Overweight and obesity in children pose a major public health concern worldwide. In Australia, it is estimated that 25% of school-aged children are overweight or obese [4]. The consequences of childhood obesity are extensive, impacting on physical and psychosocial health. The well-documented adverse physical health consequences include high blood pressure, high cholesterol, metabolic syndrome, type 2 diabetes, orthopaedic problems, sleep apnoea, asthma and fatty liver disease, e.g. [27, 31]. Research into the psychological, social and behavioural consequences of childhood obesity shows that obese children are also at increased risk of body dissatisfaction, low self-esteem, social isolation and discrimination, depression and reduced quality of life, e.g. [5, 32].

We have previously applied cross-sectional data from the Childhood Growth and Development (GAD) Study to characterize psychosocial impairment in overweight and obese children relative to healthy weight children [14]. In the GAD Study, higher body mass index (BMI) z scores were associated with increases in depression, body dissatisfaction, eating disorder symptoms, internalizing and externalizing problems and frequency of being bullied, and with decreasing self-esteem and quality of life. Similar to other studies, we also identified sex differences in these relationships, with girls reporting greater psychosocial impairment associated with excess weight than boys. While treatment-seeking samples of obese children have been found to have more severe psychosocial impairments [12], the GAD Study showed psychosocial distress to be prevalent in both treatment-seeking and community samples of overweight/obese children. Hence, our previous cross-sectional work suggests that even in young primary school-aged children, the psychosocial burden of excess weight is significant and broad-reaching.

Existing research suggests that overweight and obese children are at risk of remaining overweight or obese. An Australian cohort study found that BMI was stable over a 9-year period among boys and girls [7]. All participants who were obese at age 18 had been overweight or obese at the age of 15 years. Further, of those who were obese at age 18, 91% had been overweight or obese at the age of 12 years and 90% had been overweight or obese at the age of 9 years. Similar findings emerged in a separate population-based cohort study, where approximately two thirds of the children who were overweight or obese at age 5 were still overweight at age 14 [25]. More recently, Gibson et al. (2014) found that, on average, child BMI z scores did not change significantly in the GAD Study sample over 3 years. These results applied to overweight and obese as well as healthy weight children, and to boys and girls [15].

Given the tracking of weight over time, there is a growing interest in understanding the persistence of the health consequences of childhood obesity. In the physical health literature, there is evidence that the metabolic and cardiovascular effects of obesity in childhood persist and contribute to a link between childhood obesity and adult morbidity [31]. Conversely, there are limited studies examining the course of psychosocial functioning among children with obesity. Moreover, there is mixed evidence regarding the nature and direction of the relationship between overweight/obesity and psychosocial difficulties [21]. This makes it particularly important to examine longitudinal changes in both weight and psychosocial functioning. Viner and Cole (2005) examined socioeconomic, education, social and psychological outcomes of childhood obesity among a British birth cohort. Obesity at age 10 only was not associated with any adverse outcomes in adulthood for males or females. While persistent obesity (obesity at age 10 and age 30) was not associated with any adverse adult outcomes in men, it was associated with a higher risk of never having been gainfully employed and not having a current partner among women [39]. Results from this study suggest that obese boys may not be at risk of long-term psychosocial impairment, but that obese girls are at risk for psychosocial disadvantage if they remain obese. However, there is a need for additional research on how psychosocial functioning changes as overweight/obese children transition through adolescence, a key period for the development of mental health problems [33]. There is also a need for information on how psychosocial functioning varies over time in overweight/obese children relative to healthy weight children.

Given the above, the aim of the present study was to investigate the course of psychosocial difficulties over time for GAD Study children who were overweight or obese at baseline, and a comparison sample of GAD Study children who were healthy weight at baseline. If psychosocial difficulties in the overweight/obese group persist, evidence would be provided for the importance of attending to psychosocial functioning as part of standard management and treatment approaches for childhood overweight and obesity.

Methods

Design

The GAD Study is a prospective, enriched Western Australian cohort study designed to identify the origins and consequences of childhood obesity. It involved healthy weight, overweight and obese children aged between 6 and 13 years at baseline, together with their parent/s. As outlined elsewhere [1, 13], overweight and obese children were recruited from clinical and community settings. Healthy weight children were matched to the overweight/obese group for age and sex and drawn from primary schools. Participants were contacted twice a year for at least 2 years, with a comprehensive annual assessment and a briefer mid-year visit.

This study focuses on data collected at the baseline, 1-year follow-up and 2-year follow-up assessments between January 2004 and August 2008 for children who were aged at least 8 years at baseline. Younger children did not complete the full battery of GAD Study measures, precluding in-depth assessment of psychosocial functioning.

The study protocol was approved by the WA Women’s and Children’s Health Service Ethics Committee.

Participants

Participants were 212 GAD Study children (49% male) aged between 8 and 13 years at baseline (M = 10.05 years, SD = 1.41), who completed baseline, 1-year follow-up and 2-year follow-up assessments (M = 12.20, SD = 1.30 at the 2-year follow-up). At baseline, 54% of the children (n = 114) were a healthy weight, 30% were overweight (n = 64) and 16% (n = 34) were obese.

This sample represents 57% of the original cohort of children aged at least 8 years (n = 371). Differences between GAD Study families lost to follow-up and those retained in the study have been reported previously [15]. In brief, loss to follow-up was associated with child and maternal obesity and with family disadvantage (i.e. low maternal education, low family income). Similar variables have emerged as predictors of attrition in other longitudinal studies [42].

Measures

Measures were chosen to cover a broad range of psychosocial variables previously linked to weight in childhood and adolescence, as well as to assess anthropometric status.

Anthropometry

Children were weighed in light clothing with regularly calibrated Tanita Digital Medical Scales, and measured with a regularly calibrated portable Harpenden stadiometer. Both height and weight were measured twice, with the mean score recorded. Body mass index (BMI = weight / height2) was calculated for each child. This was used to determine age-specific and gender-specific BMI z scores using Centre for Disease Control reference data [29].

The Children’s Depression Inventory

The CDI [23] was used to assess depressive symptomatology. The CDI is the most commonly used instrument for assessing depressive symptoms in children aged 7–17 years. The short-form CDI contains 10 items, which assess negative mood, anhedonia, ineffectiveness and negative self-esteem. The CDI has good psychometric properties [9] and was internally consistent in this sample (α = 0.75).

The Self-Perception Profile for Children

The Self-Perception Profile for Children (SPPC) [19] was used to evaluate self-esteem and self-concept. It assesses six domains of self-esteem (scholastic competence, social acceptance, athletic competence, physical appearance, behavioural conduct and global self-worth). Each of these subscales contains six items, and subscale scores are calculated by averaging the appropriate item ratings. The SPPC has established psychometric properties [26], and in this sample, alpha coefficients exceeded 0.70 for all subscales.

Child quality of life (PedsQL)

The Paediatric Quality of Life Scale (PedsQL) [37] is designed to measure health-related quality of life in children and adolescents aged 2 to18 years. The self-report version was used for this study. The scale consists of 23 items that assess physical, emotional, social and school functioning. Three summary scores are obtained: total health, physical health and psychosocial health. Normative data indicate adequate reliability and validity for the self-report and parent-proxy report [24, 38], and the three subscales were internally consistent in this sample (αs > 0.80).

Bullying Questionnaire for children

The Bullying Questionnaire for children (BQ) was developed from the 38-item Friendly Schools, Friendly Families Intervention Project Bullying Questionnaire [10]. The BQ asked children how often they had been bullied by another group of students in the last term and how often they had taken part in bullying another student, either on their own or with a group of students, in the last term.

Students’ Life Satisfaction Scale

The Family Satisfaction subscale of the SLSS [20] was used to assess children’s level of satisfaction with their family, parents and the nature of family interactions. The subscale consists of seven items and has been established as reliable and valid [16, 20]. The alpha coefficient in this sample was 0.77.

The Child Eating Disorder Examination

The Child Eating Disorder Examination (ChEDE) [6] was used to assess eating disorder psychopathology. This structured interview is a modified version of the Eating Disorder Examination (EDE) for adults [11] which has been validated for use with children and previously used with overweight and healthy weight children [2, 35]. The ChEDE can generate four subscale scores (Restraint, Eating Concern, Shape Concern, Weight Concern) and a global score (the mean of the four subscales). Factor analysis has not provided support for the original four-subscale solution [22], but the global scale demonstrates good psychometric properties [2, 40] and was used for this study. The alpha coefficient was 0.90.

The Children’s Body Image Scale

The Children’s Body Image Scale (CBIS) [36] was used to measure body dissatisfaction. This gender-appropriate figure rating scale consists of seven composite photos with corresponding BMIs, ranging from 14.0 to 29.0 kg/m2 for boys and 13.0 to 28.5 kg/m2 for girls (3rd to the 97th BMI percentiles for 10-year-old children). Children are asked to select the figures that best represent their current and ideal body sizes, with the perceived-ideal discrepancy providing a measure of body dissatisfaction. The CBIS has been validated on an Australian sample of 7 to 12-year-old boys and girls [36].

Statistical analyses

As we have previously identified sex differences in the psychosocial correlates of childhood overweight/obesity [3, 14], all analyses were stratified by child sex. Models were also adjusted for child age, maternal BMI and family income, as these variables have been linked to child overweight and obesity and psychosocial functioning previously [13]. With regards to family income, 19% of the sample indicated that their income felt within the lowest population quartile for Australia.

Longitudinal changes in the psychosocial outcomes were examined using linear mixed models. These models account for correlations within individuals over time, and for the clustering of children within families. Two sets of linear mixed models were specified, and all models were adjusted for child age, maternal BMI and family income.

The first set of analyses specified a main effect of baseline weight status (overweight/obese vs. healthy weight), a main effect of time (baseline vs. 1-year follow-up vs. 2-year follow-up) and an interaction effect between weight status and time. These models focused on detecting any overall differences in psychosocial functioning across the overweight/obese and healthy weight groups (on average across the 2-year study period; the main effect of weight status), any overall changes in psychosocial functioning from baseline to 2 years (on average across both weight groups; the main effect of time) and any differences in how psychosocial functioning changes over time for overweight/obese and healthy weight children (the interaction effect). Of note, weight status was stable over time for children in the GAD Study, with fewer than 5% of children moving between different weight categories within the 2-year period. Thus, while models made use of baseline weight status as a predictor variable, this corresponded to weight status at 1 and 2-year follow-up in almost all cases.

The second set of analyses were stratified by baseline weight status, and run with a main effect of time, a main effect of BMI z score and an interaction effect between time and BMI z score. These models focused on detecting any effects of BMI z score within each weight category (main effect of BMI z score). Additionally, while BMI z scores were largely stable in the GAD Study sample over time, the models would allow associations between any small changes in BMI z score and any associated changes in psychosocial functioning to be identified (the interaction effect).

Given the large number of outcome variables (=16), we applied a Bonferroni correction to give an alpha level of p < 0.003. The clinical significance of identified between-group differences was also considered. Consistent with current approaches in psychosocial outcome research, e.g. [28, 30], a minimally important clinical difference was defined as a mean score for the overweight/obese group that was more than half a SD from the mean score of the healthy weight group.

Results

Longitudinal changes in psychosocial functioning for overweight/obese versus healthy weight children

Boys

After adjusting for child age, maternal BMI and family income, there were significant main effects of baseline weight status on boys’ physical appearance self-esteem (p = 0.001, Cohen’s d = 0.585), eating disorder symptoms (p < 0.001, Cohen’s d = 0.762) and body dissatisfaction (p < 0.001, Cohen’s d = 1.069). On average, boys who were overweight or obese at baseline had lower physical appearance self-esteem than boys who were a healthy weight at baseline, as well as higher eating disorder symptom and body dissatisfaction scores. In all cases, the mean score for the overweight/obese group was >0.5 SD from the mean score for the healthy weight group. For eating disorder symptoms and body dissatisfaction, differences were >1 SD from the mean score for the healthy weight group (see Table 1).

Table 1 Means (and standard deviations) for psychosocial outcomes in boys, by time and weight status at baseline
Full size table

There were no significant effects of weight status on the other psychosocial outcomes in boys (ps = 0.04–0.79).

There were no significant main effects of time on any of the outcome variables (ps = 0.06–0.99), nor were there interaction effects between baseline weight status and time (ps = 0.13–0.84). Thus, indices of psychosocial functioning were relatively stable over the 2-year study period for overweight and healthy weight boys (see Table 1).

Girls

After adjusting for child age, maternal BMI and family income, there were significant main effects of baseline weight status on girls’ depressive symptoms (p < 0.001, Cohen’s d = 0.520), global self-esteem (p < 0.001, Cohen’s d = 0.573), social self-esteem (p < 0.001, Cohen’s d = 0.453), athletic self-esteem (p < 0.001, Cohen’s d = 0.430), physical appearance self-esteem (p < 0.001, Cohen’s d = 0.893), total quality of life (p < 0.001, Cohen’s d = 0.683), physical quality of life (p < 0.001, Cohen’s d = 0.633), psychosocial quality of life (p < 0.001, Cohen’s d = 0.638), family satisfaction (p < 0.001, Cohen’s d = 0.278), frequency of being bullied (p < 0.001, Cohen’s d = 0.950), eating disorder symptom scores (p < 0.001, Cohen’s d = 1.014) and body dissatisfaction (p < 0.001, Cohen’s d = 1.110). Overweight and obese girls reported greater impairment in all of these areas than healthy weight girls. For depression, global self-esteem, physical appearance self-esteem, quality of life and bullying, mean scores for the overweight/obese group were >0.5 SD from the mean score for the healthy weight group. For body dissatisfaction, scores were >1 SD from the mean for the healthy weight group, and for eating disorder symptoms, scores were >2 SD from the mean for the healthy weight group. Group differences in social self-esteem, athletic self-esteem and family satisfaction did not meet criteria for clinical significance (see Table 2).

Table 2 Means (and standard deviations) for psychosocial outcomes in girls, by time and weight status at baseline
Full size table

There were no significant main effects of weight status on scholastic or behavioural self-esteem, school-related quality of life or frequency of bullying others (ps = 0.003–0.39).

There were no significant main effects of time on any of the outcome variables (ps = 0.09–0.80), nor were there interaction effects between baseline weight status and time (ps = 0.36–0.87). Thus, as seen for boys, indices of psychosocial functioning were relatively stable over the 2-year study period for overweight and healthy weight girls (see Table 2).

Longitudinal changes in psychosocial functioning within each weight group

Boys

Within the overweight/obese group, and after adjusting for covariates, significant main effects of BMI z score were identified for global self-esteem (p < 0.001), physical appearance self-esteem (p < 0.001), eating disorder symptom scores (p < 0.001) and body dissatisfaction (p < 0.001). Higher BMI z scores were associated with lower self-esteem and greater eating disorder symptoms and body dissatisfaction among overweight/obese boys. There were no significant main effects of BMI z score on the other psychosocial outcome variables (ps = 0.005–0.76). As previously, there were no significant main effects of time on any of the psychosocial outcome variables (ps = 0.08–99), nor were there any significant interaction effects between time and BMI z-score (ps = 0.03–97) (see Table 1).

Within the healthy weight group, there were no significant main effects of BMI z score (ps = 0.01–0.79), no significant main effects of time (ps = 0.23–96) and no significant interaction effects between time and BMI z score (ps = 0.11–0.66).

Girls

Within the overweight/obese group, and after adjusting for covariates, significant main effects of BMI z score were identified for global self-esteem (p = 0.002), physical appearance self-esteem (p < 0.001), eating disorder symptom scores (p = 0.002) and body dissatisfaction (p < 0.001). Increasing BMI z scores were associated with lower self-esteem and greater eating disorder symptoms and body dissatisfaction among overweight/obese girls. There were no significant main effects of BMI z score on the other psychosocial variables (ps = 0.009–0.62). There were no significant main effects of time on any of the psychosocial outcomes (ps = 0.29–0.88), nor were there any significant interaction effects between time and BMI z score (ps = 0.47–0.96) (see Table 2).

Within the healthy weight sample, significant main effects of BMI z score were identified for eating disorder symptom scores (p = 0.002) and body dissatisfaction (p < 0.001), but not for other outcomes (ps = 0.04–0.96). Again, increasing BMI z scores were associated with increasing eating disorder symptoms and body dissatisfaction. There were no significant main effects of time on any of the psychosocial outcomes (ps = 0.11–96), nor were there any significant interaction effects between time and BMI z score (ps = 0.32–0.98) (see Table 2).

Discussion

This study aimed to determine if the psychosocial difficulties associated with excess weight in childhood would persist over time. Results suggest that psychosocial difficulties show considerable stability in childhood, for overweight/obese and healthy weight children. In this sample, children with impaired psychosocial functioning at baseline were at risk for ongoing impairments over the next 2 years.

Consistent with the existing literature, overweight and obese girls showed greater psychosocial impairment across a range of areas (eating pathology, body dissatisfaction, depressive symptoms, self-esteem, quality of life, family dissatisfaction and bullying) when compared to healthy weight girls. In almost all cases, these differences satisfied criteria for clinical significance. The evidence for broad reaching psychosocial impairments among overweight and obese girls may add weight to recent suggestions that obesity and psychological problems are not separate health conditions but share common biological mechanisms [17] and may be the result of bidirectional interactions along the gut-brain axis [34]. Overweight and obese boys differed from their healthy weight counterparts on a smaller number of variables, and these were limited to weight-related constructs: body dissatisfaction, eating disorder symptoms and physical appearance self-esteem. Again, however, these differences were clinically important.

Within the overweight/obese group, higher BMI z scores (i.e. greater degree of overweight) were associated with lower self-esteem and greater body dissatisfaction and eating disorder symptoms, for boys and girls. This pattern has been reported elsewhere [5, 35] and suggests that psychosocial impairment in overweight/obese children may be expected to increase as weight increases. Moreover, even among healthy weight children, higher BMI z scores were associated with greater body dissatisfaction and eating disorder symptoms in girls.

Previous studies have tended to examine associations between weight status and psychosocial functioning cross-sectionally, or in prospective analyses that consider the effects of weight at one time point on psychosocial outcomes at a later assessment. This study adds to prior research by jointly modelling longitudinal changes in weight and psychosocial functioning. Results highlight that, in the context of stable weight, overweight/obese children who experience psychosocial difficulties are likely to continue to experience these difficulties. Given that overweight/obesity tends to track from early childhood onwards [7, 25], this is a key clinical finding and suggests that overweight/obese children may be at risk for prolonged, long-term psychosocial impairments across their formative years. This complements previous findings that suggest that physical impairments also persist in overweight/obese children who remain overweight into adulthood [31].

To avoid long-term impairments in functioning, direct attention to psychosocial well-being should be an important component of management approaches for childhood obesity. The results of this research suggest that overweight and obese children who report impairments in psychosocial well-being will not experience a natural remission of those difficulties. Accordingly, access to psychological support should be offered as part of standard care for overweight and obese young people. While primary healthcare is the ideal setting with regards to the management of childhood obesity, there are a number of limitations (e.g. support, confidence, knowledge and time) that impact on this [8, 41]. Hence, much work is needed in both the primary and secondary health setting with regards to identifying and treating childhood overweight and obesity and the associated physical and psychosocial impairments.

There are a number of strengths to this research. As noted, very little data are available on the long-term psychosocial outcomes of childhood overweight and obesity. This study adds to the available literature using a community-based sample of boys and girls, followed over a 2-year period, with data on a broad range of psychosocial variables collected using well-validated measures. We were also able to adjust for relevant covariates and, by collecting data on healthy weight as well as overweight/obese children, compare results across the two weight categories. Limitations of the work include the relatively short period of follow-up, the lack of data through to late adolescence, the associated lack of data on how pubertal status may impact on links between weight and psychosocial functioning, a modest sample size and attrition over time. The attrition rate in this study (43%) is comparable to that seen in other longitudinal studies of this type [18, 42], but makes replication of the results important, particularly in those subgroups more likely to be lost to follow-up. Lastly, it deserves note that in this study, weight status and psychosocial functioning both showed considerable stability over time. We are not able to comment on whether psychosocial functioning improves if overweight/obese children return to a healthy weight, or conversely, if healthy weight children increase in weight to become overweight.

In conclusion, this study provides new data on the course of psychosocial difficulties in overweight/obese and healthy weight children followed over a 2-year period. Results suggest that overweight and obese children are at risk of ongoing psychosocial distress from middle childhood into early adolescence.