The relationship between childhood trauma, eating behaviours, and the mediating role of metacognitive beliefs

Metacognitions Thoughts are uncontrollable and dangerous


Introduction
Disordered eating is becoming increasingly recognised as a common worldwide public health concern that has been linked to significant impairments in physical and psychological functioning (Santomauro et al., 2021;Sheehan & Herman, 2015). The prevalence of disordered eating behaviours such as overconsumption, binge-eating, and rigid dieting, have significantly increased in Western countries such as Australia across the last two decades (da Luz et al., 2017). Many individuals do not seek help for these behaviours due to the social acceptability of dieting/fasting, eating for emotional coping purposes, and skipping meals (Neumark-Sztainer et al., 2011). However, disordered eating behaviours have been associated with a plethora of comorbid physical and mental health conditions such as higher BMI lower self-esteem and body dissatisfaction (Markey et al., 2022), poorer self-rated health and psychological distress (Kärkkäinen et al., 2018) and the presence of a greater number of metabolic cardiovascular disease risk factors (Lopez-Cepero et al., 2018).
Despite the growing body of literature that recognises the prevalence and detrimental impacts of disordered eating, there is surprisingly limited understanding of the psychological mechanisms that precipitate and perpetuate such eating behaviours. Experiences of childhood trauma have been identified as one important predisposing factor to the development of disordered eating behaviours Kong & Bernstein, 2009;Zelkowitz et al., 2021). While a range of possible mediators of the association between experiences of childhood maltreatment and disordered eating behaviours have been considered (e.g. pathological dissociation, difficulty with emotion self-regulation, body dissatisfaction, negative affect/depression, anxiety, general distress, self-criticism, and alexithymia; Rabito-Alcón et al., 2021), unfortunately the psychological mechanisms linking childhood trauma to disordered eating behaviours are still poorly understood. Research by Strodl and Wylie (2020) explored the relation between distinct forms of childhood trauma (emotional abuse, emotional neglect, physical abuse, physical neglect and sexual abuse) and disordered eating behaviours (cognitive restraint, uncontrolled eating and emotional eating) and investigated whether the diminished ability to monitor and describe emotional states (alexithymia) or beliefs about emotions mediated these associations. While alexithymia did not mediate these relationships, various indirect mediation effects were found for beliefs about emotions. In particular, the beliefs that emotions are overwhelming and uncontrollable, shameful, and irrational and damaging were consistently implicated in the indirect associations. This finding provided preliminary support for the metacognitive model mediating the association between childhood maltreatment and disordered eating but raised the question of whether different forms of metacognitive knowledge are implicated in these associations.
The metacognitive model was first introduced to the cognitive developmental literature by John Flavell (Flavell, 1979). Flavell conceptualised metacognition in terms of metacognitive knowledge, metacognitive experiences and metacognitive strategies. Metacognitive knowledge refers to an individual's knowledge/beliefs about one's own and others' cognitive abilities, strengths, limitations, other internal and external factors that may affect cognition, as well as knowledge about strategies and goals of thinking (Flavell, 1979). Metacognitive experiences refer to "any conscious cognitive or affective experiences that accompany and pertain to any intellectual enterprise" (Flavell, 1979, p. 906). These include feelings, judgements/estimates, as well as awareness of thoughts and ideas experienced during a task (Efklides, 2006). Metacognitive knowledge and metacognitive experiences guide the selection and implementation of metacognitive strategies aimed to control cognition to achieve one's goals.
In addition to beliefs about emotions, another common form of metacognitive knowledge are beliefs about cognitions. The role of metacognitive beliefs, or beliefs about emotions, in the development and maintenance of psychopathology has been promoted particularly by Adrian Wells (2002). Indeed, there is evidence that metacognitive beliefs are important in understanding the experiences of depression and anxiety (Russell et al., 2021;Strodl et al., 2015), alcohol misuse (Spada et al., 2013), post-traumatic stress disorder (Capobianco et al., 2020), addictive behaviours (Hamonniere & Varescon, 2018) and eating disorders (Vann et al., 2013). Wells (2002) suggests a core construct in understanding psychopathology is the activation of a dysfunctional thinking response referred to as the cognitive-attentional syndrome (CAS). The CAS is characterised by perseverative thinking patterns, attentional bias to threat, and the use of ineffective coping strategies, such as thought avoidance and suppression (Wells, 2002). The CAS is believed to be triggered, controlled, and maintained by positive and negative beliefs about thinking (Wells, 2002). Positive metacognitive beliefs include beliefs that cognitive strategies such as worry, rumination, and threat monitoring are helpful in achieving one's goals (Wells, 2002). Negative metacognitive beliefs include the beliefs about the negative effects of thinking patterns or strategies (e.g., 'worrying is dangerous for me'; Wells, 2002). The Metacognitions Questionnaire (MCQ-30;Wells & Cartwright-Hatton, 2004) measures five metacognitive beliefs and processes considered relevant to the vulnerability and maintenance of psychopathology. This measure assesses negative metacognitive beliefs (worry is uncontrollable and dangerous), positive beliefs about worry (benefits of worry to prevent distress and improve mood), belief in the need to control thoughts, lack of cognitive confidence, and cognitive self-consciousness which characterises the tendency to monitor thoughts. These beliefs perpetuate the CAS (Wells, 2002) which is implicated in the development and maintenance of psychopathology and maladaptive coping strategies (Wells & Carter, 2009).
Applied within the context of eating, the activation of the CAS may be identified by repetitive thoughts about food/eating/body image, focused attention on food/eating/body image, and maladaptive eating behaviours (Vann et al., 2013(Vann et al., , 2014. For example, negative evaluations about one's body shape/size might be perpetuated by positive metacognitive beliefs such as "thinking about my weight will protect me from putting on weight". Attentional bias towards cues indicating weight gain (e.g. the tightness of clothes) might be perpetuated by a metacognitive belief that "identifying early signs of weight gain, through tight clothes, will protect me from putting on weight". While the presence of these two metacognitive beliefs may be deemed functional to the individual, in terms of protecting the individual from gaining weight, they also increase the probability of becoming distressed through increasing the probability of true or false positive signals of weight gain and intensifying the perseverative negative evaluations of body size. This distress may then be strengthened by the activation of negative metacognitive beliefs such as "my worry about my body shame/weight is overwhelming and uncontrollable". As distress increases, metacognitive beliefs about appropriate coping strategies become activated. If such a metacognitive belief includes "eating helps me to calm down", then emotional eating may follow.
The association between metacognitive beliefs and eating behaviours has been well supported in populations with eating disorder diagnoses (Cooper et al., 2004;Georgantopolous et al., 2020;Olstad et al., 2015;Sapuppo et al., 2018;Spada et al., 2016;Vann et al., 2013Vann et al., , 2014. Collectively, these studies highlight that metacognitive beliefs are important psychological mechanisms in eating disorder pathology. However, less is known about the association between disordered eating and metacognitions, although emerging evidence suggests promising associations within nonclinical samples. For example, a small (N = 44) cross-sectional study by Quattropani et al. (2016) found significant associations between several eating disorder symptoms and metacognitive beliefs. Building on this research, Laghi et al. (2018) reported a positive correlation between metacognitive beliefs and the frequency of binge eating. While this sample was larger (n = 804) it was limited to an adolescent sample. Similarly, Limbers et al. (2021) reported higher levels of emotional eating behaviours to be associated with negative metacognitive beliefs in a nonclinical sample of adolescents. These studies were conducted among adolescents making the results difficult to generalise to adult populations.
The plausibility of metacognitive beliefs mediating the association between childhood trauma and adult disordered eating is supported by a study by Hosseini-Ramaghani et al. (2019) that found a direct correlation between childhood trauma and the strength of maladaptive metacognitive beliefs. Moreover, a review by Mansueto et al. (2019) found that exposure to traumatic childhood experiences such as abuse or neglect is related to the development of maladaptive metacognitive beliefs in later adulthood. The authors also reported that metacognitive beliefs mediated the relationship between childhood adversity and repetitive thinking/negative affect among both clinical and healthy adult samples (Mansueto et al., 2019). Similarly, Østefjells et al. (2017) showed that metacognitive beliefs about thoughts being uncontrollable and dangerous mediated the experiences of childhood emotional abuse and adult symptoms of depression and anxiety in individuals with psychosis or bipolar disorder. While the sample limits the generalisability of these findings to a nonclinical general population, the results are consistent with metacognitive theory which should also be applicable to non-clinical populations. To date, no study has examined whether metacognition beliefs mediate the relationship between childhood trauma and disordered eating in adults. Identifying such a mediating relationship may guide the development of novel interventions for disordered eating based upon a metacognitive model.

The present study
The above evidence suggests that childhood trauma is a risk factor for the development of disordered eating behaviours with maladaptive metacognitive beliefs being a potential psychological mechanism or mediator linking the two. Thus, the present study will examine the association of five forms of childhood trauma (emotional abuse, physical abuse, sexual abuse, emotional neglect, and physical neglect) with three disordered eating behaviours (cognitive restraint, uncontrolled eating, and emotional eating). This study has two hypotheses.
1. It is hypothesised that each form of childhood trauma (emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect) will independently be associated with each type of disordered eating (cognitive restraint, uncontrolled eating, emotional eating), after controlling for covariates and other forms of childhood trauma. 2. It is hypothesised that the relationship between childhood trauma (emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect) and each type of disordered eating (cognitive restraint, uncontrolled eating, and emotional eating) will be mediated by metacognitive beliefs (uncontrollable and dangerous, positive beliefs, need to control thoughts, cognitive self-consciousness, cognitive confidence).

Participants
The inclusion criteria for this study were being 18 years or older and residing in Australia. The exclusion criteria involved self-reporting having a current or previous eating disorder diagnosis (e.g. anorexia nervosa, bulimia nervosa or binge eating disorder), and self-selecting that they would become distressed thinking about their experience of childhood trauma, beliefs, or personal eating behaviour. Six hundred and thirteen individuals began the study. Among those who commenced the questionnaire, 19 did not meet the inclusion criteria and 133 exited before completion (the majority before completing demographics). Of those remaining, 9 responses had over 10% missing data and were subsequently excluded from the analyses. The final sample comprised of 461 participants with 65.7 percent females (n = 303). Participants ranged from 18 to 82 years of age (M = 39.6; SD = 19.3). The sample demographics are detailed in Table 1.

Measures
2.2.1. Demographic questions asked about age, gender, education, employment status, annual income, marital status, and whether they came from an English-speaking background.
2.2.2. COVID-19 Questions. Data collection occurred in the context of the COVID-19 pandemic from July 2021 to September 2021. In Australia, the pandemic disrupted aspects of daily life and restrictions were imposed such as physical distancing and lockdown measures, which may have impacted an individual's eating habits (O'Sullivan et al., 2020). Therefore, questions were included to gain insight into whether the pandemic altered participants eating behaviours. Approximately 65 percent (n = 298) of participants reported taking extra precautions to socially distance due to COVID-19 at the time of completing the questionnaire. Individuals were asked if their eating behaviour (cognitive restraint, uncontrolled eating, and emotional eating) was impacted by COVID-19, for frequency of behaviour and volume of food (see Supplementary Table A).
2.2.3. Childhood Trauma Questionnaire-Short Form (CTQ-SF). The CTQ-SF is a widely used 28-item self-report measure of traumatic childhood experiences for adults (Bernstein et al., 2003). The scale is a short-form version of the initial CTQ scale developed by Bernstein et al. (1994). Responses were measured on a five-point Likert scale ranging from never true (1) to very often true (5). Higher scores represent higher levels of childhood maltreatment (Bernstein & Fink, 1998). The scale assesses emotional abuse (CEA) e.g., 'I felt that someone in my family hated me'; physical abuse (CPA) e.g., 'I was punished with a belt, a board, a cord, or some other hard object'; sexual abuse (CSA) e.g., 'Someone tried to touch me in a sexual way, or tried to make me touch them'; emotional neglect (CEN); e.g., 'My family was a source of strength and support'; and physical neglect (CPA) e.g., 'I didn't have enough to eat'. The CTQ-SF has indicated good to excellent internal reliability across the five subscales among large samples (Hernandez et al., 2013;Kongerslev et al., 2019;Mizuki & Fujiwara, 2020). The scale has also demonstrated good construct validity and good retest reliability (Kim et al., 2013;Spinhoven et al., 2014). Similarly, the internal reliability for the present study ranged from good to excellent for the three eating behaviours (CEA α = 0.91, CPA α = .85, CSA α = 0.92, CEN α = 0.94, CPN α = .73).

Three-Factor
Eating Questionnaire (TFEQ-R21). The TFEQ-R21 is a 21-item instrument that measures three types of disordered eating behaviours: cognitive restraint (CR), uncontrolled eating (UE) and emotional eating (EE; Cappelleri et al., 2009). The TFEQ was originally developed by Stunkard and Messick (1985) and was later refined by Karlsson et al., (2000) to reflect the TFEQ-R21. Items 1-20 were measured on a four-point Likert scale ranging from definitely true (1) to definitely false (4) and item 21 was measured on an eight-point numerical scale. Higher scores represent higher levels of disordered eating. The cognitive restraint subscale assesses individual control over food consumption to influence body weight and body shape e.g., 'I don't eat some foods because they make me fat'. The uncontrolled eating subscale assesses the likelihood to lose control overeating when feeling hungry or exposed to external stimuli e.g., 'Sometimes when I start eating, I just can't seem to stop.' The emotional eating subscale measures tendency to overeat with reference to negative mood states e.g., 'I start to eat when I feel anxious'. The TFEQ-R21 supports measurement invariance across genders, has good internal consistency, and has previously demonstrated appropriate discriminative and convergent validity (de Medeiros et al., 2017;Duarte et al., 2020). Lin et al. (2021) reported good to excellent internal reliability. Similarly, the internal reliability for the present study ranged from good to excellent for the three eating behaviours (CR α = 0.79, UE α = 0.88, EE α = 0.94).

Metacognitions
Questionnaire (MCQ-30). The MCQ-30 is a measure of individual differences in metacognitive beliefs, judgements, and monitoring tendencies across five distinct metacognitive factors (Wells & Cartwright-Hatton, 2004). Responses to the 30-item, self-assessment questionnaire are measured on a four-point Likert scale ranging from do not agree (1) to agree very much (4). The scale assesses five thought patterns which are negative beliefs about uncontrollability and danger (UD) e.g., 'when I start worrying, I cannot stop', positive beliefs about worry (PB) e.g., 'worrying helps me cope', negative beliefs about the need to control thoughts (NCT) e.g. 'It is bad to think certain thoughts', cognitive self-consciousness (CS) e.g., 'I am constantly aware of my thinking', cognitive confidence (CC) e.g., 'I have a poor memory'. Higher subscale scores indicate a higher maladaptive metacognitive style (Wells & Cartwright-Hatton, 2004). The MCQ-30 has previously demonstrated good to excellent internal consistency and good test-retest reliability (Spada et al., 2016;Wells & Cartwright-Hatton, 2004). The internal reliability for the present study ranged from good to excellent

Procedure
The design was a cross-sectional survey. The project was approved by the Queensland University Technology (QUT) Human Research Ethics Committee (Approval Number: 2021100236). Participants were recruited through an Australia-wide social media campaign conducted by QUT social media department, the survey link was posted on the associate researcher's Facebook page and the flyer was uploaded to various online community groups. A snowballing and convenience sampling technique was employed by the associate researcher for further dissemination (Emerson, 2015). Additionally, psychology students at QUT were invited to join the study through university email lists and first-year psychology students were able to participate through the SONA course credit system. Community participants had the opportunity enter the random prize draw to win one of two $50 vouchers while QUT students were offered 0.5 course credit. The survey was administered online using Qualtrics (2021) platform.

Statistical analyses
All analyses were conducted using IBM Statistical Package for the Social Sciences (SPSS) version 27. Data was screened, responses with over 10% missing data were excluded, items were reverse coded where appropriate, and subscales were calculated using mean values. A missing values analysis revealed less than 1% missing data per item and Little's (1988) MCAR test indicated that data was missing completely at random (χ 2 (172) = 129.58, p = .99). Missing data was imputed using the expectation maximisation procedure in SPSS. Normality statistics and visual inspection of histograms revealed no departure from linearity and normality. However, minor clustering was observed on all scatterplots containing physical neglect and sexual abuse, suggesting potential breaches to homoscedasticity. A transform (square root, log10) was conducted, and transformations did not impact the overall significance of the model, therefore the untransformed data was retained for subsequent analyses. No influential outliers were present as Cook's distance scores were less than 1, indicating that the outliers would not impact the overall solution. No breaches to multicollinearity or independence of errors were present.
The independent variable was childhood trauma which consisted of the scores for each of the five subscales in the CTQ-SF. The proposed mediators were the five subscales of the MCQ-30. The dependent variables were the cognitive restraint, uncontrolled eating and emotional eating subscales derived from the TFEQ-R21.
Bivariate correlations between the independent variables, the proposed mediators and the three dependent variables were calculated to assist the interpretation of the multivariate analyses. Cohen's (1988) standard was used to evaluate effect sizes with correlation coefficients of 0.10 classified as small, 0.30 as medium, and >0.50 as large. Three hierarchical regression analyses were conducted to determine the effect of the independent variables (all forms of childhood trauma) on each dependent variable (cognitive restraint, uncontrolled eating, and emotional eating). Age, gender, level of education and income were covariates and socioeconomic status was operationalised using annual income and level of education. Gender was dummy coded into males (reference category) and females (entered into the model). Alpha was kept at p < .05 to allow easy comparison with our two previous similar studies (e.g. Dawson et al., 2022;Strodl & Wylie, 2020).
For the mediation hypotheses, each form of childhood trauma was independently assessed with the five metacognition variables, with one eating behaviour, while controlling for covariates. To ensure a robust interpretation, indirect effects were analyses in the present study through Andrew Hayes PROCESS Macro in SPSS through bias-corrected and accelerated (BC a ) bootstrapped confidence intervals (based on 5000 samples). Historically, it is argued that a causal relationship between the independent and dependent variable is necessary for mediation effects to be investigated (Baron & Kenny, 1986). Contrary to traditional methods, growing research has suggested that mediation can occur despite a non-significant causal relationship as this method is often associated with reduced power to detect significant effects (MacKinnon, 2008;Mackinnon et al., 2007;Rucker et al., 2011). Hayes (2018) suggests that significant total effects are not a precondition for investigating indirect effects. This methodology for testing mediation relationships has been highlighted by similar studies examining the relationship between childhood trauma and disordered eating (e.g., Dawson et al., 2022;Strodl & Wylie, 2020). Thus, in the present study, the indirect effects were analysed irrespective of whether significant total effects were detected in the hierarchical regression. Beta weights of .10 were categorised as small, 0.30 was considered medium, and >0.50 classed as large (Cohen, 1988).

Bivariate analyses
A negative association was found between age and uncontrolled eating (r = -0.29, p < .001) and emotional eating (r = − 0.15, p < .01). A small positive association was found between gender and cognitive restraint (r = 0.14, p < .05) and emotional eating (r = 0.05, p < .05), with females having larger scores on these variables than males. Income was significantly negatively correlated with uncontrolled eating (r = − 0.12, p < .05) and emotional eating (r = − 0.16, p < .001). No significant associations were found between education and the three disordered eating behaviours. Given many of the demographics had significant associations with the dependent variables and have been known to account for influences on the investigated eating behaviour in previous studies (e.g., Emery et al., 2021;Minnich et al., 2017;Strodl & Wylie, 2020), they were retained as covariates in the present study. Table 2 illustrates the means and standard deviations as well as the bivariate correlation matrix for the measured variables. Emotional abuse, physical abuse, emotional neglect, and physical neglect displayed significant, small to large associations with all MCQ-30 subscales. Sexual abuse evidenced small significant associations with all subscales across the MCQ-30 scales excluding cognitive self-consciousness.
Small to medium, positive, and significant correlations were found between metacognition scales uncontrollability and danger, positive beliefs, need to control thoughts, and cognitive self-consciousness and all three eating behaviours. However, cognitive confidence yielded only small, positive significant associations with uncontrolled and emotional eating behaviours. Additionally small, positive, and significant associations were identified between emotional abuse, physical abuse and physical neglect and uncontrolled eating. All forms of childhood trauma produced significant bivariate correlations with the emotional eating subscale with associations ranging from small to moderate. Table 3 summarises the hierarchical multiple regression analysis conducted to explore the relationship between the five forms of childhood trauma and the three forms of disordered eating behaviour, after controlling for age, gender, income, and level of education. In step one, gender (female) was significantly related to Cognitive Restraint, age was significantly and negatively related to Uncontrolled Eating, while age, gender and income were significantly related to Emotional Eating.

Hierarchical multiple regression
The addition of the five childhood trauma subscales in step 2 significantly contributed to the model explaining Uncontrolled Eating, ΔR 2 = 0.06, ΔF (5, 451) = 6.18, p < .001, and Emotional Eating ΔR 2 = 0.08, ΔF (5, 451) = 8.55, p < .001. However, the additive effect of these subscales did not significantly contribute to the model predicting Cognitive Restraint ΔR 2 = 0.01, ΔF (5, 451) = 1.27, p = .227.  Fig. 1 illustrates the significant direct and indirect pathways. The total indirect effect of the combination of mediator variables was positive and significant but small from all five forms of childhood trauma to Uncontrolled Eating. Through independent analysis of the mediator variables, a significant, positive pathway through the negative belief that thoughts are uncontrollable and dangerous to Uncontrolled Eating was evident from childhood emotional abuse, physical abuse, sexual abuse, emotional neglect, and physical neglect. Fig. 2 displays the significant pathways. The total indirect effect of the combination of mediator variables was positive and significant yet small from all five forms of childhood trauma to Emotional Eating. Independent analysis of the mediator variables revealed a significant pathway through the negative belief that thoughts are uncontrollable and dangerous to uncontrolled eating was evident from childhood emotional abuse, physical abuse, sexual abuse, emotional neglect, and physical neglect.

Discussion
The present study aimed to characterise the association between childhood trauma (i.e., childhood emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect) and three disordered eating behaviours (cognitive restraint, uncontrolled eating, and emotional eating). Additionally, it sought to examine the indirect role that metacognitive beliefs (Uncontrollable and Dangerous, Positive Beliefs, Need to Control Thoughts, Cognitive Self-Consciousness, Cognitive Confidence) play in the associations between the five measures of childhood maltreatment and three measures of disordered eating.

Hypothesis 1
The results partially support the hypotheses that each form of childhood trauma (emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect) would be independently associated with each type of disordered eating (cognitive restraint, uncontrolled eating, emotional eating), after controlling for covariates and other forms of childhood trauma. Examination of total effects revealed that only childhood emotional abuse significantly independently contributed to uncontrolled eating and emotional eating behaviours. This study did not find any support for direct and independent associations between childhood trauma experiences and cognitive restraint.
The significant association detected between childhood emotional abuse and uncontrolled and emotional eating aligns with existing research. The literature suggests that this form of childhood abuse is the most prevalent (Stoltenborgh et al., 2015) and may better predict disordered eating in adulthood above and beyond that of childhood physical or sexual abuse in non-clinical community samples (Burns et al., 2012;Kent et al., 1999). Research examining childhood emotional abuse and constructs similar to uncontrolled eating (i.e. binge eating and overeating) have similarly reported significant associations in clinical samples (Amianto et al., 2018;Caslini et al., 2016) and community samples (Feinson & Hornik-Lurie, 2016). The findings in this study support the results presented by Michopoulos et al. (2015) and Hymowitz et al. (2017) who found childhood emotional abuse to be the form of abuse that was most associated with emotional eating behaviours in non-clinical community samples. Furthermore, the association between emotional abuse and emotional eating as a coping mechanism has also been characterised in previous research with non-clinical community samples (Burns et al., 2012;Kennedy et al., 2007).
While our findings are in line with other research, a couple of notable differences were identified compared with the findings from similar previous research. Strodl and Wylie (2020) and Dawson et al. (2022) both identified an independent association between childhood sexual abuse and emotional eating. This is puzzling given that the present study adopted a similar methodology and questionnaires to that of Strodl and Wylie (2020) and Dawson et al. (2022). This difference may be in part due to the current sample consisting of a more even gender split, as studies have reported gender differences in the association between childhood trauma and disordered eating behaviours (Ackard et al., 2008;Fuemmeler et al., 2009). Other studies have found childhood sexual abuse in addition to emotional neglect, and physical neglect to be associated with disordered eating behaviours (Demirci, 2018;Emery et al., 2021). However, these samples consisted of adolescents and young adults rather than adults with a broad age range and did not control for sociodemographic factors as in the present study. Similarly, while the findings of no direct independent associations between any of the five forms of childhood trauma and cognitive restraint is similar to the findings by Dawson et al. (2022), it differs from the finding by Strodl and Wylie (2020) who found a direct association between childhood emotional abuse and cognitive restraint. The inconsistencies in the literature highlight that the relationship between different forms of childhood trauma and disordered eating is complex and requires further investigation to identify potential moderators.

Hypothesis 2
The findings partially support the hypothesis that the relationship between all forms of childhood trauma and each type of disordered eating would be mediated by all forms of metacognitive beliefs. The findings indicate that all forms of childhood trauma were related to uncontrolled eating and emotional eating behaviours through one metacognitive belief that thoughts are uncontrollable and dangerous. These findings are in line with existing research that has found negative metacognitive beliefs to play a role in the association between childhood trauma and psychological disorders (Hosseini-Ramaghani et al., 2019;Myers & Wells, 2015;Østefjells et al., 2017). The findings did not detect any support for the other metacognitive beliefs. To the authors' knowledge, no existing research has explored these three factors in a mediation model so direct comparison with other studies is not possible.
The findings provide some level of support for the original cognitive developmental metacognitive model (Flavell, 1979) and Well's (2002) metacognitive model of psychopathology within a disordered eating context. That is, metacognitive knowledge, in the form of beliefs about cognitions, appear to be important in guiding the selection of metacognitive strategies, in the form of maladaptive coping strategies such as uncontrolled and emotional eating behaviours. However, the findings differ from Well's (2002) metacognitive model, in that only negative beliefs that thoughts are uncontrollable and dangerous, and not positive beliefs about thinking strategies (such as worry or rumination), mediated the association between childhood trauma and uncontrolled or emotional eating. Comparatively, research examining the association between metacognitive beliefs and disordered eating report both positive and negative metacognitive beliefs to predict various disordered eating behaviours in both clinical (Aloi et al., 2021;Spada et al., 2016) and non-clinical samples (Laghi et al., 2018;Limbers et al., 2021). Further research is therefore needed to clarify whether positive beliefs about thinking strategies is an important component of a metacognitive model explaining the link between childhood trauma and disordered eating. In addition, it is also important to highlight that the same metacognitive belief (thoughts are uncontrollable and dangerous) independently mediated the associations between each of the five forms of childhood trauma and two of the types of disordered eating behaviours measured in this study. It therefore appears that the metacognitive model identified in this study may be generic across different forms of childhood abuse and different forms of reactive disordered eating behaviours. That is, any form of childhood trauma may facilitate the development of a belief that thoughts are uncontrollable and dangerous and that, in turn, this metacognitive belief facilitates the selection of uncontrolled or emotional eating behaviours as maladaptive coping strategies. This finding is consistent with a previous review of the literature indicating that multiple forms of childhood adversity are risk factors for the metacognitive belief that thoughts are uncontrollable and dangerous (Mansueto et al., 2019). These authors propose that while childhood positive metacognitive beliefs might initially activate the CAS (according to Wells' (2002) metacognitive model), continued poor emotion regulation and heightened distress over time might result in the strengthening of the metacognitive belief that thoughts are uncontrollable and dangerous. This is a plausible explanation and in further support of this explanation, there is also evidence that both positive and negative metacognitive beliefs are present in children (Ellis & Hudson, 2010) and that parenting factors may be related to genesis and perpetuation (Chow & Lo, 2017;Gallagher & Cartwright-Hatton, 2008). As such, further longitudinal research is required to clarify the mechanisms linking childhood maltreatment with changes in positive metacognitive beliefs and the strengthening of the belief that thoughts are uncontrollable and dangerous over time.

Implications
These findings have important practical and theoretical implications.
In terms of a practical implication, the study strengthens the literature showing the role of childhood traumatic experiences in disordered eating behaviours within a community sample. Such evidence emphasises the need to further develop effective preventative programs to reduce the incidence of childhood traumatic experiences to reduce the long-term detrimental consequences of such experiences. In terms of a theoretical implication, the present study also adds to the existing body of research by identifying an important metacognitive belief that appears to be an influential psychological mechanism in uncontrolled and emotional eating. Conversely, the findings of this study indicate that positive beliefs about thinking strategies, such as worry and rumination, may not be relevant to a metacognitive model of reactive disordered eating behaviours. It is important to note that the positive beliefs about thinking subscale of the Metacognitions Questionnaire 30 was developed based upon a metacognitive model of depression and anxiety. As such, it is possible that a questionnaire that measures more precisely positive beliefs about perseverative thinking about body image, eating behaviours or food may identify a mediating role of a metacognitive beliefs about the positive role of thinking processes more relevant to disordered eating. There is, therefore, a need to develop a questionnaire that may more precisely, and more comprehensively, assess the metacognitive model in relation to eating pathology.
The consistent finding within this study of the important role of the belief that thoughts are uncontrollable and dangerous supports the need to further test the potency of this metacognitive belief in longitudinal studies. In addition, the causal role of the belief that thoughts are uncontrollable and dangerous in uncontrolled and emotional eating, should be tested by interventional studies targeting this metacognitive belief, particularly in those who have experienced childhood trauma. Metacognitive therapy was developed by Wells (2002) to modify maladaptive metacognitive beliefs and has been shown to be a successful treatment measure among various psychological disorders (Normann & Morina, 2018;Philipp et al., 2019;Wells & Colbear, 2012). Metacognitive therapy has also been found to be a promising treatment for individuals with binge eating disorder (Robertson & Strodl, 2020). Future interventional studies may consider utilising metacognitive therapy to challenge the belief that thoughts are uncontrollable and dangerous community samples with experiencing high levels of uncontrolled and emotional eating and a history of childhood trauma. This is particularly important to test given that the highest probability of poor treatment outcomes with eating pathology has been observed in individuals who have experienced childhood trauma (Kong & Bernstein, 2009).

Limitations and future directions
Despite the contribution of the current study to the broader literature and practice, various limitations may affect the accuracy of the findings reported. Firstly, the CTQ-SF does not assess experiences of noninterpersonal trauma (e.g., natural disasters, car accidents, bullying, witnessing violence, sudden loss, crime, or related events). These adversities have also been identified as risk factors for disordered eating (Hecht & Hansen, 2001;Johnson et al., 2002). Thus, while the current study is valuable in investigating the impacts of multiple domains of trauma, it may not capture the influence of wider traumatic experiences, including experiences beyond childhood. Reporting bias and social desirability may be a limitation when interpreting the results of this study. While the Childhood Trauma Questionnaire has shown good temporal stability in adults (Goltermann et al., 2023), there is also evidence of respondents commonly minimising their past history of childhood maltreatment (MacDonald et al., 2016). Furthermore, there is good evidence of a low level of agreement between prospective and retrospective measures of childhood maltreatment (Baldwin et al., 2019).
Additionally, the data collection occurred in the context of the COVID-19 pandemic which may impact the generalisability of findings as the pandemic disrupted individuals' daily life (Brown et al., 2021;Simone et al., 2021). In response to uncertainty, individuals may be more likely to engage in disordered eating behaviours as an effort to cope with emotional states and/or gain control (Schlegl et al., 2020). The results from this study found that most participants' eating behaviours were approximately the same as prior to the pandemic, however, a small proportion of individuals reported engaging in an increase or decrease in the frequency and volume of eating behaviours (see Supplementary Table A). The present study attempted to capture how COVID-19 altered eating behaviours but did not include this assessment in the analyses due to the absence of reliable measures. To examine this further, it is suggested future research replicating the design of this study are required at a time when social distancing and no stay-at-home measures are imposed. In addition, given the exploratory nature of this study, replication of the relationships found between independent and dependent variables is required.
The present study utilised a community-based sample, thus findings may not be applicable to individuals with diagnosed eating disorders. Given the logistical challenges of screening the participants using a structured clinical interview with such a large sample size, we were reliant on the participants self-report to exclude participants with a diagnosed eating disorder. This may have resulted in some misclassification. In addition, given the nature of this sample, further research is needed to investigate whether similar associations are found in a clinical sample of adults diagnosed with an eating disorder. The overrepresentation of female participants and the absence of gender diverse participants in the current sample is another notable limitation. While controlled for as a covariate, this sample resulted in an inability to make gender-difference inferences. Therefore, the impact of gender on the found effects remains unknown due to the non-representative sample. Future research should seek to recruit a more representative sample, as it would be interesting to assess the gender effects. In addition, given the presence of minor cross-country variations in correlates with disordered eating behaviours such as restrained eating , there is a need to test cultural variations in the associations identified in this study.
This study was also limited to measuring metacognitive constructs as assessed by the MCQ-30. There is evidence that other important metacognitive processes, such as attentional bias as proposed in Wells' (2002) metacognitive model, are important in understanding the relationship between traits such as impulsivity and disordered eating behaviours (Hou et al., 2011). Future research should incorporate further measures of metacognitive knowledge, metacognitive strategies and metacognitive experiences to more fully test the applicability of the metacognitive model to disordered eating. Similarly, the CTQ-SF was limited in its measure of childhood maltreatment resulting in malnutrition as it is possible that there may be a unique link between this form of childhood maltreatment and disordered eating. While the CTQ-SF did include one item measuring this issue "Did not have enough to eat", the correlations with the three measures of adult disordered eating were weak and statistically non-significant (see Supplementary Table E). It is possible though that a more rigorous assessment of childhood malnutrition might be useful in future studies. Finally, the cross-sectional design of this study precludes confident interpretations about the direction of causality. Rather the findings of this study represent preliminary evidence that may provide direction and confidence to researchers to explore the direction of causality in more resource demanding longitudinal studies.

Conclusion
Disordered eating is common in society and the consequences of these behaviours threaten peoples physical and mental wellbeing. The present study contributes to the limited literature exploring the role of metacognitive beliefs in the association between childhood trauma and disordered eating behaviours. It was found childhood emotional abuse was associated with uncontrolled eating and emotional eating behaviours. Secondly, it was found that the metacognitive belief that thoughts are uncontrollable and dangerous mediated the association between all domains of childhood trauma and uncontrolled and emotional eating behaviours. None of the metacognitive beliefs measured in this study mediated the association between childhood trauma and cognitive restraint.
These findings provide preliminary support for the role of metacognitions as an underlying mechanism in the relationship between childhood trauma and uncontrolled eating and emotional eating, but not cognitive restraint. Together, the results help to inform future treatment planning protocols when working with both clinical and subclinical disordered eating behaviours. The early identification and treatment of these beliefs could also prevent the onset of more severe psychopathology during adulthood among clients who have experienced childhood trauma or engage in disordered eating behaviours.

Author contributions
SM and ES designed the study. SM oversaw the data collection. SM and ES performed the data analysis. SM was the primary author of the manuscript with ES writing sections of the manuscript as well as revising the manuscript. All authors have approved the final article.

Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data code and availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.

Ethical statement
This study involves human data and has been performed in accordance with the Declaration of Helsinki. The project was approved by the Queensland University Technology (QUT) Human Research Ethics Committee (Approval Number: 2021100236).

Declaration of competing interest
The authors declare no competsing interests.

Data availability
Data will be made available on request.