- Email: [email protected]
Adult attachment, depression, and eating disorder symptoms: The mediating role of affect regulation strategies
Personality and Individual Differences 47 (2009) 662–667
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Adult attachment, depression, and eating disorder symptoms: The mediating role of affect regulation strategies Giorgio A. Tasca a,b,c,*, Leah Szadkowski a, Vanessa Illing a, Anne Trinneer a, Renee Grenon c, Natasha Demidenko b, Valerie Krysanski a,b, Louise Balfour a,b, Hany Bissada a,b a
University of Ottawa, 75 Laurier Avenue, East, Ottawa, ON, Canada K1N 6N5 The Ottawa Hospital, 501 Smyth Road, Box 400, Ottawa, ON, Canada K1H8L6 c Carleton University, 1125 Colonel By Drive, Ottawa, ON, Canada K1S 5B6 b
a r t i c l e
i n f o
Article history: Received 13 February 2009 Received in revised form 21 May 2009 Accepted 1 June 2009 Available online 21 June 2009 Keywords: Attachment Eating disorders Affect regulation Structural equation modeling
a b s t r a c t The study examined the role of affect regulation strategies in mediating the relationship between attachment dimensions and both depressive and eating disorder (ED) symptoms. Participants were 310 women seeking treatment for an ED. Structural equation modeling indicated that attachment anxiety contributed to both depressive symptoms and ED symptoms through emotional reactivity. In contrast, only the association between attachment avoidance and depressive symptoms was mediated by emotional deactivation; whereas attachment avoidance had a direct relationship with ED symptoms not mediated by emotional deactivation. The results suggest tailored clinical interventions that take into account attachment style and accompanying affect regulation strategies. Treatment of patients who experience attachment anxiety may emphasize impulse regulation, whereas treatment of ED patients with attachment avoidance could focus on gradual exposure to affective expression. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction During the past 30 years, attachment theory (Bowlby, 1973) has become one of the most important conceptual frameworks for understanding affect regulation and human relationships (Mikulincer & Shaver, 2007). Attachment is an inborn system that motivates an infant to seek proximity to a care-giving adult. The attachment system results in systematic patterns of interpersonal expectations, emotions, and behaviours that are associated with specific attachment-related strategies to regulate affect (Shaver & Mikulincer, 2002). Brennan, Clark, and Shaver (1998) argued that attachment can be characterized in terms of two orthogonal dimensions of attachment anxiety and attachment avoidance. Attachment anxiety may develop when attachment figures are inconsistent or unpredictable. In an attempt to maintain an engagement with inconsistent caregivers, these children ‘‘hyperactivate” their attachment systems. This involves excitatory pathways that intensify negative emotional responses by keeping them active in working memory resulting in an up-regulation of emotion (Shaver & Mikulincer, 2002). Attachment avoidance may develop when individuals perceive their primary attachment figure as rejecting or unavailable. In response, these individuals ‘‘deactivate” * Corresponding author. Address: The Ottawa Hospital, 501 Smyth Road, Box 400, Ottawa, ON, Canada K1H8L6. Tel.: +1 613 737 8035; fax: +1 613 737 8085. E-mail address: [email protected] (G.A. Tasca). 0191-8869/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2009.06.006
their attachment systems and defensively devalue their need for relationships. Thus, affective experiences are cut off from working memory, and this is accompanied by a down-regulation of emotion (Shaver & Mikulincer, 2002). Researchers link these attachment dimensions to psychological distress, including depression (e.g., Wei, Vogel, Ku, & Zakalik, 2005) and eating disorders (e.g., O’Kearney, 1996). In a recent study, Tasca et al., 2006 tested a structural equation model of the association between attachment insecurity and eating disorder (ED) symptoms. Attachment insecurity was directly related to body dissatisfaction and negative affect among a clinical sample of eating disordered women. EDs are considered to be among the most difficult to treat, and they have the highest rate of mortality among mental disorders (Agras, 2001). Treatments are not effective for 40% for those with bulimia nervosa (BN) in terms of reducing binge eating and purging (Stice, 1999), and recovery from anorexia nervosa (AN) tends to be even lower due to very high attrition rates (Tasca, Taylor, Bissada, Ritchie, & Balfour, 2004). AN restricting subtype (ANR) is often characterized by severe food restriction, body dissatisfaction, body image distortion, and very low body weight. Binge eating and/or purging also occur in the binge or purge subtype of AN (ANB). BN symptoms include body dissatisfaction and binge eating followed by inappropriate compensatory behaviours (e.g., vomiting). Individuals with an ED often suffer from current or lifetime history of depression (Pike & Striegel-Moore, 1997).
G.A. Tasca et al. / Personality and Individual Differences 47 (2009) 662–667
Problems with autonomy and separation from parents may be factors in the development and maintenance of an ED (O’Kearney, 1996). In their review, Laliberte, Boland, and Leichner (1999) concluded that patients with bulimic symptoms consistently reported their families to be less cohesive, less nurturing, less expressive, less communicative, and to have experienced more conflict. Clinton (2006) stated that difficulties with identifying, expressing, and regulating emotions are also common in patients with EDs. All of these factors can be understood as different aspects of human attachment processes (Bowlby, 1973). Researchers have become increasingly interested in the mechanisms by which attachment dimensions affect clinically relevant symptoms such as depression, anxiety and interpersonal problems (Mikulincer & Shaver, 2007). A number of studies have identified mediators between attachment and psychological distress. For example, Wei and colleagues (2005) found that affect regulation was a mediator between attachment dimensions and negative mood in a sample of college students. Affect regulation may have a role in the expression of eating disordered attitudes and behaviours. In a sample of college students, Cole-Detke and Kobak (1996) found that hyperactivating strategies were related to symptoms of depression, and that deactivating strategies were related to ED symptoms when depressive symptoms were controlled. Perry, DiTommaso, Robinson, and Doiron (2007) found that emotion focused coping partially mediated the relationship between attachment anxiety and body image disturbances and problem eating among undergraduates. Hilbert and Tuschen-Caffier (2007) studied affect regulation in a clinical sample of eating disordered women. They found that binge eating was preceded by difficulty in regulating affect in women with BN. One could argue that attachment insecurity may contribute to the development of maladaptive affect regulation strategies, which in turn may result in the expression of ED symptoms and depressive symptoms. For example, eating disordered patients with attachment anxiety may experience affect dysregulation that may result in symptoms such as purging behaviours. Conversely, those with attachment avoidance may cut off emotional experience, and this may aid extreme methods of dietary restriction. However, with the exception of the work by Wei and colleagues (2005), no studies have examined the mediating role of affect regulation to explain the relationship between attachment insecurity and distress. Further, no study has examined the mediating role of affect regulation in a sample of women with EDs. The purpose of this study was to examine the mediating role of affect regulation in a clinical sample of eating disordered women. We hypothesized that the association between attachment anxiety and both ED symptoms and depressive symptoms is mediated by emotional reactivity; and that the association between attachment avoidance and both ED symptoms and depressive symptoms is mediated by emotional deactivation. 2. Methods 2.1. Participants Participants were 310 adult females (>17 years) referred for assessment and treatment of an ED to a Center for Eating Disorders at a general hospital in a medium size urban center. Of the 459 individuals referred to the center between 2006 and 2008, 439 were women, and of those 353 met diagnostic criteria for AN, BN, or an eating disorder not otherwise specified (EDNOS; APA, 2000). Of those, 312 had valid personality assessment inventory (PAI; Morey, 1991) profiles (see definition of a valid PAI profile below) and had complete data. Two participants were identified as multivariate outliers, so 310 participants were included in the
663
analyses. Of those, 74 were diagnosed with AN, 138 with BN, and 98 with EDNOS. Mean age was 26.31 (SD = 8.76), mean body mass index (BMI = kg/m2) was 21.88 (SD = 6.20), and mean years with an ED was 7.46 (SD = 7.60). Most were Caucasian (91%), completed university (69%), and were never married (66%); and 34.5% had a comorbid affective disorder. 2.2. Instruments 2.2.1. Experiences in close relationships scale (ECR; Brennan et al., 1998) Attachment dimensions were assessed with the ECR, a 36-item self-report measure with a 7-point Likert-type response format from 1 (disagree strongly) to 7 (agree strongly), with higher scores representing greater levels of attachment anxiety or avoidance. The Attachment Anxiety subscale (18 items) assesses concern with rejection and preoccupation with abandonment. The Attachment Avoidance subscale (18 items) assesses fear of intimacy and discomfort with closeness or dependence. The coefficient alphas in this sample were .92 and .94, respectively. To generate three observed indicators for the two latent variables (i.e., attachment anxiety, attachment avoidance), we created three parcels of 6 items from each of the two scales (Russell, Kahn, Spoth, & Altmeir, 1998). Exploratory factor analyses using the maximum likelihood method were conducted separately on the items from the two scales. Items were rank ordered by magnitude of the factor loadings and successive pairs of the highest and lowest loading items were assigned to each of three parcels. 2.2.2. Differentiation of self inventory – revised (DSI–R; Skowron & Friedlander, 1998) The DSI–R contains 43 self-report items with two subscales of affect regulation used in this study, and two subscales of intimacy in relationships. Items have a 6-point Likert-type response format from 1 (not at all true of me) to 6 (very true of me). The emotional reactivity subscale (11 items) reflects the degree to which an individual responds with emotional flooding, emotional lability, or hypersensitivity. Lower scores represent greater emotional reactivity. The emotional cutoff subscale (12 items) reflects feeling threatened by intimacy, and isolating the self from others and emotions. Lower scores represent greater emotional deactivation. Coefficient alphas for this sample were .86 for each scale. The parcelling procedure described above for the attachment factors was also used to create indicator variables for the emotional reactivity and emotional deactivation factors. 2.2.3. Eating disorders inventory (EDI; Garner & Olmsted, 1984) The EDI was used to assess ED symptoms. The EDI has 64 items making up 8 scales. Factor analysis indicated that three scales represent an ED symptom factor:body dissatisfaction, bulimia, and drive for thinness (Tasca, Illing, Lybanon-Daigle, Bissada, & Balfour, 2003). Participants responded to items on a 6-point Likert-type scale ranging from ‘‘always” to ‘‘never”. The three most symptomatic choices were recoded as 1, 2, and 3, and the three least symptomatic choices all recoded zero. Higher scores on the scales represented higher ED attitudes and behaviours. Coefficient alphas for an ED sample were .92 for body dissatisfaction, .92 for bulimia, and .86 for drive for thinness (Garner & Olmsted, 1984). 2.2.4. Personality assessment inventory (PAI; Morey, 1991) The PAI has 344-items that make up 22 non-overlapping full scales measuring clinical constructs. Eleven of the full scales are made up of three or four subscales. The items are scored on a 4 point Likert-type scale ranging from 0 (not at all true) to 3 (very true). Total raw scores are converted to T-scores. Higher scores represent greater psychopathology. The three depression subscales
664
G.A. Tasca et al. / Personality and Individual Differences 47 (2009) 662–667
were used as indicator variables for the latent factor of depressive symptoms. The depression–cognitive subscale (8 items) assesses thoughts of worthlessness, hopelessness and personal failure. The depression–affective subscale (8 items) assesses sadness, loss of interest, and loss of pleasure. The depression–physiological subscale (8 items) assesses reduced energy, problems with sleep, and loss of sex drive. Among eating disordered patients, coefficient alphas for these scales were: .85 for depression–cognitive, .83 for depression– affective, and .79 for depression–physiological (Tasca, Wood, Demidenko, & Bissada, 2002). The PAI also has four scales used to identify invalid profiles caused by inconsistent responding (inconsistency scale), careless or random responding (infrequency scale), exaggerated negative responding ( negative impression scale), and exaggerated positive responding ( positive impression scale). Clinical cutoffs were used to identify invalid profiles, i.e., any T-scores above 73 for Inconsistency, 74 for Infrequency, 91 for negative impression, or 68 for positive impression (Morey, 1991). These participants were removed from the analyses. 2.3. Procedure Before being triaged to a treatment program, patients received a diagnostic assessment and completed psychological measures. All participants provided informed consent, and the study was approved by the research ethics board of the institution.
3. Results 3.1. Analysis strategy A measurement model was first tested for an acceptable fit to the data using confirmatory factor analysis (Anderson & Gerbing, 1988). Once a measurement model was estimated, a structural model was tested using the maximum likelihood method in AMOS (version 5.01). Criteria for acceptable model fit were a comparative fit index (CFI) greater than .95, and a root mean square error of approximation (RMSEA) of .08 or less (Byrne, 2001). We used the v2 difference test to compare nested models (Steiger, Shapiro, & Browne, 1985), with Bonferroni’s correction to the alpha level for the number of tests conducted, i.e., .05/5 = .01. To develop accurate estimates of standard errors of the indirect effects (i.e., mediated effects), Shrout and Bolger (2002) suggested a bootstrap procedure. From 2000 bootstrap samples, AMOS saved the estimates for indirect effects. If the 95% CI for the estimate of the indirect effect does not include zero, then the indirect effect is statistically significant at the .05 level (Shrout & Bolger, 2002). All skewness and kurtosis values for indicator variables were below |1.0| indicating a normal distribution of scores. 3.2. Measurement model A test of the measurement model resulted in a good fit to the data, v2(120, N = 310) = 228.52, p < .001, CFI = .97, RMSEA = .05 (90% CI: .04, .06). All of the loadings of the 18 measured variables on the six latent variables were significant (p < .001). Hence, all of the latent variables were adequately measured by their respective indicators. Table 1 indicates the correlations among the latent factors from the measurement model. 3.3. Structural model for testing mediated effects To test our mediation hypotheses, we assessed six alternative models. First we examined the hypothesized model (Model 1) depicted in Fig. 1. Three covariances in the model were specified
Table 1 Correlations among latent variables for the measurement model.
1. 2. 3. 4. 5. 6.
Attachment anxiety Attachment avoidance Emotional reactivity Emotional deactivation Depressive symptoms Eating disorder symptoms
1
2
–
.21 –
3
4 .68 .28
5 .36 .75 .53
–
6 .41 .42 .55 .54
–
.35 .40 .50 .39 .54
– –
N = 310. All correlations are significant at p < .001.
based on previous research indicating significant associations between: attachment anxiety and avoidance (Wei et al., 2005), emotional reactivity and deactivation (Skowron & Friedlander, 1998), and ED symptoms and depression (Pike & Striegel-Moore, 1997). Next, we compared Model 1 to four alternative models. Models 2 and 3 constrained to zero the direct paths from attachment anxiety to ED symptoms and depressive symptoms, respectively. With respect to attachment anxiety, these models assessed the mediating role of emotional reactivity separately for ED symptoms and depressive symptoms. Models 4 and 5 constrained to zero the direct paths from attachment avoidance to ED symptoms and depressive symptoms, respectively. With respect to attachment avoidance, these models assessed the mediating role of emotional deactivation separately for ED symptoms and depressive symptoms. Model 6 was tested as a final model in which all direct paths that did not significantly contribute to the fit of the hypothesized model were removed. The results indicated that Model 1 fit the data well: v2(122, N = 310) = 256.87, CFI = .96, RMSEA = .06 (90% CI: .05, .07). This model suggested that emotional reactivity was a mediator between attachment anxiety and ED and depressive symptoms, whereas, emotional deactivation was a mediator between attachment avoidance and depressive symptoms. The path between emotional deactivation and ED symptoms was not significant (Fig. 1), thus the mediating role for emotional deactivation in this relationship was not supported. Model 2, v2(123, N = 310) = 256.91, CFI = .96, RMSEA = .06 (90% CI: .05, .07); and Model 3, v2(123, N = 310) = 257.42, CFI = .96, RMSEA = .06 (90% CI: .05, .07) also fit the data well. Non significant v2 difference tests comparing Model 1 to Model 2, Dv2(1, N = 310) = .04, p > .05, and Models 1–3, Dv2(1, N = 310) = .55, p > .05, suggested that the direct paths from attachment anxiety to ED symptoms and from attachment anxiety to depressive symptoms did not contribute significantly to model fit. Hence, these direct paths were removed when the final model (Model 6) was tested. Model 4 fit the data well, v2(122, N = 310) = 268.68, CFI = .96, RMSEA = .06 (90% CI: .05, .07), however, a significant v2 difference test comparing Models 1–4, Dv2(1, N = 310) = 11.81, p < .001, indicated that Model 1 was a better fit. This confirmed that the direct path from attachment avoidance to ED symptoms should be kept in the final model (Model 6) as it contributed significantly to ED symptoms’ variance. Model 5 also fit the data well, v2(123, N = 310) = 259.02, CFI = .96, RMSEA = .06 (90% CI: .05, .07). A non significant v2 difference test comparing Models 1 and 5, Dv2(1, N = 310) = 2.15, p > .05, suggested that the direct path from attachment avoidance to depressive symptoms did not contribute significantly to model fit. Hence, this direct path was removed when the final model (Model 6) was tested. Model 6 (Fig. 2) fit the data well, v2(125, N = 310) = 256.87, CFI = .96, RMSEA = .06 (90% CI: .05, .07). When Model 6 was compared to Model 1, a non significant difference v2-test, Dv2(4, N = 310) = 3.75, p > .05, indicated that the direct paths from attachment anxiety to ED and depressive symptoms, and the direct path from attachment avoidance to depressive symptoms did not make
665
G.A. Tasca et al. / Personality and Individual Differences 47 (2009) 662–667
Emotional Reactivity
-.35***
-.64***
-.45*** Depressive Symptoms
.06
Attachment Anxiety
.15
-.02
.24***
.32***
.46***
.40*** Attachment Avoidance
Eating Disorder Symptoms
-.23*
-.73***
.13
Emotional Deactivation
Fig. 1. The hypothesized model: Model 1.
***
p < .001, *p < .05, v2(122, N = 310) = 256.87, CFI = .96, RMSEA = .06 (90% CI: .05, .07).
Emotional Reactivity
-.37***
-.64***
Attachment Anxiety
-.43***
.44***
.24***
Depressive Symptoms
.33** -.37*** .34***
Attachment Avoidance
-.74***
Eating Disorder Symptoms
.08
Emotional Deactivation
Fig. 2. The mediation model: Model 6.
***
p < .001, v2(125, N = 310) = 260.62, CFI = .96, RMSEA = .06 (90% CI: .05, .07).
a significant contribution to the hypothesized model. Therefore Model 6 was chosen as the best model. 3.3.1. Significance levels of indirect effects The results of the bootstrap method for testing indirect effects indicated that the mean indirect effect from attachment anxiety through emotional reactivity to ED symptoms, b = 1.05 (CI: .67, 1.50), b = .64 .37 = .24, p < .001, was significant and accounted for 24% of the variance in ED symptoms. The mean indirect effect from attachment anxiety to depressive symptoms, b = 2.90 (CI: 1.87, 4.06), b = .64 .43 = .28, p < .001, also was significant, and accounted for 28% of the variance in depressive symptoms. Finally, the indirect effect of attachment avoidance through emotional deactivation to depressive symptoms, b = 2.96 (CI: 2.04, 2.94), b = .74 .37 = .27, p < .001, was significant, and accounted for 27% of the variance in depressive symptoms. The indirect effect from attachment avoidance through emotional deactivation to ED symptoms was not significant, b = .36 (CI: .82, .36), b = .74 .08 = .06, p = .55, and accounted for only 6% of the variance in ED symptoms.
4. Discussion This study examined the role of affect regulation strategies in explaining the relationship between attachment insecurity and both ED and depressive symptoms. We found that the association between attachment anxiety and both ED and depressive symptoms was mediated by hyperactive affect regulation, and that the association between attachment avoidance and depressive symptoms was mediated by deactivating affect regulation. However, the significant relationship between attachment avoidance and ED symptoms was not mediated by deactivation of emotion. The results are consistent with a growing literature indicating that attachment insecurity acts upon clinically relevant indicators of distress, such as depression and interpersonal problems, through mediating psychological processes (e.g., Wei et al., 2005). Affect regulation strategies likely developed as an immediate way of dealing with a sub-optimal early environment (Wei et al., 2005). Emotional hyperactivation or deactivation initially may be adaptive for these individuals. However, the rigid and long term use of these strategies may give rise to ED and depressive
666
G.A. Tasca et al. / Personality and Individual Differences 47 (2009) 662–667
symptoms. This is consistent with interpersonal theory, which holds that rigidly adhered-to interpersonal behaviours and defences are the hallmarks of psychopathology (Kiesler, 1996). The continued use of emotional reactivity likely results in ongoing interpersonal conflict and eventual loss of relationships which reinforces the original need to hyperactivate the attachment system. Emotional deactivation, which originally was a protective strategy in response to nonresponsive attachment figures, results in loneliness and reduced social supports. Both of these affect regulation strategies leave the individual more vulnerable to depressive symptoms. As indicated by the results, emotional reactivity in the context of attachment anxiety was associated with ED symptoms. One could argue that preoccupation with relationships, a characteristic associated with attachment anxiety, in and of itself does not lead to ED symptoms. Rather, it is likely the dysregulated expression of emotions that results in ED attitudes and behaviours. Steiger and Bruce (2007) for example discussed affective instability and impulsivity as characteristic of a number of individuals with bulimia spectrum disorders including those diagnosed with BN and many diagnosed with EDNOS. The results also suggest that attachment avoidance is directly associated with ED symptoms but not mediated by emotional deactivation. The down-regulation of affect by avoidantly attached individuals had no immediate bearing on ED symptoms. However, other mediators may play a role. For example, the use of coping strategies such as denial and repression are known to be consistent with both attachment avoidance and with AN (Ward et al., 2001). Alternatively, the interpersonal aspect of attachment avoidance, i.e., devaluing relationships or viewing relationships as secondary, may mediate the expression of ED symptoms. Mixed feelings about the lack of relationships or of unsupportive relationships, coupled with a tendency to deny or repress such feelings, may exacerbate ED symptoms for avoidant individuals. Although Lopez and Brennan (2000) called for the development of attachment based interventions, such interventions for EDs are nearly non-existent. The results of this study suggest clinical avenues for treating eating disordered individuals with attachment anxiety or avoidance. Consistent with Wei and colleagues’ (2005) suggestion, clinicians would do well to recognize that individuals with EDs have varying attachment styles even within an ED diagnostic category. These variations in attachment may require different clinical strategies to help these individuals with their ED and depressive symptoms. Eating disordered individuals with attachment anxiety may benefit from a treatment approach that focuses on managing impulse regulation problems (Steiger & Bruce, 2007) and on increasing reflective functioning capacities in order to reduce the negative impact of affective instability and to improve relationships (Fonagy, Gergeley, Jurist, & Target, 2002). Attachment anxiety was associated with better treatment outcomes for eating disordered women if they received therapy that promoted self reflection, improved relationships, and affect regulation (Tasca et al., 2006). Eating disordered individuals with attachment avoidance may benefit from clinical interventions that are attuned to their discomfort with the intimacy and trust that is required in a therapeutic relationship. Keeping these patients connected and committed to their treatment may require an approach that gradually encourages them to become more open to their emotions (Wallin, 2007). Attachment avoidance has been associated with dropping out of treatment and with a decreasing therapeutic alliance (Tasca, Balfour, Ritchie, & Bissada, 2007; Tasca et al., 2004). Attending to attachment dimensions might improve the modest treatment outcomes for EDs (Stice, 1999). Some limitations of the present study are noted. First, the structural equation modeling results imply the possibility of causality,
but the design was cross sectional and correlational in nature. A longitudinal design that manipulates some variables would provide more direct evidence of causality. For example, treatment specifically aimed at reducing emotional reactivity would be expected to have positive effects on ED symptoms for individuals with attachment anxiety but not for those with attachment avoidance. Second, all of the data collected for this study were based on self-report measures that reflect consciously available self evaluations, whereas attachment interviews assess less consciously available states of mind (Shaver & Mikulincer, 2002). Replication of these results with other methods of data collection, such as the adult attachment interview (Main & Goldwyn, 1998) would strengthen the validity of the findings. Third, the majority of the sample were Caucasian women, and some researchers suggest that the expression of affect regulation is culturally bound (e.g., Wei et al., 2005). However, these participant characteristics are common among eating disordered samples (e.g., Striegel-Moore, Wilson, Wilfley, Elder, & Brownell, 1998). The results suggest that attachment dimensions and their accompanying affect regulation strategies are useful avenues to understanding symptom expression, and to inform the design and delivery of tailored clinical interventions to women with EDs. Clinicians may do well to assess attachment style among their eating disordered patients. Treatment of patients who experience attachment anxiety may emphasize impulse regulation and reflective functioning, whereas treatment of eating disordered patients with attachment avoidance could focus on gradual exposure to affective expression, and on interpersonal connectedness in the therapeutic relationship Acknowledgement The authors would like to thank Ann Barber for her help with this manuscript. References Agras, W. S. (2001). The consequences and costs of the eating disorders. The Psychiatric Clinics of North America, 24, 371–379. American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed. Text revision). Washington, DC: Author. Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice. A review and recommended two-step approach. Psychological Bulletin, 103, 411–423. Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation: Anxiety and anger. New York: Basic Books. Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measurement of adult attachment: An integrative overview. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 46–76). New York: Guilford Press. Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic concepts, applications, and programming. Mahwah, NJ: Lawrence Erlbaum Associates. Clinton, D. (2006). Affect regulation, object relations and the central symptoms of eating disorders. European Eating Disorders Review, 14, 203–211. Cole-Detke, H., & Kobak, R. (1996). Attachment processes in eating disorder and depression. Journal of Consulting and Clinical Psychology, 64, 282–290. Fonagy, P., Gergeley, G., Jurist, E. J., & Target, M. I. (2002). Affect regulation, mentalization, and the development of the self. New York: Other Press. Garner, D. M., & Olmsted, M. P. (1984). The eating disorders inventory. Odessa, FL: Psychological Assessment Resources. Hilbert, A., & Tuschen-Caffier, B. (2007). Maintenance of binge eating through negative mood: A naturalistic comparison of binge eating disorder and bulimia nervosa. International Journal of Eating Disorders, 40, 521–530. Kiesler, D. J. (1996). Contemporary interpersonal theory and research: Personality, psychopathology and psychotherapy. New York: Wiley. Laliberte, M., Boland, F. J., & Leichner, P. (1999). Family climates: Family factors specific to disturbed eating and bulimia nervosa. Journal of Clinical Psychology, 55, 1021–1040. Lopez, F. G., & Brennan, K. A. (2000). Dynamic processes underlying adult attachment organizations: Toward an attachment theoretical perspective on the healthy and effective self. Journal of Consulting and Clinical Psychology, 47, 283–301. Main, M., & Goldwyn, R. (1998). Adult attachment scoring and classification system. Unpublished manuscript. University of California at Berkeley. Mikulincer, M., & Shaver, P. R. (2007). Attachment in adulthood. New York: Guilford.
G.A. Tasca et al. / Personality and Individual Differences 47 (2009) 662–667 Morey, L. S. (1991). Personality assessment inventory professional manual. Odessa, FL: Psychological Assessment Resources. O’Kearney, R. (1996). Attachment disruption in anorexia nervosa and bulimia nervosa: A review of theory and empirical research. International Journal of Eating Disorders, 20, 115–127. Perry, S., DiTommaso, E., Robinson, B., & Doiron, Y. (2007). Attachment and coping: Implications for problematic eating behaviors and body image disturbances. Canadian Psychology, 48, 86 [Abstract]. Pike, K. M., & Striegel-Moore, R. (1997). Disordered eating and eating disorders. In S. G. Gallant, G. Puryear Keita, & R. Royak-Schaler (Eds.), Healthcare for women: Psychological social and behavioural influences (pp. 97–114). Washington, DC: American Psychological Association. Russell, D. W., Kahn, J. H., Spoth, R., & Altmeir, E. M. (1998). Analyzing data from experimental studies: A latent variable structural equation modeling approach. Journal of Counseling Psychology, 45, 18–29. Shaver, P. R., & Mikulincer, M. (2002). Attachment-related psychodynamics. Attachment and Human Development, 4, 133–161. Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and non-experimental studies: New procedures and recommendations. Psychological Methods, 7, 422–445. Skowron, E. A., & Friedlander, M. L. (1998). The differentiation of self inventory: Development and initial validation. Journal of Counseling Psychology, 45, 235–246. Steiger, H., & Bruce, K. R. (2007). Phenotypes, endophenotypes, and genotypes in bulimia spectrum eating disorders. Canadian Journal of Psychiatry, 52, 220–227. Steiger, J. H., Shapiro, A., & Browne, M. W. (1985). On the multivariate asymptotic distribution of sequential chi-square statistics. Psychometrika, 50, 253–264. Stice, E. (1999). Clinical implications of psychosocial research on bulimia nervosa and binge eating disorder. Psychotherapy in Practice, 55, 675–683.
667
Striegel-Moore, R. H., Wilson, G. T., Wilfley, D. E., Elder, K. A., & Brownell, K. D. (1998). Binge eating in an obese community sample. International Journal of Eating Disorders, 23, 27–37. Tasca, G. A., Balfour, L., Ritchie, K., & Bissada, H. (2007). The relationship between attachment scales and group therapy alliance growth differs by treatment type for women with Binge eating disorder. Group Dynamics: Theory, Research and Practice, 11, 1–14. Tasca, G. A., Illing, V., Lybanon-Daigle, V., Bissada, H., & Balfour, L. (2003). Psychometric properties of the eating disorders inventory – 2 among women seeking treatment for Binge eating disorder. Assessment, 10, 228–236. Tasca, G. A., Kowal, J., Balfour, L., Ritchie, K., Virley, B., & Bissada, H. (2006). An attachment insecurity model of negative affect among women seeking treatment for an eating disorder. Eating Behaviors, 7, 252–257. Tasca, G. A., Ritchie, K., Conrad, G., Balfour, L., Gayton, J., Daigle, V., et al. (2006). Attachment scales predict outcome in a randomized controlled trial of two group therapies for Binge eating disorder: An aptitude by treatment interaction. Psychotherapy Research, 16, 106–121. Tasca, G. A., Taylor, D., Bissada, H., Ritchie, K., & Balfour, L. (2004). Attachment predicts treatment completion in an eating disorders partial hospital program among women with anorexia nervosa. Journal of Personality Assessment, 83, 201–212. Tasca, G. A., Wood, J., Demidenko, N., & Bissada, H. (2002). Using the PAI in an eating disordered population: Scale characteristics and differences among diagnostic groups. Journal of Personality Assessment, 79, 337–356. Wallin, D. J. (2007). Attachment in psychotherapy. New York: Guilford. Ward, A., Ramsey, R., Turnbull, S., Steele, M., Steele, H., & Treasure, J. (2001). Attachment in anorexia nervosa: A transgenerational perspective. British Journal of Medical Psychology, 74, 497–505. Wei, M., Vogel, D. L., Ku, T., & Zakalik, R. A. (2005). Adult attachment, affect regulation, negative mood, and interpersonal problems: The mediating roles of emotional reactivity and emotional cutoff. Journal of Counseling Psychology, 52, 14–24.
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Adult attachment, depression, and eating disorder symptoms: The mediating role of affect regulation strategies Giorgio A. Tasca a,b,c,*, Leah Szadkowski a, Vanessa Illing a, Anne Trinneer a, Renee Grenon c, Natasha Demidenko b, Valerie Krysanski a,b, Louise Balfour a,b, Hany Bissada a,b a
University of Ottawa, 75 Laurier Avenue, East, Ottawa, ON, Canada K1N 6N5 The Ottawa Hospital, 501 Smyth Road, Box 400, Ottawa, ON, Canada K1H8L6 c Carleton University, 1125 Colonel By Drive, Ottawa, ON, Canada K1S 5B6 b
a r t i c l e
i n f o
Article history: Received 13 February 2009 Received in revised form 21 May 2009 Accepted 1 June 2009 Available online 21 June 2009 Keywords: Attachment Eating disorders Affect regulation Structural equation modeling
a b s t r a c t The study examined the role of affect regulation strategies in mediating the relationship between attachment dimensions and both depressive and eating disorder (ED) symptoms. Participants were 310 women seeking treatment for an ED. Structural equation modeling indicated that attachment anxiety contributed to both depressive symptoms and ED symptoms through emotional reactivity. In contrast, only the association between attachment avoidance and depressive symptoms was mediated by emotional deactivation; whereas attachment avoidance had a direct relationship with ED symptoms not mediated by emotional deactivation. The results suggest tailored clinical interventions that take into account attachment style and accompanying affect regulation strategies. Treatment of patients who experience attachment anxiety may emphasize impulse regulation, whereas treatment of ED patients with attachment avoidance could focus on gradual exposure to affective expression. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction During the past 30 years, attachment theory (Bowlby, 1973) has become one of the most important conceptual frameworks for understanding affect regulation and human relationships (Mikulincer & Shaver, 2007). Attachment is an inborn system that motivates an infant to seek proximity to a care-giving adult. The attachment system results in systematic patterns of interpersonal expectations, emotions, and behaviours that are associated with specific attachment-related strategies to regulate affect (Shaver & Mikulincer, 2002). Brennan, Clark, and Shaver (1998) argued that attachment can be characterized in terms of two orthogonal dimensions of attachment anxiety and attachment avoidance. Attachment anxiety may develop when attachment figures are inconsistent or unpredictable. In an attempt to maintain an engagement with inconsistent caregivers, these children ‘‘hyperactivate” their attachment systems. This involves excitatory pathways that intensify negative emotional responses by keeping them active in working memory resulting in an up-regulation of emotion (Shaver & Mikulincer, 2002). Attachment avoidance may develop when individuals perceive their primary attachment figure as rejecting or unavailable. In response, these individuals ‘‘deactivate” * Corresponding author. Address: The Ottawa Hospital, 501 Smyth Road, Box 400, Ottawa, ON, Canada K1H8L6. Tel.: +1 613 737 8035; fax: +1 613 737 8085. E-mail address: [email protected] (G.A. Tasca). 0191-8869/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2009.06.006
their attachment systems and defensively devalue their need for relationships. Thus, affective experiences are cut off from working memory, and this is accompanied by a down-regulation of emotion (Shaver & Mikulincer, 2002). Researchers link these attachment dimensions to psychological distress, including depression (e.g., Wei, Vogel, Ku, & Zakalik, 2005) and eating disorders (e.g., O’Kearney, 1996). In a recent study, Tasca et al., 2006 tested a structural equation model of the association between attachment insecurity and eating disorder (ED) symptoms. Attachment insecurity was directly related to body dissatisfaction and negative affect among a clinical sample of eating disordered women. EDs are considered to be among the most difficult to treat, and they have the highest rate of mortality among mental disorders (Agras, 2001). Treatments are not effective for 40% for those with bulimia nervosa (BN) in terms of reducing binge eating and purging (Stice, 1999), and recovery from anorexia nervosa (AN) tends to be even lower due to very high attrition rates (Tasca, Taylor, Bissada, Ritchie, & Balfour, 2004). AN restricting subtype (ANR) is often characterized by severe food restriction, body dissatisfaction, body image distortion, and very low body weight. Binge eating and/or purging also occur in the binge or purge subtype of AN (ANB). BN symptoms include body dissatisfaction and binge eating followed by inappropriate compensatory behaviours (e.g., vomiting). Individuals with an ED often suffer from current or lifetime history of depression (Pike & Striegel-Moore, 1997).
G.A. Tasca et al. / Personality and Individual Differences 47 (2009) 662–667
Problems with autonomy and separation from parents may be factors in the development and maintenance of an ED (O’Kearney, 1996). In their review, Laliberte, Boland, and Leichner (1999) concluded that patients with bulimic symptoms consistently reported their families to be less cohesive, less nurturing, less expressive, less communicative, and to have experienced more conflict. Clinton (2006) stated that difficulties with identifying, expressing, and regulating emotions are also common in patients with EDs. All of these factors can be understood as different aspects of human attachment processes (Bowlby, 1973). Researchers have become increasingly interested in the mechanisms by which attachment dimensions affect clinically relevant symptoms such as depression, anxiety and interpersonal problems (Mikulincer & Shaver, 2007). A number of studies have identified mediators between attachment and psychological distress. For example, Wei and colleagues (2005) found that affect regulation was a mediator between attachment dimensions and negative mood in a sample of college students. Affect regulation may have a role in the expression of eating disordered attitudes and behaviours. In a sample of college students, Cole-Detke and Kobak (1996) found that hyperactivating strategies were related to symptoms of depression, and that deactivating strategies were related to ED symptoms when depressive symptoms were controlled. Perry, DiTommaso, Robinson, and Doiron (2007) found that emotion focused coping partially mediated the relationship between attachment anxiety and body image disturbances and problem eating among undergraduates. Hilbert and Tuschen-Caffier (2007) studied affect regulation in a clinical sample of eating disordered women. They found that binge eating was preceded by difficulty in regulating affect in women with BN. One could argue that attachment insecurity may contribute to the development of maladaptive affect regulation strategies, which in turn may result in the expression of ED symptoms and depressive symptoms. For example, eating disordered patients with attachment anxiety may experience affect dysregulation that may result in symptoms such as purging behaviours. Conversely, those with attachment avoidance may cut off emotional experience, and this may aid extreme methods of dietary restriction. However, with the exception of the work by Wei and colleagues (2005), no studies have examined the mediating role of affect regulation to explain the relationship between attachment insecurity and distress. Further, no study has examined the mediating role of affect regulation in a sample of women with EDs. The purpose of this study was to examine the mediating role of affect regulation in a clinical sample of eating disordered women. We hypothesized that the association between attachment anxiety and both ED symptoms and depressive symptoms is mediated by emotional reactivity; and that the association between attachment avoidance and both ED symptoms and depressive symptoms is mediated by emotional deactivation. 2. Methods 2.1. Participants Participants were 310 adult females (>17 years) referred for assessment and treatment of an ED to a Center for Eating Disorders at a general hospital in a medium size urban center. Of the 459 individuals referred to the center between 2006 and 2008, 439 were women, and of those 353 met diagnostic criteria for AN, BN, or an eating disorder not otherwise specified (EDNOS; APA, 2000). Of those, 312 had valid personality assessment inventory (PAI; Morey, 1991) profiles (see definition of a valid PAI profile below) and had complete data. Two participants were identified as multivariate outliers, so 310 participants were included in the
663
analyses. Of those, 74 were diagnosed with AN, 138 with BN, and 98 with EDNOS. Mean age was 26.31 (SD = 8.76), mean body mass index (BMI = kg/m2) was 21.88 (SD = 6.20), and mean years with an ED was 7.46 (SD = 7.60). Most were Caucasian (91%), completed university (69%), and were never married (66%); and 34.5% had a comorbid affective disorder. 2.2. Instruments 2.2.1. Experiences in close relationships scale (ECR; Brennan et al., 1998) Attachment dimensions were assessed with the ECR, a 36-item self-report measure with a 7-point Likert-type response format from 1 (disagree strongly) to 7 (agree strongly), with higher scores representing greater levels of attachment anxiety or avoidance. The Attachment Anxiety subscale (18 items) assesses concern with rejection and preoccupation with abandonment. The Attachment Avoidance subscale (18 items) assesses fear of intimacy and discomfort with closeness or dependence. The coefficient alphas in this sample were .92 and .94, respectively. To generate three observed indicators for the two latent variables (i.e., attachment anxiety, attachment avoidance), we created three parcels of 6 items from each of the two scales (Russell, Kahn, Spoth, & Altmeir, 1998). Exploratory factor analyses using the maximum likelihood method were conducted separately on the items from the two scales. Items were rank ordered by magnitude of the factor loadings and successive pairs of the highest and lowest loading items were assigned to each of three parcels. 2.2.2. Differentiation of self inventory – revised (DSI–R; Skowron & Friedlander, 1998) The DSI–R contains 43 self-report items with two subscales of affect regulation used in this study, and two subscales of intimacy in relationships. Items have a 6-point Likert-type response format from 1 (not at all true of me) to 6 (very true of me). The emotional reactivity subscale (11 items) reflects the degree to which an individual responds with emotional flooding, emotional lability, or hypersensitivity. Lower scores represent greater emotional reactivity. The emotional cutoff subscale (12 items) reflects feeling threatened by intimacy, and isolating the self from others and emotions. Lower scores represent greater emotional deactivation. Coefficient alphas for this sample were .86 for each scale. The parcelling procedure described above for the attachment factors was also used to create indicator variables for the emotional reactivity and emotional deactivation factors. 2.2.3. Eating disorders inventory (EDI; Garner & Olmsted, 1984) The EDI was used to assess ED symptoms. The EDI has 64 items making up 8 scales. Factor analysis indicated that three scales represent an ED symptom factor:body dissatisfaction, bulimia, and drive for thinness (Tasca, Illing, Lybanon-Daigle, Bissada, & Balfour, 2003). Participants responded to items on a 6-point Likert-type scale ranging from ‘‘always” to ‘‘never”. The three most symptomatic choices were recoded as 1, 2, and 3, and the three least symptomatic choices all recoded zero. Higher scores on the scales represented higher ED attitudes and behaviours. Coefficient alphas for an ED sample were .92 for body dissatisfaction, .92 for bulimia, and .86 for drive for thinness (Garner & Olmsted, 1984). 2.2.4. Personality assessment inventory (PAI; Morey, 1991) The PAI has 344-items that make up 22 non-overlapping full scales measuring clinical constructs. Eleven of the full scales are made up of three or four subscales. The items are scored on a 4 point Likert-type scale ranging from 0 (not at all true) to 3 (very true). Total raw scores are converted to T-scores. Higher scores represent greater psychopathology. The three depression subscales
664
G.A. Tasca et al. / Personality and Individual Differences 47 (2009) 662–667
were used as indicator variables for the latent factor of depressive symptoms. The depression–cognitive subscale (8 items) assesses thoughts of worthlessness, hopelessness and personal failure. The depression–affective subscale (8 items) assesses sadness, loss of interest, and loss of pleasure. The depression–physiological subscale (8 items) assesses reduced energy, problems with sleep, and loss of sex drive. Among eating disordered patients, coefficient alphas for these scales were: .85 for depression–cognitive, .83 for depression– affective, and .79 for depression–physiological (Tasca, Wood, Demidenko, & Bissada, 2002). The PAI also has four scales used to identify invalid profiles caused by inconsistent responding (inconsistency scale), careless or random responding (infrequency scale), exaggerated negative responding ( negative impression scale), and exaggerated positive responding ( positive impression scale). Clinical cutoffs were used to identify invalid profiles, i.e., any T-scores above 73 for Inconsistency, 74 for Infrequency, 91 for negative impression, or 68 for positive impression (Morey, 1991). These participants were removed from the analyses. 2.3. Procedure Before being triaged to a treatment program, patients received a diagnostic assessment and completed psychological measures. All participants provided informed consent, and the study was approved by the research ethics board of the institution.
3. Results 3.1. Analysis strategy A measurement model was first tested for an acceptable fit to the data using confirmatory factor analysis (Anderson & Gerbing, 1988). Once a measurement model was estimated, a structural model was tested using the maximum likelihood method in AMOS (version 5.01). Criteria for acceptable model fit were a comparative fit index (CFI) greater than .95, and a root mean square error of approximation (RMSEA) of .08 or less (Byrne, 2001). We used the v2 difference test to compare nested models (Steiger, Shapiro, & Browne, 1985), with Bonferroni’s correction to the alpha level for the number of tests conducted, i.e., .05/5 = .01. To develop accurate estimates of standard errors of the indirect effects (i.e., mediated effects), Shrout and Bolger (2002) suggested a bootstrap procedure. From 2000 bootstrap samples, AMOS saved the estimates for indirect effects. If the 95% CI for the estimate of the indirect effect does not include zero, then the indirect effect is statistically significant at the .05 level (Shrout & Bolger, 2002). All skewness and kurtosis values for indicator variables were below |1.0| indicating a normal distribution of scores. 3.2. Measurement model A test of the measurement model resulted in a good fit to the data, v2(120, N = 310) = 228.52, p < .001, CFI = .97, RMSEA = .05 (90% CI: .04, .06). All of the loadings of the 18 measured variables on the six latent variables were significant (p < .001). Hence, all of the latent variables were adequately measured by their respective indicators. Table 1 indicates the correlations among the latent factors from the measurement model. 3.3. Structural model for testing mediated effects To test our mediation hypotheses, we assessed six alternative models. First we examined the hypothesized model (Model 1) depicted in Fig. 1. Three covariances in the model were specified
Table 1 Correlations among latent variables for the measurement model.
1. 2. 3. 4. 5. 6.
Attachment anxiety Attachment avoidance Emotional reactivity Emotional deactivation Depressive symptoms Eating disorder symptoms
1
2
–
.21 –
3
4 .68 .28
5 .36 .75 .53
–
6 .41 .42 .55 .54
–
.35 .40 .50 .39 .54
– –
N = 310. All correlations are significant at p < .001.
based on previous research indicating significant associations between: attachment anxiety and avoidance (Wei et al., 2005), emotional reactivity and deactivation (Skowron & Friedlander, 1998), and ED symptoms and depression (Pike & Striegel-Moore, 1997). Next, we compared Model 1 to four alternative models. Models 2 and 3 constrained to zero the direct paths from attachment anxiety to ED symptoms and depressive symptoms, respectively. With respect to attachment anxiety, these models assessed the mediating role of emotional reactivity separately for ED symptoms and depressive symptoms. Models 4 and 5 constrained to zero the direct paths from attachment avoidance to ED symptoms and depressive symptoms, respectively. With respect to attachment avoidance, these models assessed the mediating role of emotional deactivation separately for ED symptoms and depressive symptoms. Model 6 was tested as a final model in which all direct paths that did not significantly contribute to the fit of the hypothesized model were removed. The results indicated that Model 1 fit the data well: v2(122, N = 310) = 256.87, CFI = .96, RMSEA = .06 (90% CI: .05, .07). This model suggested that emotional reactivity was a mediator between attachment anxiety and ED and depressive symptoms, whereas, emotional deactivation was a mediator between attachment avoidance and depressive symptoms. The path between emotional deactivation and ED symptoms was not significant (Fig. 1), thus the mediating role for emotional deactivation in this relationship was not supported. Model 2, v2(123, N = 310) = 256.91, CFI = .96, RMSEA = .06 (90% CI: .05, .07); and Model 3, v2(123, N = 310) = 257.42, CFI = .96, RMSEA = .06 (90% CI: .05, .07) also fit the data well. Non significant v2 difference tests comparing Model 1 to Model 2, Dv2(1, N = 310) = .04, p > .05, and Models 1–3, Dv2(1, N = 310) = .55, p > .05, suggested that the direct paths from attachment anxiety to ED symptoms and from attachment anxiety to depressive symptoms did not contribute significantly to model fit. Hence, these direct paths were removed when the final model (Model 6) was tested. Model 4 fit the data well, v2(122, N = 310) = 268.68, CFI = .96, RMSEA = .06 (90% CI: .05, .07), however, a significant v2 difference test comparing Models 1–4, Dv2(1, N = 310) = 11.81, p < .001, indicated that Model 1 was a better fit. This confirmed that the direct path from attachment avoidance to ED symptoms should be kept in the final model (Model 6) as it contributed significantly to ED symptoms’ variance. Model 5 also fit the data well, v2(123, N = 310) = 259.02, CFI = .96, RMSEA = .06 (90% CI: .05, .07). A non significant v2 difference test comparing Models 1 and 5, Dv2(1, N = 310) = 2.15, p > .05, suggested that the direct path from attachment avoidance to depressive symptoms did not contribute significantly to model fit. Hence, this direct path was removed when the final model (Model 6) was tested. Model 6 (Fig. 2) fit the data well, v2(125, N = 310) = 256.87, CFI = .96, RMSEA = .06 (90% CI: .05, .07). When Model 6 was compared to Model 1, a non significant difference v2-test, Dv2(4, N = 310) = 3.75, p > .05, indicated that the direct paths from attachment anxiety to ED and depressive symptoms, and the direct path from attachment avoidance to depressive symptoms did not make
665
G.A. Tasca et al. / Personality and Individual Differences 47 (2009) 662–667
Emotional Reactivity
-.35***
-.64***
-.45*** Depressive Symptoms
.06
Attachment Anxiety
.15
-.02
.24***
.32***
.46***
.40*** Attachment Avoidance
Eating Disorder Symptoms
-.23*
-.73***
.13
Emotional Deactivation
Fig. 1. The hypothesized model: Model 1.
***
p < .001, *p < .05, v2(122, N = 310) = 256.87, CFI = .96, RMSEA = .06 (90% CI: .05, .07).
Emotional Reactivity
-.37***
-.64***
Attachment Anxiety
-.43***
.44***
.24***
Depressive Symptoms
.33** -.37*** .34***
Attachment Avoidance
-.74***
Eating Disorder Symptoms
.08
Emotional Deactivation
Fig. 2. The mediation model: Model 6.
***
p < .001, v2(125, N = 310) = 260.62, CFI = .96, RMSEA = .06 (90% CI: .05, .07).
a significant contribution to the hypothesized model. Therefore Model 6 was chosen as the best model. 3.3.1. Significance levels of indirect effects The results of the bootstrap method for testing indirect effects indicated that the mean indirect effect from attachment anxiety through emotional reactivity to ED symptoms, b = 1.05 (CI: .67, 1.50), b = .64 .37 = .24, p < .001, was significant and accounted for 24% of the variance in ED symptoms. The mean indirect effect from attachment anxiety to depressive symptoms, b = 2.90 (CI: 1.87, 4.06), b = .64 .43 = .28, p < .001, also was significant, and accounted for 28% of the variance in depressive symptoms. Finally, the indirect effect of attachment avoidance through emotional deactivation to depressive symptoms, b = 2.96 (CI: 2.04, 2.94), b = .74 .37 = .27, p < .001, was significant, and accounted for 27% of the variance in depressive symptoms. The indirect effect from attachment avoidance through emotional deactivation to ED symptoms was not significant, b = .36 (CI: .82, .36), b = .74 .08 = .06, p = .55, and accounted for only 6% of the variance in ED symptoms.
4. Discussion This study examined the role of affect regulation strategies in explaining the relationship between attachment insecurity and both ED and depressive symptoms. We found that the association between attachment anxiety and both ED and depressive symptoms was mediated by hyperactive affect regulation, and that the association between attachment avoidance and depressive symptoms was mediated by deactivating affect regulation. However, the significant relationship between attachment avoidance and ED symptoms was not mediated by deactivation of emotion. The results are consistent with a growing literature indicating that attachment insecurity acts upon clinically relevant indicators of distress, such as depression and interpersonal problems, through mediating psychological processes (e.g., Wei et al., 2005). Affect regulation strategies likely developed as an immediate way of dealing with a sub-optimal early environment (Wei et al., 2005). Emotional hyperactivation or deactivation initially may be adaptive for these individuals. However, the rigid and long term use of these strategies may give rise to ED and depressive
666
G.A. Tasca et al. / Personality and Individual Differences 47 (2009) 662–667
symptoms. This is consistent with interpersonal theory, which holds that rigidly adhered-to interpersonal behaviours and defences are the hallmarks of psychopathology (Kiesler, 1996). The continued use of emotional reactivity likely results in ongoing interpersonal conflict and eventual loss of relationships which reinforces the original need to hyperactivate the attachment system. Emotional deactivation, which originally was a protective strategy in response to nonresponsive attachment figures, results in loneliness and reduced social supports. Both of these affect regulation strategies leave the individual more vulnerable to depressive symptoms. As indicated by the results, emotional reactivity in the context of attachment anxiety was associated with ED symptoms. One could argue that preoccupation with relationships, a characteristic associated with attachment anxiety, in and of itself does not lead to ED symptoms. Rather, it is likely the dysregulated expression of emotions that results in ED attitudes and behaviours. Steiger and Bruce (2007) for example discussed affective instability and impulsivity as characteristic of a number of individuals with bulimia spectrum disorders including those diagnosed with BN and many diagnosed with EDNOS. The results also suggest that attachment avoidance is directly associated with ED symptoms but not mediated by emotional deactivation. The down-regulation of affect by avoidantly attached individuals had no immediate bearing on ED symptoms. However, other mediators may play a role. For example, the use of coping strategies such as denial and repression are known to be consistent with both attachment avoidance and with AN (Ward et al., 2001). Alternatively, the interpersonal aspect of attachment avoidance, i.e., devaluing relationships or viewing relationships as secondary, may mediate the expression of ED symptoms. Mixed feelings about the lack of relationships or of unsupportive relationships, coupled with a tendency to deny or repress such feelings, may exacerbate ED symptoms for avoidant individuals. Although Lopez and Brennan (2000) called for the development of attachment based interventions, such interventions for EDs are nearly non-existent. The results of this study suggest clinical avenues for treating eating disordered individuals with attachment anxiety or avoidance. Consistent with Wei and colleagues’ (2005) suggestion, clinicians would do well to recognize that individuals with EDs have varying attachment styles even within an ED diagnostic category. These variations in attachment may require different clinical strategies to help these individuals with their ED and depressive symptoms. Eating disordered individuals with attachment anxiety may benefit from a treatment approach that focuses on managing impulse regulation problems (Steiger & Bruce, 2007) and on increasing reflective functioning capacities in order to reduce the negative impact of affective instability and to improve relationships (Fonagy, Gergeley, Jurist, & Target, 2002). Attachment anxiety was associated with better treatment outcomes for eating disordered women if they received therapy that promoted self reflection, improved relationships, and affect regulation (Tasca et al., 2006). Eating disordered individuals with attachment avoidance may benefit from clinical interventions that are attuned to their discomfort with the intimacy and trust that is required in a therapeutic relationship. Keeping these patients connected and committed to their treatment may require an approach that gradually encourages them to become more open to their emotions (Wallin, 2007). Attachment avoidance has been associated with dropping out of treatment and with a decreasing therapeutic alliance (Tasca, Balfour, Ritchie, & Bissada, 2007; Tasca et al., 2004). Attending to attachment dimensions might improve the modest treatment outcomes for EDs (Stice, 1999). Some limitations of the present study are noted. First, the structural equation modeling results imply the possibility of causality,
but the design was cross sectional and correlational in nature. A longitudinal design that manipulates some variables would provide more direct evidence of causality. For example, treatment specifically aimed at reducing emotional reactivity would be expected to have positive effects on ED symptoms for individuals with attachment anxiety but not for those with attachment avoidance. Second, all of the data collected for this study were based on self-report measures that reflect consciously available self evaluations, whereas attachment interviews assess less consciously available states of mind (Shaver & Mikulincer, 2002). Replication of these results with other methods of data collection, such as the adult attachment interview (Main & Goldwyn, 1998) would strengthen the validity of the findings. Third, the majority of the sample were Caucasian women, and some researchers suggest that the expression of affect regulation is culturally bound (e.g., Wei et al., 2005). However, these participant characteristics are common among eating disordered samples (e.g., Striegel-Moore, Wilson, Wilfley, Elder, & Brownell, 1998). The results suggest that attachment dimensions and their accompanying affect regulation strategies are useful avenues to understanding symptom expression, and to inform the design and delivery of tailored clinical interventions to women with EDs. Clinicians may do well to assess attachment style among their eating disordered patients. Treatment of patients who experience attachment anxiety may emphasize impulse regulation and reflective functioning, whereas treatment of eating disordered patients with attachment avoidance could focus on gradual exposure to affective expression, and on interpersonal connectedness in the therapeutic relationship Acknowledgement The authors would like to thank Ann Barber for her help with this manuscript. References Agras, W. S. (2001). The consequences and costs of the eating disorders. The Psychiatric Clinics of North America, 24, 371–379. American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed. Text revision). Washington, DC: Author. Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice. A review and recommended two-step approach. Psychological Bulletin, 103, 411–423. Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation: Anxiety and anger. New York: Basic Books. Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measurement of adult attachment: An integrative overview. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 46–76). New York: Guilford Press. Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic concepts, applications, and programming. Mahwah, NJ: Lawrence Erlbaum Associates. Clinton, D. (2006). Affect regulation, object relations and the central symptoms of eating disorders. European Eating Disorders Review, 14, 203–211. Cole-Detke, H., & Kobak, R. (1996). Attachment processes in eating disorder and depression. Journal of Consulting and Clinical Psychology, 64, 282–290. Fonagy, P., Gergeley, G., Jurist, E. J., & Target, M. I. (2002). Affect regulation, mentalization, and the development of the self. New York: Other Press. Garner, D. M., & Olmsted, M. P. (1984). The eating disorders inventory. Odessa, FL: Psychological Assessment Resources. Hilbert, A., & Tuschen-Caffier, B. (2007). Maintenance of binge eating through negative mood: A naturalistic comparison of binge eating disorder and bulimia nervosa. International Journal of Eating Disorders, 40, 521–530. Kiesler, D. J. (1996). Contemporary interpersonal theory and research: Personality, psychopathology and psychotherapy. New York: Wiley. Laliberte, M., Boland, F. J., & Leichner, P. (1999). Family climates: Family factors specific to disturbed eating and bulimia nervosa. Journal of Clinical Psychology, 55, 1021–1040. Lopez, F. G., & Brennan, K. A. (2000). Dynamic processes underlying adult attachment organizations: Toward an attachment theoretical perspective on the healthy and effective self. Journal of Consulting and Clinical Psychology, 47, 283–301. Main, M., & Goldwyn, R. (1998). Adult attachment scoring and classification system. Unpublished manuscript. University of California at Berkeley. Mikulincer, M., & Shaver, P. R. (2007). Attachment in adulthood. New York: Guilford.
G.A. Tasca et al. / Personality and Individual Differences 47 (2009) 662–667 Morey, L. S. (1991). Personality assessment inventory professional manual. Odessa, FL: Psychological Assessment Resources. O’Kearney, R. (1996). Attachment disruption in anorexia nervosa and bulimia nervosa: A review of theory and empirical research. International Journal of Eating Disorders, 20, 115–127. Perry, S., DiTommaso, E., Robinson, B., & Doiron, Y. (2007). Attachment and coping: Implications for problematic eating behaviors and body image disturbances. Canadian Psychology, 48, 86 [Abstract]. Pike, K. M., & Striegel-Moore, R. (1997). Disordered eating and eating disorders. In S. G. Gallant, G. Puryear Keita, & R. Royak-Schaler (Eds.), Healthcare for women: Psychological social and behavioural influences (pp. 97–114). Washington, DC: American Psychological Association. Russell, D. W., Kahn, J. H., Spoth, R., & Altmeir, E. M. (1998). Analyzing data from experimental studies: A latent variable structural equation modeling approach. Journal of Counseling Psychology, 45, 18–29. Shaver, P. R., & Mikulincer, M. (2002). Attachment-related psychodynamics. Attachment and Human Development, 4, 133–161. Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and non-experimental studies: New procedures and recommendations. Psychological Methods, 7, 422–445. Skowron, E. A., & Friedlander, M. L. (1998). The differentiation of self inventory: Development and initial validation. Journal of Counseling Psychology, 45, 235–246. Steiger, H., & Bruce, K. R. (2007). Phenotypes, endophenotypes, and genotypes in bulimia spectrum eating disorders. Canadian Journal of Psychiatry, 52, 220–227. Steiger, J. H., Shapiro, A., & Browne, M. W. (1985). On the multivariate asymptotic distribution of sequential chi-square statistics. Psychometrika, 50, 253–264. Stice, E. (1999). Clinical implications of psychosocial research on bulimia nervosa and binge eating disorder. Psychotherapy in Practice, 55, 675–683.
667
Striegel-Moore, R. H., Wilson, G. T., Wilfley, D. E., Elder, K. A., & Brownell, K. D. (1998). Binge eating in an obese community sample. International Journal of Eating Disorders, 23, 27–37. Tasca, G. A., Balfour, L., Ritchie, K., & Bissada, H. (2007). The relationship between attachment scales and group therapy alliance growth differs by treatment type for women with Binge eating disorder. Group Dynamics: Theory, Research and Practice, 11, 1–14. Tasca, G. A., Illing, V., Lybanon-Daigle, V., Bissada, H., & Balfour, L. (2003). Psychometric properties of the eating disorders inventory – 2 among women seeking treatment for Binge eating disorder. Assessment, 10, 228–236. Tasca, G. A., Kowal, J., Balfour, L., Ritchie, K., Virley, B., & Bissada, H. (2006). An attachment insecurity model of negative affect among women seeking treatment for an eating disorder. Eating Behaviors, 7, 252–257. Tasca, G. A., Ritchie, K., Conrad, G., Balfour, L., Gayton, J., Daigle, V., et al. (2006). Attachment scales predict outcome in a randomized controlled trial of two group therapies for Binge eating disorder: An aptitude by treatment interaction. Psychotherapy Research, 16, 106–121. Tasca, G. A., Taylor, D., Bissada, H., Ritchie, K., & Balfour, L. (2004). Attachment predicts treatment completion in an eating disorders partial hospital program among women with anorexia nervosa. Journal of Personality Assessment, 83, 201–212. Tasca, G. A., Wood, J., Demidenko, N., & Bissada, H. (2002). Using the PAI in an eating disordered population: Scale characteristics and differences among diagnostic groups. Journal of Personality Assessment, 79, 337–356. Wallin, D. J. (2007). Attachment in psychotherapy. New York: Guilford. Ward, A., Ramsey, R., Turnbull, S., Steele, M., Steele, H., & Treasure, J. (2001). Attachment in anorexia nervosa: A transgenerational perspective. British Journal of Medical Psychology, 74, 497–505. Wei, M., Vogel, D. L., Ku, T., & Zakalik, R. A. (2005). Adult attachment, affect regulation, negative mood, and interpersonal problems: The mediating roles of emotional reactivity and emotional cutoff. Journal of Counseling Psychology, 52, 14–24.