Prospective Pathways to Depressive Symptoms and Disordered Eating in Adolescence: A 7-Year Longitudinal Cohort Study

Research indicates a co-occurrence and increase in disordered eating and depressive symptoms throughout adolescence (Goldschmidt et al. 2016), both of which predict negative impacts on health in adulthood (Bornioli et al. 2019; Johnson et al. 2009). This highlights the importance of early prevention, which could be aided through the identification of common risk factors. The Dual-Pathway model proposes pathways to the development of eating pathology and depressive symptoms via BMI, perceived sociocultural pressure for thinness, and body dissatisfaction (Stice et al. 1996). Central components of the model have received individual empirical support for the prediction of eating pathology and depressive symptoms among pre-adolescents (e.g., Evans et al. 2017). However, the interrelated pathways have not undergone prospective exploration among this younger group. Puberty has also emerged as a risk factor for eating pathology and depressive symptoms (Lewis et al. 2018), yet research has neglected the individual impact of externally visible pubertal indicators, including breast development for girls and height for boys, which may elicit social reactions that have an adverse impact on the individual. The present study therefore aimed to fill these gaps in knowledge by examining risk factors for, and interrelated pathways to, eating pathology and depressive symptoms in adolescence. This was done by evaluating an adaptation of the Dual-Pathway model of disordered eating across seven years, spanning childhood to early adolescence (see Fig. 1).

With regard to the first hypothesis, findings differed by gender. For girls, early BMI was found to predict disordered eating and depressive symptoms in adolescence via different pathways in the model. With regard to the outcome of disordered eating, effects of childhood BMI were mediated by pre-adolescent BMI, depressive symptoms, and body dissatisfaction. The mediating influence of body dissatisfaction supports previous findings (Jendrzyca and Warschburger 2016) and indicates that girls of higher adiposity are vulnerable to engaging in disordered eating due to dissatisfaction with their bodies. However, the differing impact of BMI at ages 7 and 11 on pre-adolescent depressive symptoms was unexpected. This is the first study to reveal adverse impacts associated with lower childhood BMI on subsequent depressive symptoms and contradicts previous prospective research conducted among this group (Evans et al. 2017). There may be several different reasons for these inconsistent findings. One may be the differing use of measures. For example, Evans et al. ( 2017) employed the short form of the Children’s Depression Inventory (CDI-S; Kovacs 1992), which unlike the SMFQ employed in the present study, includes an item related to appearance: (“I look ugly”). Children of a higher BMI have been found to experience greater weight stigma (Jendrzyca and Warschburger 2016), and thus may deem themselves less attractive, which may give rise to greater leading them to indicate a higher degree of agreement with this item on the CDI-S. An alternative consideration is that girls as young as 7 may be less aware of, or vulnerable to, societal standards for thinness. Instead, they may value heavier bodies, as they have been associated with muscle and stature relating to sporting success (Tatangelo and Ricciardelli 2013). Finally, this unexpected finding could be an anomaly, and BMI at such an early age may not be associated with depressive symptoms. Indeed, a prospective study which examined this association at regular intervals between the ages of 2 and 12 found BMI to be influential from 8 years of age, with the authors suggesting that the social-cognitive processes responsible for forging this connection (e.g., experiencing weight stigma, body dissatisfaction) may not undergo development until this age (Bradley et al. 2008). Nonetheless, these are all speculations and further examination of this unexpected finding is needed. The subsequent impact of depressive symptoms on disordered eating mirrors previous findings (Gardner et al. 2000), and suggests that girls who engage in disordered eating may do so due to negative affect, whereby they engage in “comfort” eating to cope with their feelings rather than trying to alter their body weight and shape (Stice et al. 1996). Finally, the direct pathway from pre-adolescent BMI to disordered eating supports previous findings (Reed et al. 2017). With regard to the outcome of depressive symptoms in adolescence, the first two identified pathways via pre-adolescent depressive symptoms and BMI at 11 years have already been discussed. Contrary to predictions, body dissatisfaction was not found to predict later depressive symptoms. This suggests that while girls of higher adiposity may experience negative effect, this may not be due to feelings of body dissatisfaction. This emphasizes that the direct effects of BMI on both disordered eating and depressive symptoms warrant further exploration, as other variables not currently accounted for in the model may be acting as mediators, including internalization of appearance ideals (Evans et al. 2013), weight teasing/stigma (Vartanian and Porter 2016), and “body talk” (conversations about appearance which reinforce appearance pressures; Strandbu and Kvalem 2014).

With regard to the first hypothesis in relation to the boys, there was greater support for the pathways leading to eating pathology within the model. The identification of the first impact of childhood BMI via pre-adolescent BMI, and subsequently body dissatisfaction, supports previous research which found that boys with higher adiposity were more likely to develop disordered eating due to dissatisfaction with their body (Jendrzyca and Warschburger 2016). The second direct pathway from pre-adolescent BMI to disordered eating mirrors previous findings (e.g., Reed et al. 2017) and simultaneously suggests that BMI may exert additional effects via other mediators. One potential mediator may be weight stigma (including weight-related teasing). Cross-sectional research has consistently identified an association between the experience of weight stigma and disordered eating (Vartanian and Porter 2016), with prospective research also having identified weight teasing as a predictor among boys (Haines et al. 2006). Therefore, it could be argued that boys of higher adiposity experience teasing about their weight, and therefore engage in restrictive eating to avoid future experiences of weight stigma and teasing. Finally, the independent direct effect of pre-adolescent depressive symptoms on disordered eating may be unrelated to BMI or body dissatisfaction. This previously observed effect (Gardner et al. 2000) may be due to boys engaging in disordered eating for self-soothing of depressive symptoms and not in an attempt to control their weight. Contrary to predictions, only the prevalence of pre-adolescent depressive symptoms emerged as a risk factor for depressive symptoms in adolescence. This risk factor has been identified previously (Heerde et al. 2018) and highlights the persistent nature of negative affect among boys. The absence of evidence connecting BMI with depressive symptoms supports previous findings (Evans et al. 2017), which indicates that the association between adiposity and negative affect tends to emerge later in childhood among boys. Similarly, the present findings suggest that body dissatisfaction does not foster depressive symptoms, and thus replicates previous studies (Paxton et al. 2006) while challenging others (Goldschmidt et al. 2016). However, research has also indicated that body dissatisfaction becomes a salient risk factor for depressive symptoms later in adolescence (Paxton et al. 2006). Collectively, this implies that BMI and body dissatisfaction may be less influential in the development of depressive symptoms among boys in earlier adolescence. The conflicting findings in the literature also highlight the importance of developing measures to capture male-specific body image concerns. Many traditional measures, including the figure rating scale utilized in the present study (Dowdney et al. 1995), tend to focus on weight and shape concerns, and are thus unlikely to be capturing male concerns, such as muscularity (Ricciardelli and McCabe 2004). Nonetheless, there are likely to be other aspects of boys’ self-esteem which might influence and contribute towards depressive symptoms, such as perceived sports ability (Slutzky and Simpkins 2009).

The second hypothesis regarding the adverse impact of more advanced breast development on the development of eating pathology and depressive symptoms among girls was supported. This study is also the first to examine breast development as a risk factor for disordered eating, and advances prior research which found earlier puberty, measured via other indicators, to predict disordered eating (Baker et al. 2012). The present findings may be attributable to girls with more advanced breast development feeling unprepared for the emotional and social changes associated with their visible indicator of maturity (e.g., Ge et al. 2003). This group may feel uncomfortable with the associated objectifying and sexualising attention they experience from adolescent boys, while also experiencing teasing from later-developing peers (Grower et al. 2019). This may lead them to engage in disordered eating in an attempt to “lose” their breasts, and thus receive less attention. An alternative explanation may be that adolescent girls with advanced breast development are being pushed away from the Westernized appearance ideal for women, and thus pursue weight-control behaviors in an attempt to change their bodies (Bulik 2002). These postulations may also relate to the observed adverse impact of advanced breast development on the development of depressive symptoms, which mirrors previous findings (e.g., Joinson et al. 2012). However, these potential mediating mechanisms are all speculative and warrant further exploration.

The third hypothesis regarding the adverse impact of shorter stature on the development of eating pathology and depressive symptoms among boys was not supported. While the effects of childhood and pre-adolescent BMI on height strengthen previous cross-sectional findings (Bosy-Westphal et al. 2009), the absence of subsequent effects on depressive symptoms contradicts other cross-sectional research (Rees et al. 2009). However, the latter study found that among adolescents aged 12–19 years, an association between height and depressive symptoms only emerged among boys aged 14 and above (Rees et al. 2009), which suggests that height may become more influential on depressive symptoms later in adolescence. In contrast, the impact of shorter stature on disordered eating approached significance ( p = 0.08), which lends partial support to previous prospective findings (Gardner et al. 2000). As a Western appearance ideal for men (Fornari and Dancyger 2003), it is not surprising that height has been found to be an important source of appearance satisfaction among adult men (Griffiths et al. 2019). In view of this, the present findings raise the possibility that boys of shorter height may experience dissatisfaction with this uncontrollable aspect of their appearance, which consequently leads them to engage in compensatory disordered eating behaviors in an attempt to change other aspects of their appearance; for example, to attain a slim and muscular body shape. However, this supposition warrants further examination, in addition to previously indicated moderating factors, such as conformity to masculine norms (O’Gorman et al. 2019). As may be the case with depressive symptoms, it is possible that height becomes increasingly influential on disordered eating later in adolescence. Indeed, increasing numbers of boys engage in romantic relationships the further they move through adolescence (Carver et al. 2003), and research indicates that those who are taller or at later stages of puberty are deemed more popular, attractive, and are more likely to be dating compared to their shorter peers (Cawley et al. 2006). Thus, there is likely to be increasing pressure relating to height as boys move through adolescence. This warrants further exploration and highlights the importance of assessing multiple aspects associated with male appearance ideals, such as muscularity and height.

While the present findings have important implications, the limitations warrant consideration. First, the sample lacks socio-economic and ethnic diversity, indicating that the findings may be less generalizable. Second, the measure of disordered eating was not a validated tool, and instead was based on a series of questions relating to eating behaviors that were available for the cohort. However, the questions were based on those of a validated instrument (e.g., Youth Risk Behavior Surveillance System questionnaire; Kann et al. 2000). Third, while a simple cross-sectional measure of one indicator of puberty was employed, the difficulty of identifying truly unique effects of a particular pubertal indicator should be acknowledged, given that developmental maturation across the different indicators tends to be highly correlated (Schubert et al. 2005). Fourth, although this is a longitudinal study, three of the key variables (BMI, body image, and depression) were measured at approximately the same age (11 years). Therefore, while strong evidence of associations between these variables has been found, it is not possible to draw conclusions about the direction of causation. Finally, the data was collected in the 1990s and 2000s, and thus does not account for the more recent impact of social media, which has been found to predict disordered eating and depressive symptoms (Primack et al. 2017). It is therefore recommended that future research replicates the present examination in more recent cohorts, while accounting for this additional influence.

Nonetheless, this study has a number of strengths. First, it has tested an adaptation of a well-supported theory, the Dual-Pathway of disordered eating (Stice et al. 1996), and has identified novel etiological pathways to the development of eating pathology and depressive symptoms among adolescents, including the impact of breast development stage among girls. Second, the present study has focused on symptoms of disordered eating and depressive symptoms, as distinct from their clinical threshold classifications. This has enabled a broader exploration of etiological pathways to disordered eating and depressive symptoms at a population level and has important population-based implications. Given the risk factors of BMI, body dissatisfaction, and external indicators of puberty; the implementation of early interventions which target these variables is indicated. However, instead of implementing programmes promoting weight loss among children, school-based interventions which foster positive body image and promote the acceptance of diverse bodies and appearance are recommended (Bray et al. 2018). Such an approach would also help prepare preadolescents for puberty and it associated bodily changes.