*4.1. Eating-Related Symptomatology*

Although not included in the DSM-5 diagnostic criteria [2], strong concerns about shape and weight are core psychopathologies that BED shares with other EDs like anorexia nervosa or bulimia nervosa [23,24]. Significant concerns of shape and weight and elevated measures of eating disturbances also emerge in connection with obesity in large community samples [25]. For example, Hilbert et al. [67] reported an increased risk of 11 to 20 times for obese individuals to show eating disorder psychopathology compared with individuals with normal weight. Consistent with these findings, OB-BED patients scored consistently higher on the EDEQ than OB participants, who scored higher than controls, with the subscale restraint as the only exception (for similar findings, see [68]). The LDA also suggests that the OB and OB-BED groups can be distinguished from controls along a continuum best described by elevated eating disturbances. Although more research on the relative importance of restraint appears warranted (cf. [69]), these findings suggest that shape and weight concerns in obese BED patients should be of special interest since they may be related to the condition's pathogenesis and can determine the therapy outcome [70].

Related to the question of the role of restraint, we assessed the PSRS. This short questionnaire yields the individuals' self-assessment of their own success in dieting, which has been shown to negatively correlate with BMI, rigid dietary control, food cravings, food addiction symptoms, and binge eating, but to correlate positively with flexible dietary control [59]. In our study, the OB-BED and OB groups reported much lower scores in dieting successfully than controls but remained comparable, which was also indicated by the LDA associating the PSRS with the continuum relevant to the distinction of healthy controls from OB and OB-BED groups. A possible explanation for the lack of distinction between OB-BED and OB groups according to restraint is that the questionnaire measures the attempt to lose weight, rather than actual restraint eating behavior, which more specifically relates to disordered eating [71,72].

However, OB-BED and OB groups were clearly distinguishable according to features of emotional eating, as suggested by the LDA and individual analyses. In the SEES, which was developed with the expectation that persons with lower eating pathologies tend to eat rather more when happy, and persons with higher eating pathologies more when having negative emotions [54]; the average score of eating under negative emotions was increased in OB-BED compared with both OB and controls. Controls, instead, reported eating more when feeling happy. The differences between OB-BED and OB were most pronounced for sadness as a low arousal emotion, which is consistent with similar findings for binge eating in patients with bulimia nervosa [73]. These findings may suggest that mechanisms of decreased food uptake while experiencing high arousal in the form of emotions are decoupled in BED with comorbid obesity but not in obesity without BED. This idea is further supported by similar patterns found for stress eating tendencies in the SSES. Again, the OB-BED group showed the highest score by far, while the OB group reported to eat only slightly more. Controls instead reported eating less under stress. These findings are consistent with laboratory studies showing an increased speed of food uptake after stress exposure in BED patients [74].

In line with the above-summarized findings, the DEBQ, which also measures emotional eating on one subscale, was found to distinguish between OB-BED and OB groups. The OB-BED group in our study showed the highest scores in emotional eating in the DEBQ, including the subscales for diffuse emotions [56]. It stands to reason that these differences hint at the role of emotional dysregulation in predicting binge eating behavior [75].

Further lines of distinction between OB-BED and OB groups emerged along with external eating and food craving tendencies. Consistent with previous findings [75–77], OB-BED patients scored higher than both OB and controls on the DEBQ external eating subscale, which measures the tendency to eat after being exposed to food cues. In a similar vein, OB-BED, OB, and controls were found to differ consistently across FCQT's food craving subscales. Although strong food cravings can be found in healthy individuals, too, it has been shown that those with binge eating symptoms score higher in food craving

assessments [37]. Consistent with our findings, this association is stronger for binge eating than for only obesity [35–37]. Innamorati et al. [78] even developed a potential cut-off score (157.5) of the FCQT for identifying clinical-level binge eating. Although our clinically diagnosed BED sample did not meet this criterion on average (142.3), findings from the LDA, which identified FCQT scores among the most important predictors for distinguishing between OB-BED and OB groups, generally support this contention.

As BED has been linked to high impulsivity and related conditions like substance use disorders [32,79], it has been suggested that BED might share features with food addiction [41,48]. Extending this line of research, we compared OB-BED, OB, and controls in terms of food addiction as measured by the YFAS 2.0. In our study, the highest prevalence of 76% for food addiction was found in the OB-BED group, followed by OB (40%) and controls (0%). For the OB-BED group, this is a lower prevalence than shown in previous studies [38,80], though it should be mentioned that food addiction prevalence rates are inhomogeneous across different samples [41]. Given the clear difference in food addiction prevalence between OB-BED and OB groups, it might be considered surprising to find that food addiction in the LDA was associated with the more general distinction between healthy and clinical groups, rather than between OB-BED and OB groups. It should be noted, however, that the LDA included scores for food addiction symptoms rather than scores for food addiction severity, rendering these findings only partly comparable. The decision to include food addiction symptoms rather than severity scores in the LDA was mainly due to statistical reasons, as the absence of variance in severity scores of controls (i.e., none of the controls were classified as food addicted) violated the LDAs assumptions.
