits in addiction and obesity: Evidence of systems pathology. Philosophical Transactions of the Royal Society B: Biological Sciences 363: 3191–3200.

55.

Wilkinson

Rowe

Heath

(2013) Eating me up inside: Priming attachment security and anxiety, and their effects on snacking. Journal of Social and Personal Relationships 30: 795–804.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.48 MB