سابقه خودآسیبی مقاومت به درمان در بیماران بستری از شکل گریزی بدن در زنان مبتلا به اختلالات غذا خوردن را پیش بینی می کند: نقش عاطفه منفی

Abstract Although self-harm has been observed among patients with eating disorders, the effects of such tendencies on treatment outcomes are unclear. The current study employed structural equation modeling to (a) evaluate the relationship between self-harm and changes in body dissatisfaction and drive for thinness in a large sample of patients (n = 2061) who underwent inpatient treatment, and (b) to examine whether the relationship between self-harm and changes in body dissatisfaction and drive for thinness during inpatient treatment remains significant when controlling for change in negative affect during treatment. Results revealed that patients with a history of self-harm reported significantly less reduction in body dissatisfaction and drive for thinness following treatment. Patients experiencing less change in negative affect also reported significantly less reduction in body dissatisfaction and drive for thinness after discharge from treatment. However, the association between history of self-harm and reduction in body dissatisfaction and drive for thinness after treatment became non-significant when controlling for change in negative affect. This pattern of findings was also replicated among patients with a primary diagnosis of anorexia nervosa (n = 845), bulimia nervosa (n = 565), and eating disorder not otherwise specified (n = 651). The implications of these findings for delineating the specific role of self-harm in the nature and treatment of eating disorders are discussed.

1. Introduction Eating disorders such as anorexia nervosa (AN) and bulimia nervosa (BN) are characterized by disturbed patterns of eating behavior, extreme weight regulation, preoccupation with body weight and shape, and a distorted body image (American Psychiatric Association, 2013). Although eating disorders may not be highly prevalent in the general population (e.g., 0.5% and 1.0% for AN and BN, respectively; 0.9% and 1.5% among women; Hudson et al., 2007), reported rates may be due to current stringent criteria (e.g., frequency of binge episodes, Wilson and Sysko, 2009). In fact, most individuals with pathological eating behaviors fall in the eating disorder not otherwise specified (EDNOS) category (Fairburn and Bohn, 2005), suggesting needed changes to the existing diagnostic criteria. Eating disorders are associated with severe health detriments including renal and gastrointestinal complications, electrolyte imbalance, and cardiovascular problems among other issues (Mehler et al., 2004). Further, approximately 4.0%, 3.9%, and 5.2% of cases associated with AN, BN, and EDNOS, respectively, result in death due to medical complication and suicide (Crow et al., 2009). Eating disorders are also associated with self-harm which may reflect broader deficiencies in effective coping (e.g., Dingemans et al., 2007 and Woodside et al., 2004). Self-harm is defined in various ways in the literature and various terms are used to label self-harm, including self-injurious behavior, intentional self-injury, nonsuicidal self-injury, and self-mutilation. Self-harm, broadly defined, is the intentional injuring of one's own body without apparent suicidal intent (Pattison and Kahan, 1983). Self-harm occurs in various forms, including cutting, branding or burning, picking at skin or reopening wounds, pulling hair, hitting or punching, and head banging (Nock, 2008 and Skegg, 2005). Research has found high rates of self-harm among patients with eating disorders (Solano et al., 2005, Peebles et al., 2011 and Claes et al., 2001), and recent research has emphasized the importance of targeting self-harm tendencies in the context of eating disorder treatment (Fischer and Peterson, 2015). Research has also shown that self-harm is even more common among patients with binging/purging symptoms (Peebles et al., 2011 and Stein et al., 2004) who employ multiple purging methods (Stein et al., 2004). Drive for thinness and body dissatisfaction represent significant risk factors for disordered eating (Dobmeyer and Stein, 2003 and Stice et al., 2011). Drive for thinness consists of an excessive concern with dieting, preoccupation with weight, and fear of weight gain whereas body dissatisfaction involves not being satisfied with one's physical appearance (Garner, 1991). A growing body of research suggests that drive for thinness and body dissatisfaction may represent core endophenotypes of eating disorders (Keski-Rahkonen et al., 2005). Drive for thinness and body dissatisfaction have also been linked to self-harm (see Kostro et al., 2014 for review). For example, research has shown that body dissatisfaction is associated with higher rates of suicide attempts in adolescent boys and girls (Crow et al., 2008). A subsequent study also found that those who engage in self-harm display significantly more eating pathology than those who do not, including increased body dissatisfaction and bulimic tendencies. These findings suggest that body dissatisfaction and drive for thinness may also represent significant risk factors for engagement in self-harm.

3. Results 3.1. Rates of self-harm As seen in Table 2, the proportion of those who reported a history of self-harm and a history of suicide attempts differed significantly across all eating disorder subtypes. Specifically, 26.4% of those in the AN group, 40.7% of those in the BN group, and 49.0% of those in the Eating Disorder NOS group reported a history of self-harm—and all proportions were significantly different from each other (p < .05). Similarly, 12.7% of those in the AN group, 19.1% of those in the BN group, and 22.4% of those in the Eating Disorder NOS group reported a history of suicide attempts—and again, all proportions were significantly different from each other (p < .05). 3.2. Measurement model The measurement model—including the six latent factors of Self-Harm (at pre), Negative Affect (at pre), Body Size Aversion (at pre), Negative Affect (at post), and Eating Disorder Symptoms (at post)—was associated with good model fit (i.e., RMSEA = .050, CFI = .959). This measurement model may be seen in Fig. 1. All factor loadings and correlations between latent factors were significant. Notably, as seen in Fig. 1, the Body Size Aversion latent factors (at pre and post) are identified by only two subscales of the Eating Disorders Inventory–2 (i.e., Drive for Thinness and Body Dissatisfaction). We initially included a third Bulimia subscale in the measurement model; however, this model resulted in the Bulimia subscale indicator loading relatively low on the Eating Disorder Symptoms factor (i.e., <.30). Model fit associated with this measurement model was also somewhat low (i.e., CFI = .828; RMSEA = .086). Upon removal of the Bulimia subscale, model fit improved substantially (i.e., CFI = .959; RMSEA = .050), and all factor loadings were now significantly loaded on their respective latent factors throughout the measurement model. We therefore used this measurement model to underlie all subsequent SEM models. Table 3 contains the descriptive statistics and correlations for all observed indicators. Measurement model of the association between (pre) self-harm, (pre and post) ... Fig. 1. Measurement model of the association between (pre) self-harm, (pre and post) negative affect, and (pre and post) body size aversion (all path estimates are completely standardized). Figure options Table 3. Descriptive statistics and correlations of indicators of negative affect, harm exposure, and body Size aversion change (N = 2061). Indicators 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 1. DT pre-treatment – 2. DT post-treatment .52** – 3. B pre-treatment .28** .05* – 4. B post-treatment .18** .40** .44** – 5. BD pre-treatment .73** .44** .23** .19** – 6. BD post-treatment .47** .81** .04 .32** .58** – 7. BDI-II pre-treatment .53** .34** .25** .20** .52** .36** – 8. BDI-II post-treatment .27** .61** −.04 .33** .29** .58** .46** – 9. BAI pre-treatment .39** .24** .08** .16** .38** .24** .67** .38** – 10. BAI post-treatment .13** .37** −.10** .13** .12** .24** .23** .43** .28** – 11. Self-Harm History .17** .13** .05* .09** .25** .16** .29** .23** .29** .17** – 12. Suicide Attempt .10** .08** .06* .10** .16** .10** .25** .17** .22** .07** .37** – M 13.81 8.23 5.49 1.49 16.65 13.15 28.73 13.11 22.24 15.51 0.37 0.18 SD 6.37 6.75 6.01 2.97 8.29 8.86 12.61 11.67 11.22 10.50 0.48 0.38 Note. DT Difference Score = Eating Disorders Inventory - 2 Drive for Thinness Subscale Pre-Post Treatment Difference Score; B Difference Score = Eating Disorders Inventory – 2, Bulimia subscale Pre-Post Treatment Difference Score; BD Difference Score = Eating Disorders Inventory – 2 Body Dissatisfaction subscale Pre-Post Treatment Difference Score; BDI-II Difference Score = Beck Depression Inventory-II Pre-Post Treatment Difference Score; BAI Difference Score = Beck Anxiety Inventory Pre-Post Treatment Difference Score; Self-Harm History = Yes/No History of Self Harm; Suicide Attempt = Yes/No History of Suicide Attempt(s). **p < .01; *p < .05. Table options 3.3. SEM models and relationship between factors In the context of a latent difference score model, we regressed latent change of Body Size Aversion on latent change of Negative Affect (while controlling for differences in treatment duration). This model revealed that latent change of Negative Affect significantly predicted latent change in Body Size Aversion (following eating disorder treatment), ζ = .41, S.E. = .066, p < .01. We then examined the model whereby latent change in Body Size Aversion was regressed on latent Self Harm (while controlling for differences in treatment duration). This model revealed that Self Harm significantly predicted Body Size Aversion Change following eating disorder treatment (when Self Harm Exposure is examined in isolation), ζ = .15; S.E. = .041, p < .01. 2 Fig. 2 shows the results of the model whereby latent Body Size Aversion Change was regressed on latent Negative Affect Change and latent Self Harm simultaneously (while controlling for differences in treatment duration). In this model, Negative Affect Change significantly predicted Body Size Aversion Change when controlling for both treatment duration and Self-Harm (ζ = .41; S.E. = .052, p < .01). When controlling for treatment duration and Negative Affect Change, however, the effect on Self-Harm on latent Body Size Aversion Change was no longer significantly different from zero (ζ = .004; S.E. = .035, p = .36). Latent change in body size aversion regressed on latent self-harm and latent ... Fig. 2. Latent change in body size aversion regressed on latent self-harm and latent change of negative affect (all path estimates are completely standardized). Figure options 3.4. Structural model equivalence across eating disorder subtypes The subsample-specific results of the SEM model (whereby latent Self-Harm and latent change of Negative Affect simultaneously predicted latent change in Body Size Aversion in the Anorexia Nervosa only (n = 845), Bulimia Nervosa only (n = 565), and Eating Disorder Not Otherwise Specified (n = 651) subsamples) appears in Fig. 3. The overall pattern of results was similar across all three groups, including model fit, which was good across all subsamples (RMSEA range: .045–.059; CFI range: .947–.968). Across all three diagnostic subcategories, only latent change of Negative Affect significantly predicted latent change in Body Size Aversion when simultaneously predicted along with latent Self-Harm. In this simultaneous model, latent Self Harm no longer significantly predicted latent change in Body Size Aversion once controlling for latent change of Negative Affect. Latent change in body size aversion regressed on latent self-harm and latent ... Fig. 3. Latent change in body size aversion regressed on latent self-harm and latent change of negative affect in the anorexia nervosa (AN), bulimia nervosa (BN), and eating disorder not otherwise specified subsamples (all path estimates are completely standardized