Introduction

Alexithymia, a construct defined by difficulty identifying feelings, difficulty describing feelings, and an externally oriented cognitive style (i.e., focus on external and concrete events and experiences versus internal experiences) [1], has received extensive attention in studies of eating and weight disorders [2, 3]. This construct has been implicated in the development and maintenance of disordered eating [4], and has also been linked with worse eating disorder treatment outcomes [5]. In particular, two core alexithymia subcomponents, difficulty identifying feelings and difficulty describing feelings, have been found to be positively associated with disordered-eating symptoms in clinical and non-clinical samples of adults and youth [3, 6,7,8]. Notably, research has shown that alexithymia is associated with both disordered-eating attitudes [8, 9], such as body weight and shape concerns, and disinhibited-eating behaviors, such as binge eating, eating in the absence of hunger, and emotional eating [10,11,12,13,14,15]. Additionally, meta-analytic evidence indicates that alexithymia is elevated among individuals with obesity [2], perhaps due in part to its association with disinhibited-eating behaviors linked with high weight, including binge eating and emotional eating [2, 14, 16].

The relationship of alexithymia with disordered and disinhibited eating has been found to be at least partially accounted for by emotional distress, including in the form of anxiety and depressive symptoms [17,18,19]. Alexithymia is moderately correlated with, but distinct from, symptoms of depression and anxiety [17, 18], and as such can persist among patients with disordered eating even when mood improves with treatment [19]. Regarding depressive symptoms, it has been posited that alexithymia is associated with several aspects of how one perceives and responds to emotional situations and stimuli, giving rise to maladaptive emotional coping and emotion dysregulation processes that promote those symptoms [9, 20,21,22,23,24]. Further, ineffectiveness with regulating and processing social emotions may impede social relationships, contributing to low self-esteem and the development of self-critical depressive thinking [25,26,27,28,29]. In turn, depressive symptoms arising from alexithymia-related processes may promote certain types of disordered-eating attitudes and disinhibited-eating behaviors. For example, disordered-eating attitudes, such as concerns about one’s eating, shape, and weight, may be more likely to develop or occur in the context of self-dislike and low self-esteem that are characteristic of depressive symptomatology [12, 30]. Consistent with this theory, prior research has found significant associations of depressive self-criticism and dislike with disordered-eating symptoms [31, 32]. Similarly, theoretical models and empirical evidence suggest that negative emotions and aversive self-oriented cognitions are common precipitants of disinhibited-eating behaviors, which may reflect maladaptive efforts to cope with or avoid these unwanted internal experiences [33,34,35,36]. Moreover, certain specific forms of disinhibited eating, such as emotional eating (i.e., tendency to overeat in response to negative emotions such as anxiety or irritability) [37], are specifically conceptualized as arising from the experience of aversive affect (e.g., depressive symptoms) [38, 39].

Much of the literature investigating associations between alexithymia, depression, and disordered and disinhibited eating has focused on adult samples. A limited body of research on adolescents suggests that alexithymia is a stable trait that is associated with disordered-eating symptoms [15, 40, 41], maladaptive coping [42], and internalizing and externalizing behaviors [43]. However, there remains a need to better understand these interrelationships in adolescence, given that this is a unique and important period of the lifespan in terms of both socio-emotional development [44, 45] and risk for onset of disordered-eating attitudes and behaviors [46, 47]. One subgroup of adolescents for whom these associations may be salient is adolescent military dependents. Children of military service members face unique psychosocial and emotional stressors, including frequent parental deployments and relocations, concerns about parental safety, and military emphasis on fitness and weight standards that may be modeled or communicated to youth [48,49,50]. Further, compared to civilian peers, adolescent military dependents have been found to display greater disinhibited-eating behaviors, as well as more severe depressive symptoms [51, 52]. Notably, given that adolescents with high weight also are at greater risk for increased depressive symptoms and disordered eating [53,54,55], military dependent youth with high weight may represent an especially vulnerable population.

We therefore investigated the indirect association of alexithymia with disordered-eating attitudes and disinhibited-eating behaviors through depressive symptoms. Specifically, two core components of alexithymia (i.e., difficulty identifying feelings and difficulty describing feelings) were examined, along with both generalized disordered-eating attitudes and emotional eating, the latter of which is a form of disinhibited-eating behavior of particular relevance to youth [39, 56]. We hypothesized that in the current sample of adolescent military dependents with high weight, adjusting for trait anxiety, there would be significant indirect paths between each of the alexithymia components and the disordered- and disinhibited-eating variables through depressive symptoms.

Method

Participants and procedure

At-risk adolescent military dependent boys and girls (between 12 and 17 years) were recruited as part of an adult obesity and binge-eating disorder prevention trial (ClinicalTrials.gov ID#: NCT02671292). Participants were TRICARE-eligible (i.e., a healthcare program for military service members, retirees, and their families), and were considered at-risk based on having a body mass index (BMI, kg/m2) ≥ 85th percentile for their age and sex [57] and either elevated symptoms of anxiety (≥ 32 on the Trait scale of the State-Trait Anxiety Inventory for children) [58,59,60] or at least one self-reported episode of loss-of-control (LOC) eating during the previous 3 months [60, 61]. Individuals were excluded if they had a major psychiatric or medical condition (including pregnancy), current participation in psychotherapy or a structured weight-loss program, regular use of medications affecting appetite or body weight (unless dose and weight were stable in the last 3months), or weight loss exceeding 3% of their body weight in the last 3 months [60]. Potential participants were recruited through provider referrals, flyers posted military bases, listservs ads, and direct mailings to parents after being identified by the Defense Enrollment Eligibility Reporting System.

Participants attended a baseline screening visit where they completed questionnaires and interviews and had their height and fasting weight measured. Written assent and consent were obtained from participants and parents/guardians, respectively. Approval was provided by the Uniformed Services University of the Health Sciences Institutional Review Board and the Fort Belvoir Community Hospital Research Office.

Measures

Body composition

Height and fasting weight were measured with light clothing on, and shoes removed. BMI standard deviation scores (BMIz) for age and sex were calculated based on the Centers for Disease Control and Prevention standards [57].

Alexithymia

The Alexithymia Questionnaire for Children (AQC) [62] is an adaptation of the widely used Toronto Alexithymia Scale for adults [63]. The AQC has 20 items that are rated on a three-point Likert-type scale ranging from 0 = “Not true” to 2 = “Often true”, with higher scores indicating higher levels of alexithymia. The measure is comprised of three subscales: Difficulty Identifying Feelings (DIF), Difficulty Describing Feelings (DDF), and Externally Oriented Thinking. Based on conceptual and psychometric concerns regarding the Externally Oriented Thinking subscale [64], only the DDF and DIF subscales were used in this investigation. Cronbach’s alpha for all measures in the current sample are reported in Table 1.

Table 1 Bivariate correlations, descriptive statistics, and internal consistencies for primary variables
Full size table

Disordered-eating attitudes

The Eating Disorder Examination (EDE) interview v.14 OC/C.2 [65] assesses key attitudinal and behavioral symptoms of eating disorder psychopathology and was administered by trained interviewers. The EDE global score was used as a measure of generalized disordered-eating attitudes and is calculated by averaging scores across four subscales assessing restraint, eating concern, shape concern, and weight concern. The EDE also was used to assess the presence of LOC-eating episodes in the past 3 months for inclusion criteria.

Emotional eating

The Emotional Eating Scale adapted for use in children and adolescents (EES-C) [39] assesses eating in response to feeling states, and is based on the original Emotional Eating Scale for adults [38]. The EES-C lists 26 emotions across three domains (anger/anxiety/frustration, depression, and unsettled), and participants rate the extent to which each emotion stimulates a desire to eat using a five-point Likert-type scale from 0 = “No desire” to 4 = “Very strong desire.” In this study, the total score was used given evidence supporting its construct validity and the unidimensionality of the scale [66, 67].

Depressive symptoms

The Beck Depression Inventory-II (BDI-II) [68] assesses symptoms of depression (e.g., fatigue, sadness, and self-dislike) over the past 2 weeks. This scale has 21 items that are rated on a 4-point scale from 0 to 3, with higher scores indicating more severe depressive symptoms.

Trait anxiety

The State-Trait Anxiety Inventory for Children is a self-report measure of anxious arousal [59]. The 20-item trait subscale was used in this study, on which participants rate each item using a 3-point Likert-type scale from 1 = “Hardly ever” to 3 = “Often.” Higher scores reflect greater propensity to experience anxious arousal.

Data analysis

Data were screened for normality (skewness statistic values ranged from 0.131 to 0.911 and kurtosis from -0.869 to 0.451) and for multi-collinearity through evaluation of bivariate correlations, VIF (1.49 to 1.84), and tolerance values (0.543 to 0.672). No concerns were identified, and there were no multivariate outliers. Missing data for each variable was minimal, ranging from 0% (DDF) to 2.7% (EDE Global), and listwise deletion was used for all analyses.

Indirect paths between the alexithymia components (DIF; DDF) and the disordered-eating attitudes and emotional eating variables (EDE global and EES-C total, respectively) were analyzed using the PROCESS macro for SPSS 25.0 [63]. Specifically, four indirect path models through depressive symptoms (BDI-II) were examined (two for each dependent variable): (1) DIF to EDE global via BDI-II; (2) DDF to EDE global via BDI-II; (3) DIF to EES-C total via BDI-II; (4) DDF to EES-C total via BDI-II. This proposed ordering of variables is based on existing theory and findings suggesting that alexithymia can be conceptualized as a trait that may predispose individuals to onset of depressive symptoms that are episodic in nature, which may in turn promote disinhibited and disordered eating. In particular, the variable ordering reflected in these models is supported by findings that alexithymia typically develops in childhood and is relatively stable in adolescence [40, 69], alexithymia and emotional clarity prospectively predict depressive symptoms, alexithymia often persists when depressive symptoms remit [70,71,72,73], and internalizing symptoms (such as those characterizing depression) can precede the development of disordered eating [22].

In all models, depressive symptoms, disordered-eating attitudes, and emotional eating were initially adjusted for the following covariates: age, self-identified sex (coded as female = 0 or male = 1), race (coded as 0 = non-Hispanic White or 1 = People of Color), trait anxiety, LOC-eating status (coded as 0 = absence or 1 = presence), and BMIz. Trait anxiety was the only significant covariate in any model, thus for parsimony, trait anxiety was the only variable retained in final models. Coefficients were estimated with 95% bias-corrected bootstrapped confidence intervals with 5000 resamples; intervals that excluded zero were considered to indicate statistical significance [74]. Tests of direct paths were two-tailed and ps < 0.05 were considered significant.

Results

Participants were 149 adolescents (14.4 ± 1.6 years; 55.0% female; 37.3% non-Hispanic White, 20.0% non-Hispanic Black, 18.5% Hispanic, 12.6% multiple races, 3.7% Asian; BMIz: 1.9 ± 0.4). Means, standard deviations, internal consistencies, and Pearson pairwise correlations for the measures are presented in Table 1. Significant small-to-moderate positive correlations (rs = 0.19-0.26) were found between the alexithymia components and disordered-eating attitudes (DIF: p = 0.002; DDF: p = 0.002) and emotional eating (DIF: p = 0.01; DDF: p = 0.02). Depressive symptoms had moderate-to-large positive correlations (rs = 0.29-0.47) with the alexithymia components (DIF: p < 0.001; DDF: p < 0.001), disordered-eating attitudes (p < 0.001), and emotional eating (p < 0.001).

Disordered-eating attitudes

Adjusting for trait anxiety, there were significant indirect paths from each of the alexithymia components (DIF and DDF) to disordered-eating attitudes through depressive symptoms. Specifically, for the DIF model (see Fig. 1), greater DIF was associated with more depressive symptoms (p < 0.05), and in turn, greater depressive symptoms were associated with higher disordered-eating attitudes (p < 0.05). Similarly, for the DDF model (see Fig. 2), greater DDF was associated with more depressive symptoms (p < 0.01), and in turn greater depressive symptoms were associated with higher disordered-eating attitudes (p < 0.05). Neither the total associations of DIF and DDF with disordered-eating attitudes (DIF: p = 0.06; DDF: p = 0.13) nor the direct associations (DIF: p = 0.21; DDF: p = 0.52) were significant.

Fig. 1
figure 1

Model for the indirect path from difficulty identifying feelings to disordered-eating attitudes through depressive symptoms. *p < .05; for indirect path, †95% confidence interval excludes zero. Model adjusted for trait anxiety

Full size image
Fig. 2
figure 2

Model for the indirect path from difficulty describing feelings to disordered-eating attitudes through depressive symptoms. *p < .05; **p < .01; for indirect path, †95% confidence interval excludes zero. Model adjusted for trait anxiety

Full size image

Emotional eating

Adjusting for trait anxiety, the indirect paths from each of the alexithymia components to emotional eating through depressive symptoms were not significant. Specifically, for the DIF model (see Fig. 3), greater DIF was associated with more depressive symptoms (p < 0.05), but the association between depressive symptoms and emotional eating was not significant (p = 0.12). Similarly, for the DDF model (see Fig. 4), greater DDF was associated with more depressive symptoms (p < 0.01), but the association between depressive symptoms and emotional eating was not significant (p = 0.07). Neither the total associations of DIF and DDF with emotional eating (DIF: p = 0.13; DDF: p = 0.14) nor the direct associations (DIF: p = 0.28; DDF: p = 0.37) were significant.

Fig. 3
figure 3

Model for the indirect path from difficulty identifying feelings to emotional eating through depressive symptoms. *p < .05; Model adjusted for trait anxiety

Full size image
Fig. 4
figure 4

Model for the indirect path from difficulty describing feelings to emotional eating through depressive symptoms. **p < .01; Model adjusted for trait anxiety

Full size image

Discussion

This study examined indirect associations of alexithymia with disordered-eating attitudes and emotional eating through depressive symptoms in adolescent military dependents with high weight. As hypothesized, adjusting for trait anxiety, we found significant indirect paths from both core alexithymia components (difficulty identifying feelings and difficulty describing feelings) to disordered-eating attitudes through depressive symptoms. This finding is congruent with results from previous studies reporting similar indirect paths in adults [11, 75, 76]. However, in contrast, there were no significant indirect paths from either of the core alexithymia components to emotional eating through depressive symptoms. The total associations of the alexithymia components with disordered-eating attitudes and emotional eating were not significant, potentially due in part to the relationships of the trait anxiety covariate with the core variables in the models. Importantly, the paths from the alexithymia components to disordered-eating attitudes through depressive symptoms were small in magnitude, but comparable to indirect paths found in similar studies [5, 11, 77], and represent associations that may be modifiable in clinical settings given the range of effective treatments available for depressive symptoms.

Notably, both alexithymia components were significantly and positively associated with depressive symptoms. This is consistent with theories and empirical evidence highlighting the role of alexithymia as a potential risk factor for dysregulated emotionality broadly and depressive symptoms in particular [28, 42]. Conversely, the associations of depressive symptoms with disordered-eating attitudes and emotional eating in the respective models were not fully in accord with previous data. We did find that the association between depressive symptoms and disordered-eating attitudes was significant, which is consistent with the results of prior studies [30, 78], and could reflect the link between disordered-eating attitudes and cognitive symptoms of depression in particular. For example, cognitive symptoms, which have been found to be at least partially distinct from emotional and somatic symptoms [79], have been posited as sharing a key vulnerability with disordered eating in the form of low self-esteem [34, 80]. Moreover, empirical associations have been found between disordered-eating attitudes and specific cognitive depressive symptoms such as self-dislike and low self-worth [12, 30, 78].

However, the association of depressive symptoms with emotional eating was non-significant in the relevant models. This finding was unexpected given theories linking low mood with disinhibited-eating behaviors [34], and prior results in adults and youths suggesting a positive association [81, 82]. One possible explanation for this discrepancy is that emotional eating, which reflects the propensity to experience an urge to eat in response to a variety of emotional states (including anger, frustration, and anxiety), may not be prompted by depressive symptoms alone, but also by related aversive emotional states. For example, in this study the association between depressive symptoms and emotional eating may have been attenuated by adjusting for trait anxiety, which has been found to be significantly associated with emotional eating in the current and prior studies [83,84,85]. Additionally, the distinction between cognitive and emotional symptoms of depression may also account for the discrepant findings with regard to the significant association of depressive symptoms with disordered-eating attitudes, but not emotional eating.

Strengths and limitations

Study strengths include the unique sample of male and female adolescent military dependents, use of well-validated measures, and the theoretical foundation for the models. A primary limitation is the use of cross-sectional data. Although the order of variables in the model was based on theory and prior findings that alexithymia promotes depressive symptoms [40, 72, 73], scholars have also argued that alexithymia may result from depressed mood or represents a state (versus trait) variable occurring in the context of psychological distress [86]. However, the models investigated in this study were supported by evidence that emotional clarity prospectively predicts depression symptoms [72, 73], alexithymia operates as a stable trait in adolescence [40], and internalizing symptoms (such as those characterizing depression) can precede the development of disordered eating [22]. Second, given the nature of the sample, findings may not generalize to civilian youth, non-treatment-seeking adolescents, or youth without high weight. Third, the ACQ may have limitations as a measure of alexithymia, including potential relationships of the DIF scale with measures of emotional distress [87], and the use of self-report items to assess difficulties with identifying and describing emotions, which may be challenging for respondents [88]. Finally, the inclusion/exclusion criteria (including trait anxiety) and the restricted weight range of the sample may have contributed to different findings than might be observed in an adolescent sample with a greater range in body weight and severity of disordered eating or other psychopathology. Future research might further explicate the role of these and other factors that have been shown to influence the variables examined, including gender [89, 90], menstrual symptoms [91], impulsivity [92, 93], and anxiety [5, 75].

Conclusion

Disordered eating among adolescent military dependents may have negative impacts on already challenging military recruitment and retention efforts [94, 95], particularly given that a substantial proportion of U.S. military recruits have a family member who served [77]. Additionally, disordered-eating attitudes, disinhibited eating, and high weight can have wide-ranging impacts on health and quality of life [96, 97]. Understanding factors that may contribute to disordered-eating and the mechanisms by which certain factors influence risk are thus important areas of investigation. Although these relationships have been examined in adults, it is important to also evaluate the extent to which they are evident in adolescents, given the unique socio-emotional development [44, 45] that occurs during this period, which is also associated with elevated risk for onset of disordered-eating symptoms [46, 47]. The findings of this study suggest that increased depressive symptoms, while already an important psychosocial outcome to consider, may also represent an intervening variable between alexithymia and disordered-eating attitudes. In particular, links between depressive symptoms and disordered-eating attitudes may be salient among military dependent youth of higher weight. Further, military fitness and weight standards, which may be communicated to youths implicitly or explicitly by parents [48,49,50], may promote negative self-evaluative judgements characteristic of depressive thoughts and disordered-eating attitudes [98]. These youth also tend to experience greater weight stigma, which may exacerbate depressive self-dislike and self-critical attitudes regarding shape, weight, and eating [99,100,101]. Future studies will be needed to determine if the current findings generalize to the broader adolescent population, and whether the cross-sectional associations found here replicate with prospective data.

What is already known on this subject?

Prior research has found significant associations of depressive self-criticism and dislike with disordered-eating symptoms. Similarly, theoretical models and empirical evidence suggest that negative emotions and aversive self-oriented cognitions are common precipitants of disordered- and disinhibited-eating behaviors, which may reflect maladaptive efforts to cope with or avoid these unwanted internal experiences. Further, certain specific forms of disinhibited eating, such as emotional eating (i.e., tendency to overeat in response to negative emotions such as anxiety or irritability), are specifically conceptualized as arising in response to the experience of aversive affective states.

What does this study add?

The findings of this study suggest that increased depressive symptoms, while an important psychosocial outcome to consider independently, may also represent an intervening variable between alexithymia and disordered-eating attitudes. In particular, the potential link between depressive symptoms and disordered-eating attitudes may be salient for military-dependent youth of higher weight. Accordingly, there may be clinical utility in helping military-dependent youth learn skills to effectively identify, describe, and communicate their emotions in efforts to reduce emotional distress and depressive symptoms that are associated with disordered eating.