Background

Eating disorder (ED) psychopathology is becoming more prevalent in non-Western societies, such as Iran [1,2,3,4]. For example, Iranian college samples have been shown to adopt Western ED-related body comparisons and thin-ideal internalization [5,6,7] and engage in ED behaviors with similar frequency to college women in the United States (US) [7]. Therefore, it is important to examine factors that may initiate ED psychopathology in Iran. One of the most prominent risk factors and contributors to the development of ED psychopathology is negative affect [8, 9]. Specifically, negative affect has shown to play an important role in initiating and maintaining ED psychopathology [10,11,12,13,14]. In Iran, negative affect as well as related variables (e.g., poor emotion regulation) are also emerging as important potential vulnerability factors for ED psychopathology [15, 16].

Most research on links between ED psychopathology and negative affect, in both Iran and the West, has been conceptualized from the perspective that psychopathology symptoms result from a latent common variable (e.g., negative affect) [17]. Because of this, little is known about which individual negative affective states are most relevant to specific aspects of ED psychopathology. Whereas some studies have examined specific negative affective states (e.g., guilt, shame, fear, hostility, sadness) that may be relevant to ED psychopathology in both clinical and non-clinical samples [18,19,20,21,22,23,24,25,26,27,28,29], these studies have examined each affective states separately and have not incorporated negative affective and ED psychopathology in one integrated model.

Network theory is an alternative to traditional approaches that can increase understanding of which specific aspects of negative affect (e.g., shame vs. hostility) are most relevant to specific ED symptoms (e.g., fear of weight gain, binge eating, restraint, etc.). Network theory conceptualizes psychopathology such as EDs as interacting symptoms that activate and maintain one another [30, 31]. Network analysis can be used to elucidate connections between various vulnerability factors (e.g., negative affect) and ED psychopathology. Specifically, in network models, symptoms from one cluster (e.g., negative affect) may activate or weaken symptoms in another cluster (e.g., ED psychopathology) and vice versa via ‘bridge’ symptoms within the network. Bridge symptoms can elucidate which specific individual negative affective states (e.g., guilt vs. shame) may be linked with specific ED symptoms (e.g., binge eating vs. restricting). Network theory suggests that disrupting bridge pathways may weaken the relation between negative affect and ED psychopathology [32]. There are currently several network studies that have examined bridge symptoms between various psychological constructs (e.g., anxiety, depression, fear) and ED symptoms [33,34,35,36,37,38,39,40]. Network theory also enables for the identification of core symptoms, which are symptoms with the highest centrality that are posited to have the most overall relationships and therefore impact most other symptoms in the network. Core symptoms are suggested to be potential intervention targets (e.g., in prevention programs) because disruption of a core symptoms would weaken the overall network. Several studies [33, 39, 41,41,42,43,44,45,46,47,48,49] and a systematic review [50] examining core symptoms in EDs have been conducted, and concluded that overvaluation of weight or shape and desire to lose weight are core symptoms of EDs.

Although researchers have evaluated networks of emotional distress (e.g., depression, anxiety, fear, shame; [33,34,35,36,37,38,39,40] and EDs, to the best of our knowledge, there is only one study to date examining a network of negative affect and ED symptoms. Wong and colleagues [51] conducted a network analysis of negative affect, positive affect, and ED symptoms in a sample of US patients with EDs, and found that guilt about eating and shame were central bridge symptoms and that guilt about eating was the most central symptom [51]. However, no study to date has examined a negative affect and ED symptom network in a non-Western (e.g., Iranian) sample, and it is unclear if findings would translate in this population. There are various cultural differences in Iran that may result in differences in bridge symptoms between negative affect and ED symptoms in Iran vs. the West. For example, Iranian culture is unique because, since the 1979 Iranian revolution, modest dress (i.e., wearing Islamic-head and body coverings such as a hijab and a manteau) has been legally mandated for women and mainstream Western media. On the other hand, women in Iran do not use hijab in private places (i.e., home). This exposure may increase body dissatisfaction and/or negative affect (e.g., shame, nervousness) when seeing one’s body at home. For example, a recent study indicated that body-related shame was associated with ED psychopathology in Iran [52]. Of note, bridge symptoms between other psychological constructs (e.g., depression, anxiety) and ED symptoms have differed among individuals in Iran [4, 53] compared to those from other countries (e.g.. US; Australia; [38, 54]. Finally, the sample used by Wong and colleagues [51] mostly consisted of women, and it is unclear if findings would translate to a sample of both men and women.

Furthermore, only three studies to date have examined ED networks in Iranian individuals [4, 53, 55]. Sahlan and colleagues [4] conducted a network analysis of depression, self-esteem, and EDs in Iranian adolescents and college students, and found that desire to lose weight and discomfort when seeing one’s own body were the most central symptoms of EDs. Sahlan and colleagues [53] conducted a network analysis of social anxiety and EDs in Iranian preadolescents, and found that discomfort eating sweets was the most central symptom. Sahlan and Sala [55] conducted a network analysis of resilience and EDs in Iranian college students, and found that discomfort in seeing one’s own body, feeling guilty about eating due to shape/weight, and thinking about shape and weight making it difficult to concentrate were the most central symptoms. Of note, at least some core symptoms have differed from US, Italian and Australian samples [38, 39, 41,42,43,44,45, 47, 56]. For example, whereas overvaluation of weight or shape has consistently emerged as a core symptom in research conducted in the US [33, 41, 43, 46,47,48,49], it did not in one study conducted in Iran [4]. In contrast, a desire to lose weight has emerged as one of the most central symptoms in several cultures, including the US [33, 41, 43, 46,47,48,49], Italy [47], Australia [38], and in one study in Iran [4].

Current study

The current study used network analysis to test links between specific negative affective states and ED psychopathology among Iranian college students. Our primary aim was to identify bridge symptoms within a negative affect and ED symptom network. We also aimed to identify central (i.e., most densely connected) symptoms as a secondary aim. Based on previous studies in US [51, 57] and Iran [4], we hypothesized that shame would bridge negative affect and ED symptoms, and that guilt about eating and desire to lose weight would be the most central symptoms in the network.

Method

Participants

Data from this study (n = 637; 60.3% women) were also used in Sahlan and colleagues’ study [4]. However, the specific aims and primary analyses of the current project are unique and have not been previously published. Specifically, bridge symptoms between negative affective states and ED symptoms have not been investigated previously. College students were recruited from two cities with diverse geographic zones and ethnicities (i.e., Tabriz [North-Western, Turk], Shiraz [South-Central, Persian]). Participants’ age ranged from 18 to 54 years (M = 21.89, SD = 3.62) and self-reported BMI ranged from 15.57 to 39.18 kg/m2 (M = 22.19, SD = 3.50). In the current study, 19.40% of participants reported clinical levels of ED (i.e., ≥ 2.5; Eating Disorder Examination-Questionnaire [EDE-Q], a global score; [58]. Additionally, between 1.80 and 26.70% of participants reported recurrent (i.e., ≥ 4 times during the past 28 days) binge eating, or purging (i.e., self-induced vomiting, laxative misuse, and over-exercise) (See Table 1 for more information).

Table 1 Means and standard deviations for eating disorder psychopathology and negative affect among college students (N = 637)
Full size table

Procedure

All the potential participants from a broad range of departments (e.g., Psychology, Sociology, Agriculture, Tourism Management) were approached on campus or during class and were invited to participate in a study to test psychological issues among college students. Interested students completed paper–pencil questionnaires. All questionnaires were anonymous with no identifying information to protect the confidentiality of college participants. The study was approved by the Ethical Board of Iran University of Medical Sciences and all participants provided informed consent.

Measures

Demographic information

Participants completed questions regarding age, gender, as well as height and weight to calculate body mass index (BMI).

Eating disorder symptoms

We used a Farsi version of the EDE-Q (F-EDE-Q; [1, 59]) that assesses ED symptoms over the past 28 days. The validity and reliability of the F-EDE-Q has been supported in Iran [1]. Twenty-two items are rated on a seven-point scale ranging from 0 (No days) to 6 (Every day). Five items assess the frequency of ED behaviors. Higher scores indicate a greater level of pathology. Cronbach’s α was 0.92 in this sample.

Negative affective states

We used a Farsi version of the 10-item Negative Affect scale [60] that assesses negative affect. The validity and reliability of the F-PANAS has been supported in Iran [60]. Responses are rated on a 5-point Likert scale from 1 (Very slightly or not at all) to 5 (Extremely). Higher scores indicate higher trait level negative affective states. Cronbach’s α was 0.86 in this sample. Of note, we also assessed positive affect, but did not include it in the model due to low stability when including positive affect.

Data analyses

Missing data

One participant (0.4%) did not provide weight and height to calculate BMI. There were no missing data on any of the scales administered.

Item selection

We used the goldbricker function in the R package networktools to select the final items to include in the network [61]. Goldbricker compares every possible combination of correlations in the network and suggests nodes that may be redundant (i.e., measuring the same construct). Using goldbricker avoids inflating centrality by eliminating items that may measure the same construct [62]. We then used best_goldbricker to suggest which of the redundant nodes to remove. After applying goldbricker, six eating disorder items (i.e., food avoidance, preoccupation with eating, fear of weight gain, feelings of fatness, importance of shape, dissatisfaction with shape) and three negative affect items (i.e., distressed, guilty, scared) were removed. The final network had 29 nodes (22 eating disorder nodes, 7 negative affect nodes).

Glasso networks

We used the GLASSO estimator in the bootnet package in R to estimate networks. GLASSO networks estimate edges that are likely to be spurious as zero, thus resulting in a more accurate network [63]. Of note, correlations among nodes in GLASSO networks represent partial correlations in the network. We used Spearman correlations because the initial network estimated using polychoric correlations was densely connected [63]. Furthermore, Spearman correlations produce more stable networks [63].

Network stability

We used the bootnet package to estimate network stability [64]. We computed the following stability parameters: (1) an EI stability correlation coefficient (EI-coefficient); (2) an edge stability correlation coefficient (ES-coefficient), and (3) bridge EI stability correlation coefficient (BEI-coefficient). Coefficients between 0.20 and 0.50 are considered acceptable, coefficients above 0.50 and below 0.70 are considered good, and coefficients above 0.70 are considered excellent [86, 87].

Limitations and future directions

The study has several strengths. To our knowledge, this is the first network study examining symptoms that may bridge negative affect and ED symptoms in Iran. However, it is worth mentioning some limitations. First, several ED psychopathology symptoms are affective in nature (e.g., fear of weight gain, fear of losing control over eating, feeling guilty about eating), which may have influenced findings. Specifically, the extent to which findings relate to generic affective states vs. ED-related affective states is unclear. Second, similar to most network studies, this study was cross-sectional. Although other studies have shown that network structures among cross-sectional and longitudinal studies do not differ [39, 44, 56], to fully conceptualize how ED symptoms and negative affect symptoms dynamically relate to one another, longitudinal network analyses are needed [56]. Additionally, the results of the current study are limited by the measures used. For example, previous studies reported that risk factors such as muscular-athletic-ideal internalization also contribute to ED psychopathology among college men and women in Iran [5,6,7]. However, the EDE-Q did not assess muscularity/exercise-oriented attitudes and behaviors [88]. Therefore, future research should examine a network of ED symptoms using measures that include muscularity and exercise constructs (i.e., Farsi Eating Pathology Symptoms Inventory ([F-EPSI]; [79]). Relatedly, we did not include positive affect in the model. Lastly, another limitation is the use of a non-clinical sample with a narrow range of F-EDE-Q scores (i.e., 0.16–3.23). However, other studies have found that network structures (e.g., centrality of specific ED symptoms, edges) among non-clinical and clinical ED samples are similar [46, 47]. Furthermore, examining networks in non-clinical samples may have important implications for prevention research.

Conclusion

Our findings suggest that negative affective states of hostility and shame may increase vulnerability to ED psychopathology in Iran. These findings are somewhat consistent with findings in US samples, where shame bridged negative affect and ED psychopathology. Unique to our sample, hostility also emerged as a central bridge symptom. Findings have important implications for ED prevention programs in Iran that should be examined in future research.