Metacognition and emotion regulation as treatment targets in binge eating disorder: a network analysis study

Background This study aims to examine the underlying associations between eating, affective and metacognitive symptoms in patients with binge eating disorder (BED) through network analysis (NA) in order to identify key variables that may be considered the target for psychotherapeutic interventions. Methods A total of 155 patients with BED completed measures of eating psychopathology, affective symptoms, emotion regulation and metacognition. A cross-sectional network was inferred by means of Gaussian Markov random field estimation using graphical LASSO and the extended Bayesian information criterion (EBIC-LASSO), and central symptoms of BED were identified by means of the strength centrality index. Results Impaired self-monitoring metacognition and difficulties in impulse control emerged as the symptoms with the highest centrality. Conversely, eating and affective features were less central. The centrality stability coefficient of strength was above the recommended cut-off, thus indicating the stability of the network. Conclusions According to the present NA findings, impaired self-monitoring metacognition and difficulties in impulse control are the central nodes in the psychopathological network of BED whereas eating symptoms appear marginal. If further studies with larger samples replicate these results, metacognition and impulse control could represent new targets of psychotherapeutic interventions in the treatment of BED. In light of this, metacognitive interpersonal therapy could be a promising aid in clinical practice to develop an effective treatment for BED. Supplementary Information The online version contains supplementary material available at 10.1186/s40337-021-00376-x.


Background
Binge eating disorder (BED) is characterized by recurrent episodes of binge eating with a sense of loss of control over eating, accompanied by negative feelings [1]. To date, the guidelines recommend cognitive behavioural therapy (CBT) as the first-line treatment option for BED [2,3]. Although CBT is quite effective in patients with BED, about 50% do not fully respond to treatment [4][5][6]. A possible explanation could be that only a small portion of patients with BED report the overvaluation of body shape and weight that forms the core of the CBT protocol [7]. Other treatments such as dialectical behaviour therapy [8,9] and interpersonal psychotherapy [10,11] have shown promising results but failed to bridge the efficacy gap in treating BED. In other words, the available data do not favour one treatment over the other.
New therapeutic approaches able to target the key elements of the complex psychopathology of BED are therefore a priority. Investigating the specific weight of each psychopathological dimension could help in developing more tailored psychological interventions for BED.
Network analysis (NA) emerged as a novel approach to conceptualize mental disorders [12]. According to this approach, symptoms of psychiatric disorders are distinct entities that can influence, maintain and/or interact with other symptoms [13]. Mental disorders can be characterized as complex systems in which symptoms are represented as distinct nodes, connected by edges that represent the strength (e.g. strong/weak correlations) and direction (e.g. positive/negative correlations) between pairs of symptoms. NA allows the identification of the central symptoms (i.e. when a node has many strong associations with other nodes and strong correlations with other nodes within the network) [14].
Development of the NA approach over the past decade has provided a theoretical framework that was adopted to identify the central symptoms of different psychiatric disorders, such as bipolar disorder [15], depression [16], obsessive compulsive disorder [17] and schizophrenia [18]. More recently, researchers in the field of eating disorders have applied NA to examine the symptoms of anorexia nervosa [19][20][21][22] and bulimia nervosa [23][24][25].
To date, only three studies [26][27][28] dealing with BED have used the NA approach. In the first investigation, overvaluation of shape and weight emerged as central symptoms of BED whereas behavioural symptoms (i.e. binge eating, restriction, secret eating) were less central [26]. The study by Solmi et al. revealed affective symptoms, interoceptive awareness, ineffectiveness, interpersonal functioning and drive for thinness as the central variables among patients with BED [27]. Finally, the third study showed that CBT provides high integration and connectivity of the psychopathology network in BED, suggesting an improved patient understanding of associations between binge eating and other symptoms [28].
However, no research has used NA to investigate the complex connections between the eating (i.e. binge eating and eating psychopathology), affective (i.e. anxiety and mood) and psychological (i.e. metacognition and emotion regulation) features of patients with BED.
Prior research evidenced a significant relationship among negative affect, difficulties with emotion regulation and binge eating symptoms [29][30][31][32][33]. For example, binge eating can be the result of a dysfunctional strategy to avoid interpersonal difficulties and negative emotions [34], especially in individuals who experience difficulties with regulating their emotional state [32]. However, the role of metacognition in BED has received less research attention. In the current study, we refer to metacognition as a psychological function that plays a key role in identifying mental states and ascribing them to oneself and others, reflecting and reasoning on mental states and, finally, using this information to manage interpersonal conflicts [35]. According to this model, metacognition is made up of different sub-functions that interact with each other and can be singularly impaired [35]. A previous study suggested that the severity of BED can worsen in relation to the impaired self-monitoring metacognition through the mediation of emotion dysregulation [36].
In the present study, we sought to extend the research on the clinical characteristics of BED by applying an NA model to provide an examination of the pathways that underlie eating symptoms and their relations to metacognition, emotion regulation and distress. These NA results may lead to more nuanced insights regarding the core targets for psychotherapeutic interventions. Given the explorative nature of our study, no a priori hypotheses were formulated.

Procedure
We performed a consecutive sampling of male and female patients attending the Outpatient Unit for Clinical Research and Treatment of Eating Disorders in Catanzaro (Italy). Patients were invited to participate in the present study if they met the following criteria: age 18-65 years; current diagnosis of BED according to the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) criteria; absence of current Axis I comorbid psychiatric disorders; and capable of answering self-report questionnaires and expressing valid consent.
Participants were deemed ineligible if they had: IQ < 70 [37]; drug dependence and/or abuse; severe mental illness that could interfere with clinical assessment (i.e. psychosis); history of chronic medical illness (i.e. chronic cardiovascular diseases) or neurological conditions (i.e. dementia) affecting cognitive functioning; other severe medical comorbidities (i.e. epilepsy); medical conditions that influenced eating/weight (i.e. diagnosis of diabetes mellitus); or a history of malignant disease.
Trained psychiatrists interviewed all participants using the Structured Clinical Interview for DSM-5 Disorders -Research Version [38] for diagnostic purposes and collected sociodemographic and clinical data. Researchers informed participants about the aims, procedures, anonymity and voluntary participation in this research. Participants gave their written informed consent to participate in accordance with the latest version of the Declaration of Helsinki [39] and the local ethical committee.

Measures
The Eating Disorders Inventory-2 (EDI-2) [40,41] is a self-report questionnaire made up of 91 items that evaluates the psychopathology and symptomatology of eating disorders. The EDI-2 provides 11 subscale scores and a global measure of eating disorder severity obtained from the sum of all the items (ranging from 0 to 273). Higher scores indicate more severe symptoms. Cronbach's alpha for the total score in this study was good (.840).
The Binge Eating Scale (BES) [42] measures the severity of BED. It consists of 16 items that describe the behaviours, feelings and cognitions associated with binge eating. Total BES scores of < 17, 17-27 and > 27 indicate improbable, possible and probable BED, respectively. The internal consistency in this study was .880. The Metacognition Self-Assessment Scale (MSAS) [43] is an 18-item five-point Likert-type (1 = never, 5 = almost always) self-report questionnaire that evaluates metacognitive functioning. The raw score ranges from 18 to 90 and lower scores indicate impaired self-evaluation of metacognitive function. Specifically, the MSAS measures four abilities of metacognition: monitoring, differentiation/decentration, integration and mastery. In this study, Cronbach's alpha ranges from .820 to .840.
The Difficulties in Emotion Regulation Scale (DERS) [44] is a 36-item five-point Likert-type scale that assesses emotion dysregulation across six subscales: nonacceptance of emotions; difficulties in pursuing goals when having strong emotions; difficulties in controlling impulsive behaviours when experiencing negative emotions; lack of emotional awareness; limited access to emotion regulation strategies; and lack of emotional clarity. Higher scores indicate more problems in emotion regulation. In the current study, the internal consistency ranges from .870 to .895.
The State-Trait Anxiety Inventory (STAI) consists of 20 items that assess state anxiety (STAI-St) and 20 items that measure trait anxiety (STAI-Tr) [46]. The present study only included the STAI-Tr for statistical purposes. Cronbach's alpha was .795.

Network estimation and accuracy
NA was performed using the R (version 3.6.2) qgraph and bootnet packages in accordance with Epskamp and colleagues [47].
The network has been inferred by means of Gaussian Markov random field estimation, applying 'Least Absolute Shrinkage and Selection Operator'(LASSO) regularization to limit the number of spurious associations [48]. Moreover, the extended Bayesian information criterion (EBIC) [49], a tuning parameter that sets the degree of regularization/penalty applied to sparse correlations, was set to 0.20 in the current study (values between 0 and 0.5 are typically chosen). Network estimation was performed using the estimateNetwork routine of the bootnet package [50].
The centrality of a node is used to infer its influence, or structural importance, in the network. Three main indices estimate the centrality: betweenness (how a node influences the average path between other pairs of nodes); closeness (how a node is indirectly connected to the other nodes); and strength (how a node is directly connected to the other nodes). The centrality Plot function in qgraph was used to calculate indices of centrality.
According to the recommendations of Epskamp et al. [51], in order to assess the internal reliability of the network we calculated the correlation stability (CS) coefficient, which is the maximum proportion of the population that can be dropped so that the correlation between the re-calculated indices of the obtained networks and those of the original network is at least 0.7. It is recommended that the minimum cut-off to consider a network stable is 0.25 for betweenness, closeness and strength [51]. The CS coefficient was computed using case-drop bootstrapping (nboots = 2000). Then we estimated the accuracy of edge-weights by drawing bootstrapped confidence intervals calculated using nonparametric bootstrapping (nboots = 2000). Both for casedrop and non-parametric bootstrapping, network stability analyses were performed using the bootnet function in the bootnet package.
Visual inspection of the network reveals that thicker edges indicate stronger associations between symptoms, with positive associations typically illustrated in blue and negative associations typically represented in red.

Sample characteristics
In total, 155 BED patients (86.5% females) aged 41.2 ± 13.2 years and with body mass index 37.9 ± 10.4 kg/m 2 , took part in the current study. Table 1 displays the clinical characteristics of the sample. Figure 1 illustrates the network of BED symptoms. Nodes belonging to each domain (i.e. eating symptoms, emotion dysregulation and metacognition) are generally associated and close to each other. There is a strong negative connection between self-monitoring and DERS-Clarity, and a strong positive connection among selfmonitoring, differentiation and mastery. The associations between BED symptoms and depression, and between EDI-2 total score, depression and anxiety, are moderately strong. The psychopathological variables (BES, EDI-2 total score, STAI-Tr and BDI) and emotion regulation (DERS) are moderately connected. The BED symptom node (BES) has a direct connection with nonacceptance of emotions, whereas the depression node (BDI) connects both with difficulties in controlling impulsive behaviour and lack of emotional clarity. Figure 2 displays the strength centrality index of the variables included in the network. The CS coefficient is 0.301 for strength, which is above the recommended cut-off value (i.e. 0.25); however, the CS coefficients for betweenness and closeness are below 0.25. Therefore, we decided to choose the strength index as the main CS coefficient. This choice is not surprising because the interpretation of betweenness and closeness in networks is somewhat unclear [52] and the strength index is considered a more stable centrality index than betweenness and closeness [53]. Furthermore, because we aimed to understand the core symptoms to target using psychological treatment, we relied on the strength index because it exactly performs this function. Additional file 1 (Fig. S1) shows the accuracy of the CS indices.

Discussion
This is the first study to investigate the associations between eating (i.e. binge eating and eating psychopathology), affective (i.e. anxiety and depression) and psychological features (i.e. metacognition and emotion regulation) through the NA method among patients with BED. Our results showed that impaired self-monitoring metacognition and difficulties in impulse control were the nodes with the highest centrality strength and thus the nodes most directly connected to the other nodes in the network [53]. According to the NA approach, activation of a node may cause the development of the connected symptoms; therefore, the most central nodes have been conceptualized as core symptoms [54]. Our findings suggest that impaired selfmonitoring metacognition and difficulties in impulse control may be important clinical characteristics among patients with BED. Although the high centrality of a node may be the effect of connections with other symptoms [55] and a cross-sectional study cannot show causal associations, the metacognitive and emotion regulation dysfunctions may represent potential targets for treatment, therefore these outcome variables of BED warrant further research.
This finding is in line with our previous study where low self-monitoring led BED-obese patients to express the worsening of binge severity through emotion dysregulation [36]. Consistent with this hypothesis, other researchers found that difficulties in emotion recognition could play a key role in the development and maintenance of BED [56,57].
Another important finding of the current NA was the strong correlation of the self-monitoring node with mastery strategies. According to metacognitive theory, a high level of self-monitoring allows the use of functional mastery strategies. In more detail, mastery is 'the ability to work through one's representations and mental states, with a view to implementing effective action strategies, in order to accomplish cognitive tasks or cope with problematic mental states' [35,58]. Thus, it could be inferred that enhancing metacognitive abilities leads to reduced dysfunctional strategies among patients with BED, who usually manage intense emotions with binges [8,59]. It is worth noting that both dysfunctional eating (i.e. BES and EDI-2 total scores) and affective symptoms (i.e. BDI and STAI-Tr) were peripheral to the network structure of patients with BED, indicating that they had weaker connections to the rest of the network compared with other nodes. Regarding eating psychopathology, in the current study the lowest strength was found for the BES (M = − 1.39) and the EDI-2 total score (M = − 1.22). Notwithstanding the weak centrality of eating symptoms in the network structure, our findings suggest that the BES score is connected to non-acceptance of emotions, whereas the EDI-2 total score is connected to poor metacognitive ability to distance from one's own thoughts and evaluate them critically. Overall, our results confirm recent literature data on NA in BED (that binge eating was not central to the psychopathology) [26,28] but contrast with the typical approach to diagnosing BED (relying upon the presence of binge eating behaviours).
Consistent with the present findings, we could argue that the clinical constructs such as impaired selfmonitoring, difficulties in impulse control and lack of emotional clarity could be the vulnerability factors of BED whereas the pathological eating behaviour (i.e. binge eating) itself seems to be the consequent behaviour. This observation is in line with recent literature that investigated predisposing and precipitating factors in BED [32,60,61].
Furthermore, depressive and anxious symptoms were not central nodes in our network model whereas they had high centrality in Solmi and colleagues' model [27]. This discrepancy could be due to the use of different psychometric instruments. Solmi and colleagues used the Symptom Checklist-90 (SCL-90), which is not so specific and only takes into consideration the prior week; instead, the BDI-II and the STAI-Tr are more specific for diagnostic purposes and consider a longer temporal range of assessment (i.e. 2 weeks for BDI following the DSM-5 temporal criterion for major depressive episode; 'usually feeling' for STAI-Tr). Therefore, their study could have overestimated the weight of anxious and depressive symptoms in BED.
The present results should be read in light of some limitations. First, the sample size is smaller than in other studies that used NA in BED. Nevertheless, according to the recommendations of Levinson and colleagues [62] on the use of NA in the field of eating disorder ('to date, the best recommendation is to use the largest sample size possible and make sure that your network is stable'), our model was demonstrated to be stable. Second, it was not possible to evaluate the differences in NA according to gender; however, a recent NA study among patients with eating disorders showed more similarities than differences between men and women [63]. Finally, the crosssectional design does not allow the investigation of causality in the associations between dimensions, therefore future longitudinal research could explore whether psychotherapeutic interventions that target metacognitive and impulsive dimensions may be more effective in treating BED.

Conclusions
The current study suggests a link between reduced ability to identify and describe mental states and the lack of emotion awareness and clarity among patients with BED. Moreover, according to the present NA findings, impaired self-monitoring metacognition and difficulties in impulse control are the central nodes in the psychopathological network of BED, whereas eating symptoms seem to be marginal.
These results could lead to a change in the current conceptualization of BED and the consideration of new targets of psychotherapeutic interventions, if confirmed in larger samples. Also, approaches focused on the improvement of metacognitive dysfunctions could be considered. With this aim, metacognitive interpersonal therapy [64] could be a promising aid in clinical practice to develop an effective treatment for BED.
Additional file 1: Figure S1. Results of case-dropping subset bootstrap procedure to assess stability of network centrality indices. Average correlations between centrality indices of networks sampled with persons dropped and the original sample.