- Email: [email protected]
Outcome of group cognitive-behavior therapy for bulimia nervosa: the role of core beliefs
Behaviour Research and Therapy 38 (2000) 145±156
www.elsevier.com/locate/brat
Outcome of group cognitive-behavior therapy for bulimia nervosa: the role of core beliefs Newman Leung a,b,*, Glenn Waller c, Glyn Thomas b a
Eating Disorders Service, Queen Elizabeth Psychiatric Hospital, Mindelsohn Way, Edgbaston, Birmingham B15 2QZ, UK b School of Psychology, University of Birmingham, Birmingham, UK c Department of Psychology, University of Southampton, Southampton, UK Received 3 December 1998; accepted 3 December 1998
Abstract It is hypothesized that individuals who bene®t less from CBT will be those who have more pathological core beliefs (unconditional beliefs, unrelated to food, shape and weight). Twenty bulimic women were treated using 12 sessions of conventional group CBT. Eating behavior and attitudes were assessed pre- and posttreatment. Core beliefs were assessed at the beginning of the programme, and were used as predictors of change across treatment (once any eect of pretreatment psychopathology was taken into account). Group CBT was eective, with reductions of over 50% in bulimic symptoms. Outcome on most indices was associated with pretreatment levels of pathological core beliefs. Possible reasons for these ®ndings are discussed. # 2000 Elsevier Science Ltd. All rights reserved. Keywords: Bulimia nervosa; Cognitive-behavioral therapy; Core beliefs
1. Introduction Cognitive behavioral therapy (CBT) is generally considered to be the treatment of choice for bulimia nervosa (e.g. Fairburn, 1988; Craighead & Agras, 1991; Wilson & Fairburn, 1993). For example, in a survey of clinicians, Herzog, Keller, Strober, Yeh and Pai (1992) found that 85± 94% of respondents indicated that they would consider using cognitive behavioral therapy alone or in conjunction with other approaches in the treatment of bulimia nervosa. This strong
* Corresponding author. 0005-7967/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 5 - 7 9 6 7 ( 9 9 ) 0 0 0 2 6 - 1
146
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
preference of clinicians is supported by empirical ®ndings, which indicate a high ecacy for CBT. In a review of its eectiveness, Vitousek (1996) concluded that individual CBT typically produces an 80% reduction of bingeing and purging symptoms, and a total elimination of bulimic episodes in 50% of treated clients. In addition to its use in individual treatment, CBT has been used in group settings. Group CBT has been demonstrated to be highly cost-eective (e.g. Garner, 1987; Mitchell et al., 1990; Agras et al., 1992; Fettes & Peters, 1992), although the degree of reduction in bulimic symptomatology may be more modest than with individual treatment (Freeman, 1995). Despite these suggestions that CBT should be considered as the treatment of choice for bulimia nervosa, there is a substantial amount of evidence in the literature that calls such claims into question. First, there is comparatively little evidence that CBT in its existing form is better than other therapies (rather than simply eective in its own right). Meta-analytic studies have consistently shown that the overall eects of dierent psychotherapeutic approaches to bulimia nervosa (including individual and group CBT) are very similar (e.g. Fettes & Peters, 1992; Hartmann, Herzog & Drinkmann, 1992). Few studies directly compare outcomes across psychotherapies. However, in the most rigorous of such studies, Fairburn et al. (1995) have shown that interpersonal therapy is at least as eective as CBT at long-term follow-up. Similarly, Cooper, Cooper and Hill (1989) conclude that adding a cognitive component to a behavioral intervention programme for bulimic individuals does not substantially improve outcome. The second reason for doubting the claim that CBT (as currently practiced) should be considered as the universal ®rst line of treatment is that it is known to have limited success with particular groups of bulimics, particularly those with comorbid borderline personality disorder (e.g. Johnson, Tobin & Dennis, 1990; Sansone & Fine, 1992). This limitation is an important one, given the high incidence of borderline personality disorder amongst bulimics (e.g. Skodol et al., 1993; Braun, Sunday & Halmi, 1994; Wonderlich, 1995; Carroll, Touyz & Beumont, 1996). The failure of CBT in some instances might be explained by its cognitive focus. The essence of existing cognitive behavioral models (e.g. Fairburn & Cooper, 1989; Vitousek, 1996; Cooper, 1997; Fairburn, 1997) is that bulimic symptoms are precipitated and maintained by a set of maladaptive thinking patterns (negative automatic thoughts and dysfunctional assumptions) regarding body weight, size and shape. Within this model, the individual typically overvalues slimness, usually as a means of restoring self-esteem. However, there is evidence to suggest the presence of a broader, more general dysfunctional thinking style in bulimics, rather than just maladaptive beliefs regarding food, weight and shape (e.g. Phillips, Tiggemann & Wade, 1997). Apart from their content, the key element of these other cognitions is that they can be seen as unconditional, re¯ecting core beliefs rather than negative automatic thoughts or dysfunctional assumptions. Relevant patterns of core belief include: insucient self-control (e.g. Newton, Freeman & Munro, 1993), vulnerability to threat and harm (Root & Fallon, 1989; Waller & Ruddock, 1995; McManus, Waller & Chadwick, 1996) and shame (Andrews, 1997). Such core beliefs, along with others, have been considered by Young (1994) in his work with individuals with personality disorders. Young's schema-focused model was developed to explore this deepest level of cognitive representation (which he terms `early maladaptive schemas'), as it was recognized that people with personality or characterological problems might not bene®t from conventional cognitive therapy, with its focus on primary and
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
147
secondary cognitions (i.e. automatic thoughts and underlying dysfunctional assumptions). Waller, Ohanian, Meyer and Osman (in press) have shown that the Schema Questionnaire (YSQ) of Young (1994) is a useful tool in understanding the psychopathology of bulimic disorders, at both the dimensional and the categorical level. Given CBT's failure to induce change in a substantial minority of bulimics (particularly those with personality disorder pathology), it is potentially valuable to explore the role of core beliefs in the ecacy of treatment for bulimic psychopathology. Such an understanding would demonstrate those core beliefs that might be addressed in order to enhance the value of existing models of CBT. Therefore, the aim of this study was to investigate the role of core beliefs (as conceptualized by Young, 1994) in the outcome of group CBT for bulimia nervosa. First, based on the existing literature (e.g. Fettes & Peters, 1992), it was hypothesized that group CBT would be eective in the treatment of bulimia nervosa. Second, it was hypothesized Table 1 Comparison of initial psychopathology of the women who completed group CBT (N = 20) and those who dropped out (N = 7) Group Completers mean Age Bulimic symptomatology Bulimia Test (BULIT-R) Mizes Anorexic Cognitions (MAC) Frequency of bingeing (per week) Frequency of vomiting YSQ scales Abandonment Functional dependence/incompetence Defectiveness/shame Emotional deprivation Emotion inhibition Enmeshment Entitlement Failure to achieve Insucient self-control/discipline Mistrust/abuse Subjugation Social isolation Self sacri®ce Social undesirability Unrelenting standards Vulnerability to harm
Drop-outs S.D.
mean
Mann±Whitney S.D.
z
p
26.2
5.57
23.0
3.34
0.92
NS
117.7 115.6 5.6 12.6
8.71 19.3 4.8 7.2
116.3 116.9 7.2 8.6
7.65 13.9 6.4 5.5
0.50 0.28 0.73 1.42
NS NS NS NS
1.04 0.57 1.10 1.11 0.67 0.40 0.45 1.27 0.55 0.60 0.92 0.99 0.65 1.03 0.89 0.54
0.75 1.55 0.44 0.78 1.80 1.50 0.17 1.47 1.16 1.00 1.66 0.06 1.55 1.55 0.30 1.28
NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS
3.41 2.91 3.45 3.87 3.29 2.38 2.96 3.33 3.49 3.97 3.34 3.43 3.78 3.44 4.30 3.12
1.12 1.05 1.12 1.33 1.03 0.88 0.88 1.41 0.95 1.08 1.19 1.25 0.76 1.24 0.93 0.95
3.75 3.54 3.78 3.46 3.78 3.00 2.86 4.06 3.17 3.55 4.09 3.50 4.24 4.29 4.48 3.53
148
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
that less healthy levels of core beliefs would be associated with a poorer outcome from such treatment.
2. Method 2.1. Design This study used a pre±post design, where a sample of 20 bulimics completed measures of eating pathology before and after treatment (group CBT for bulimia nervosa). Pretreatment levels of core beliefs were used to predict the degree of change of eating attitudes and behaviors. 2.2. Participants The participants in the study were 20 female outpatients, who were consecutive referrals to an eating disorders service. Each had a diagnosis of bulimia nervosa (DSM-IV; American Psychiatric Association, 1994) and agreed to take part in group treatment. Their mean age was 26.2 years (range=18±39 years; S.D.=5.57). Four CBT groups were run consecutively, starting with a total of 27 patients across the groups. Thus, there were a further seven patients who failed to complete the treatment groups, divided relatively evenly across those groups (two from the seven patients in the ®rst group, one each from the seven patients in the second and fourth groups and three from the six patients in the third group). Preliminary analysis showed that there were no signi®cant dierences on any measure between those who completed treatment and those who dropped out (see Table 1). However, it is worthy of note that the drop-outs tended to have higher levels of vomiting than those who completed the group, possibly supporting the need for a close `®t' between group aims and patients' goals (e.g. McKisack & Waller, 1997). Kruskal±Wallis tests showed no dierences between the four groups' pretreatment levels of bulimic symptoms or core beliefs when the whole group (27 women) were considered ( p > 0.05 in all cases). When comparing the 20 women who remained in treatment throughout, an almost identical pattern emerged. The one exception was that the fourth group had lower levels of subjugation belief than the ®rst group ( p = 0.04). This overall similarity across groups suggests that any role of core beliefs in predicting change is a general one (i.e. not a product of a particular subset of the sample). The dierence in subjugation beliefs is of minimal importance, as this variable did not prove to be important in the multivariate analyses. 2.3. Measures and procedure Each patient completed three self-report questionnaires, as follows. The ®rst two were completed at the beginning of treatment and immediately after the ®nal group session. The third was completed only at the beginning of treatment. Each patients also recorded the frequency of bingeing and vomiting before and after the group CBT.
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
149
2.3.1. Bulimia Test Ð Revised (BULIT-R; Thelen, Farmer, Wonderlich & Smith, 1991) The BULIT-R consists of 36 items designed to measure bulimic symptoms, including estimated frequency of binge-eating and purging behaviors, negative aect and weight ¯uctuations. Each item is answered on a ®ve-point Likert scale. Higher scores (range=28±140) indicate greater levels of bulimic psychopathology. 2.3.2. Mizes Anorectic Cognition scale (MAC; Mizes & Klesges, 1989) The MAC is a 33-item questionnaire, measuring cognitions associated with anorexic and bulimic behaviors. Each item is answered on a ®ve-point Likert scale. Higher scores (range=33±165) indicate greater levels of cognitions regarding weight and eating regulation, and their connection to self-esteem, self-control and approval from others. 2.3.3. Young Schema Questionnaire (YSQ; Young, 1994) The YSQ is a 205-item questionnaire, developed to measure 16 core beliefs. Schmidt, Joiner, Young and Telch (1995) have demonstrated that the YSQ has good levels of psychometric and clinical utility. The items are answered on a six-point Likert scale. Higher item mean scores (range=1±6) re¯ect a more unhealthy level of core beliefs. The 16 schemas assessed by the YSQ are: abandonment (the belief that close relationships will end imminently), functional dependence/incompetence (the belief that one is not competent and cannot be independent), defectiveness/shame (the belief that one is internally ¯awed), emotional deprivation (the belief that one's emotional needs will never be met), emotional inhibition (the belief that emotions must be inhibited to avoid adverse consequences), enmeshment (the lack of individual identity, due to emotional overinvolvement with others), entitlement (the belief that one can act without consideration for others), failure to achieve (the belief that one is incapable of performing well), insucient self-control/self discipline (the belief that one cannot control one's impulses or feelings), mistrust/abuse (the belief that one will be taken advantage of by others), subjugation (the belief that one must submit to the control of others to avoid negative consequences), social isolation (the belief that one is dierent and isolated from the world), self-sacri®ce (the belief that one must sacri®ce one's own needs to help to satisfy others' needs), social undesirability (the belief that one is unattractive to and disliked by others), unrelenting standards (the belief that one should strive for unrealistic standards) and vulnerability to harm and illness (the belief that one has no control over the threat of disasters). 2.3.4. Frequency of bulimic behaviors Frequency of bingeing and self-induced vomiting were also recorded before and after the group CBT, using diary measures. A binge was de®ned in terms of both quantity of food and the experience of loss of control (American Psychiatric Association, 1994). 2.4. Treatment Each of the four CBT groups consisted of 12 weekly sessions, and was run by the same two therapists. The treatment was based on an existing cognitive behavioral model of the aetiology and treatment of bulimia nervosa (Fairburn, Cooper & Cooper, 1986; Fairburn, Marcus & Wilson, 1993). The main aims of the group were: to increase patients' knowledge about
150
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
bulimia nervosa; to teach them behavioral techniques to overcome the bingeing and purging behaviors; to equip them with cognitive skills to challenge their maladaptive thoughts surrounding food, weight and size and to provide an opportunity for them to gain support and mutual understanding from one another. The weekly themes for the group were: principles of normal eating and breaking the binge-purge cycle (session 1), behavioral techniques to overcome bingeing and purging (session 2), physical and psychological eects of bulimia (session 3), thinking errors, changing the way you think and looking for rational answers (sessions 4 and 5), dietary advice (session 6), relaxation training (sessions 7 and 8), body image and its distortion (session 9), assertiveness (sessions 10 and 11) and forward planning and relapse prevention (session 12). 2.5. Data analysis Due to the non-normal distribution of some of the scores, nonparametric tests were used in the early analyses. The results were essentially identical to those obtained when using parametric tests. Where no nonparametric equivalent was available (i.e. when multivariate methods were used), parametric analyses were used, but the results of those analyses will be presented with the appropriate caution. Wilcoxon tests were used to measure the change in eating symptomatology across treatment. Spearman's correlation coecients were calculated to test for associations of pretreatment core beliefs with changes in eating psychopathology. Multiple regression analyses were then used to determine whether these associations were a product of the pretreatment core beliefs or of the pretreatment level of the relevant symptom. Finally, multiple regression analyses were used to partial out the intercorrelation of the YSQ scales, in order to determine the most parsimonious models of associations between pretreatment core beliefs (and symptoms) and changes in symptomatology after CBT. Where there were directional hypotheses, one-tailed tests were used. 3. Results 3.1. Eectiveness of group CBT in reducing eating psychopathology Table 2 shows the women's levels of eating psychopathology before and after group CBT. Table 2 Changes in eating psychopathology following 12-session group CBT (N = 20)
Bulimia Test (BULIT-R) Mizes Anorexic Cognitions (MAC) Frequency of bingeing (per week) Frequency of vomiting (per week)
Pretreatment
Posttreatment
Wilcoxon test
mean
S.D.
mean
S.D.
z
p
117.7 115.6 5.6 12.6
8.7 19.3 4.8 7.2
93.5 100.3 2.3 5.9
20.4 21.1 2.8 6.1
3.58 3.45 3.36 3.82
0.001 0.001 0.001 0.001
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
151
The mean pretreatment scores on the BULIT-R and MAC are consistent with the published norms for bulimic individuals (Mizes & Klesges, 1989; Thelen et al., 1991). There was a signi®cant decrease in all aspects of eating psychopathology over the course of the group CBT. As might be expected from a group treatment (Freeman, 1995), the reduction in symptoms was not as great as if the CBT had been delivered individually (Vitousek, 1996). However, the reduction in binge frequency (59%) and vomiting (53%) was high, suggesting that the treatment had been relatively eective. 3.2. Association of core beliefs with changes in eating behaviors and cognitions Table 3 shows the association of changes in eating psychopathology with pretreatment core beliefs (YSQ scores). Change on the BULIT-R (re¯ecting more bulimic attitudes) was negatively correlated with 10 of the 16 YSQ scales (unhealthy pretreatment core beliefs predicted less change in bulimic attitudes). In contrast, only one of the YSQ scales (`emotional deprivation') was reliably associated with change on the MAC, and the correlation was in an unexpected direction (i.e. higher levels of pretreatment `emotional deprivation' belief were linked with greater changes in anorectic cognitions). Change in the frequency of vomiting was negatively correlated with scores on four YSQ scales (less healthy `abandonment', `defectiveness/shame', `social isolation' and `social undesirability' beliefs predicted less of a
Table 3 Associations (Spearman's rho) between levels of core belief (YSQ scores) and changes in eating psychopathology following group CBT. (p < 0.05; p < 0.01) Changes in eating psychopathology
YSQ scale Abandonment Dependence/incompetence Defectiveness/shame Emotional deprivation Emotional inhibition Enmeshment Entitlement Failure to achieve Insucient self control Mistrust/abuse Subjugation Social isolation Self sacri®ce Social undesirability Unrelenting standards Vulnerability to harm
BULIT-R
MAC
Bingeing
Vomiting
ÿ0.53 ÿ0.73 ÿ0.45 0.18 ÿ0.38 ÿ0.54 ÿ0.11 ÿ0.70 ÿ0.19 0.22 ÿ0.58 ÿ0.28 ÿ0.53 ÿ0.64 ÿ0.55 ÿ0.60
0.15 ÿ0.20 0.07 0.86 0.10 0.12 ÿ0.06 0.02 0.09 0.17 0.20 0.39 0.16 ÿ0.03 ÿ0.07 ÿ0.21
ÿ0.22 ÿ0.16 ÿ0.32 ÿ0.10 ÿ0.27 ÿ0.06 ÿ0.15 ÿ0.17 ÿ0.22 ÿ0.17 ÿ0.13 ÿ0.34 ÿ0.09 ÿ0.46 ÿ0.04 ÿ0.19
ÿ0.46 ÿ0.30 ÿ0.64 ÿ0.10 ÿ0.43 ÿ0.07 ÿ0.23 ÿ0.40 ÿ0.42 ÿ0.31 ÿ0.35 ÿ0.49 ÿ0.22 ÿ0.60 ÿ0.11 ÿ0.15
152
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
reduction in vomiting). Finally, the women's reduction in frequency of bingeing was lower if they had high pretreatment `social undesirability' beliefs. While these results appear to indicate that pretreatment core beliefs are associated with the outcome of treatment on all four measures, it is important to account for the impact of pretreatment symptomatology, since such pathology is also likely to be related to core beliefs. In other words, do the associations between core beliefs and treatment change hold up when one also considers the predictive power of pretreatment symptomatology? Multiple regression analyses were used to test this possibility. For each case where there was a signi®cant correlation between a pretreatment core belief and the level of change in psychopathology Table 3, the pretreatment level of that symptom and the core belief were entered simultaneously as independent variables, predicting levels of change in the symptom. First, associations of core beliefs with changes in bulimic behaviors (bingeing and vomiting) were considered. In the case of the link between Social undesirability beliefs and binge frequency, there was a signi®cant overall eect (F = 70.7; p < 0.0001) of the two independent variables, but this was a product of a positive association of greater change in bingeing with pretreatment binge frequency (t = 10.6, p < 0.001), rather than any eect of the Social undesirability belief (t = 0.59; NS). When considering the impact of core beliefs and pretreatment symptom levels upon changes in vomiting, all of the overall eects were signi®cant (F > 11.2, p < 0.001 in all cases). However, the patterns of signi®cant predictors diered across analyses. In the case of the association with Abandonment beliefs, a higher pretreatment level of vomiting was a signi®cant predictor of greater reduction in symptoms (t = 10.6, p < 0.001), but there was no signi®cant eect of Abandonment beliefs (t = 1.33; NS). In contrast, a greater degree of change in vomiting was predicted by both high pretreatment levels of vomiting (t = 4.81, p < 0.001) and low levels of Defectiveness/shame beliefs (t = 3.29, p < 0.004). Similarly, a greater reduction in vomiting was predicted by both high pretreatment levels of vomiting (t = 4.87, p < 0.001) and low levels of Social isolation beliefs (t = 2.54, p < 0.02). Such a change was also predicted by both high pretreatment levels of vomiting (t = 4.50, p < 0.001) and low levels of Social undesirability beliefs (t = 2.57, p < 0.02). Next, the associations of core beliefs with changes in eating attitudes (MAC and BULIT-R) were examined. The overall association of Emotional deprivation beliefs and pretreatment MAC scores with changes in MAC scores was signi®cant (F = 16.1; p < 0.0001), but was due to a relationship between greater change and high Emotional deprivation beliefs (t = 5.59, p < 0.001) rather than pretreatment MAC scores (t = 0.60, NS). The overall association of the pairs of variables with BULIT-R scores was signi®cant in all cases (F > 2.54, p R 0.05, in all cases). In each individual analysis, the eect of pretreatment BULIT-R scores was nonsigni®cant (t < 1.40, NS, in all cases), whereas the eect of the YSQ scale was signi®cant (t > 2.20, p < 0.05, in all cases). To summarize, more unhealthy YSQ Defectiveness/shame, Social isolation and Social undesirability core beliefs were robust predictors of a subsequent failure to reduce frequency of vomiting. In the case of eating attitudes (MAC and BULIT-R scores), all of the associations with pretreatment core beliefs (shown in Table 3) were robust. However, the apparent impact of YSQ Social undesirability beliefs upon change in binge frequency was not robust when the eect of pretreatment pathology was taken into account.
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
153
3.3. Developing a parsimonious model of predictors of bulimic behaviors and attitudes Multiple regression analyses were used to determine the most parsimonious set of the core beliefs that predicted change in bulimic attitudes (BULIT-R) and frequency of vomiting. (No such analysis was conducted for the other two indices of therapeutic gain, since each was predicted by a single YSQ scale). Due to the number of variables relative to the number of patients, YSQ scales were used only if they correlated strongly (r r0.60) with changes in symptomatology. As before, the impact of pretreatment level of symptomatology was also entered into the analysis as an independent variable. The ®rst regression analysis showed that changes in BULIT-R were reliably predicted by the YSQ scale scores and pretreatment pathology (overall F = 4.62; p < 0.02; explained variance=43%). However, the only signi®cant individual predictor was a negative eect of Functional dependence/ incompetence beliefs (t = 2.00; p < 0.05). Change in frequency of vomiting was also reliably predicted by the YSQ scale scores and pretreatment levels of vomiting (overall F = 13.1; p < 0.01; explained variance=66%). This association was a product of signi®cant individual eects of Defectiveness/shame (t = 1.75; p < 0.05) and pretreatment vomiting (t = 4.64; p < 0.01).
4. Discussion Each of the two hypotheses was supported. First, group cognitive behavioral therapy was eective in treating bulimia nervosa, and the size of the treatment eect was comparable with that obtained in other controlled and uncontrolled CBT trials (e.g. Garner, 1987; Mitchell et al., 1990; Agras et al., 1992; Fettes & Peters, 1992). Second, even after the impact of pretreatment symptoms was taken into account, pretreatment core beliefs were associated with the degree of change in bulimic psychopathology (vomiting, bulimic attitudes, restrictive attitudes). This con®rms that such cognitive representations constitute a further factor that may need to be taken into account when considering the ecacy of group or individual CBT in such cases (e.g. Baell & Wertheim, 1992; McKisack & Waller, 1997). With one exception, the associations between core beliefs and the change in eating psychopathology were in the expected direction Ð less healthy pretreatment core beliefs were followed by lower gains. A failure to show an improvement in general bulimic attitudes (BULIT-R scores) was related to a high level of unconditional `functional dependence/ incompetence' beliefs, as suggested by previous researchers (e.g. Garner et al., 1990; Baell & Wertheim, 1992). These ®ndings are broadly in line with earlier suggestions (e.g. Fairburn, Kirk, O'Connor, Anastasiades & Cooper, 1987) that pretreatment level of self-esteem can reliably predict treatment outcome. In contrast, the strongest cognitive predictor of maintained vomiting was a high level of unconditional `defectiveness/shame' beliefs Ð the YSQ variable that has the strongest association with levels of purging (Waller et al., in press). This ®nding is compatible with earlier models, which suggest that purging reduces awareness of intolerable cognitions about oneself (Pitts & Waller, 1993) or removes the problem from within oneself (Schupak-Neuberg & Nemero, 1993).
154
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
The one ®nding that ran counter to prediction was that unconditional `emotional deprivation' beliefs were strongly correlated with a positive change in anorectic cognitions. In other words, restrictive attitudes and cognitions were likely to be more reduced if the women believed pretreatment that their emotional needs would never be met. It can be hypothesized that this bene®t was a function of the group format, challenging the emotional deprivation belief in vivo by providing an emotionally supportive environment. Such an environment might have the value of encouraging the individual with such beliefs to participate more fully in the therapy, thus leading to greater gains. In contrast, the individual with lower emotional deprivation beliefs at the outset might ®nd the group format less supportive (and hence a less safe environment for change). In order to test this hypothesis, it would be necessary to determine whether this positive association still holds in individual therapy. At a general level, these ®ndings support the clinical suggestion that existing models of CBT (with their emphasis on negative automatic thoughts and dysfunctional assumptions) are more appropriate for some bulimic patients than for others. It could be argued that CBT for bulimia nervosa would be more eective if it included a schema-focused component (e.g. Padesky, 1994; Young, 1994), addressing core beliefs that are unrelated to food, shape and weight. Alternatively, it might be more ecient if dierent therapeutic `packages' were designed for dierent bulimic groups, where the level of unhealthy core beliefs played a role in determining the targeting of therapy. Since schema-focused CBT is usually more consuming of time and resources than routine CBT (e.g. Fairburn & Cooper, 1989; Young, 1994; Fairburn, 1997), it will be important to develop appropriate heuristics for making such decisions. At the diagnostic level, the presence of a comorbid personality disorder or a history of trauma might be a useful initial determinant of who should receive schema-focused CBT, but a more re®ned use of psychometrics (such as the YSQ) might assist in the task of assigning bulimic patients to the most appropriate treatment. These preliminary ®ndings suggest that the clinician might consider addressing the core beliefs of defectiveness/shame and functional dependence/ incompetence in order to enhance CBT's eectiveness at reducing vomiting behaviors and bulimic attitudes.
References Agras, W. S., Rossiter, E. M., Arnow, B., Schneider, J., Telch, C. F., Raeburn, S. D., Bruce, B., Perl, M., & Koran, L. M. (1992). Pharmacological and cognitive-behavioral treatment for bulimia nervosa: a controlled comparison. American Journal of Psychiatry, 149, 82±87. American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: American Psychiatric Association. Andrews, B. (1997). Bodily shame in relation to abuse in childhood and bulimia: a preliminary investigation. British Journal of Clinical Psychology, 36, 41±49. Baell, W. K., & Wertheim, E. H. (1992). Predictors of outcome in the treatment of bulimia nervosa. British Journal of Clinical Psychology, 31, 330±332. Braun, D. L., Sunday, S. R., & Halmi, K. A. (1994). Psychiatric comorbidity in patients with eating disorders. Psychological Medicine, 24, 859±867. Carroll, J. M., Touyz, S. W., & Beumont, P. J. V. (1996). Speci®c comorbidity between bulimia nervosa and personality disorders. International Journal of Eating Disorders, 19, 159±170.
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
155
Cooper, M. (1997). Cognitive theory in anorexia nervosa and bulimia nervosa: a review. Behavioral and Cognitive Psychotherapy, 25, 113±145. Cooper, P. J., Cooper, Z., & Hill, C. (1989). Behavioral treatment of bulimia nervosa. International Journal of Eating Disorders, 8, 87±92. Craighead, L. W., & Agras, W. S. (1991). Mechanisms of action in cognitive-behavioral and pharmacological interventions of obesity and bulimia nervosa. Journal of Consulting and Clinical Psychology, 59, 115±125. Fairburn, C. G. (1988). The current status of psychological treatments for bulimia nervosa. Journal of Psychosomatic Research, 32, 635±645. Fairburn, C. G. (1997). Eating disorders. In D. M. Clark, & C. G. Fairburn, Science and practice of cognitive behavior therapy. Oxford: Oxford University Press. Fairburn, C. G., & Cooper, P. (1989). Eating disorders. In K. Hawton, P. M. Salkovskis, J. Kirk, & D. M. Clark, Cognitive behavior therapy for psychiatric problems. Oxford: Oxford University Press. Fairburn, C. G., Cooper, Z., & Cooper, P. J. (1986). In K. D. Brownell, & J. P. Foreyt, Handbook of eating disorders: physiology, psychology and treatment of obesity, anorexia and bulimia (pp. 389±404). New York: Basic Books. Fairburn, C. G., Kirk, J., O'Connor, M., Anastasiades, P., & Cooper, P. J. (1987). Prognostic factors in bulimia nervosa. British Journal of Clinical Psychology, 26, 223±224. Fairburn, C. G., Marcus, M. D., & Wilson, G. T. (1993). Cognitive-behavioral therapy for binge eating and bulimia nervosa: a comprehensive treatment manual. In C. G. Fairburn, & G. T. Wilson, Binge eating: nature, assessment and treatment (pp. 361±404). New York: Guilford. Fairburn, C. G., Norman, P. A., Welch, S. L., O'Connor, M. E., Doll, H. A., & Peveler, R. C. (1995). A prospective study of outcome in bulimia nervosa and the long term eects of three psychological treatments. Archives of General Psychiatry, 52, 304±312. Fettes, P. A., & Peters, J. M. (1992). A meta-analysis of group treatments for bulimia nervosa. International Journal of Eating Disorders, 11, 97±110. Freeman, C. (1995). Cognitive therapy. In G. Szmukler, C. Dare, & J. Treasure, Handbook of eating disorders: theory, treatment and research. New York: Wiley. Garner, D. M. (1987). Psychotherapy outcome research with bulimia nervosa. Psychotherapy and Psychosomatics, 48, 129±140. Garner, D. M., Olmsted, M. P., Davies, R., Rockert, W., Goldbloom, D., & Eagle, M. (1990). The association between bulimic symptoms and reported psychopathology. International Journal of Eating Disorders, 9, 1±15. Hartmann, A., Herzog, T., & Drinkmann, D. (1992). Psychotherapy for bulimic nervosa: what is eective? A meta analysis. Journal of Psychosomatic Research, 36, 159±167. Herzog, D. B., Keller, M. B., Strober, M., Yeh, C., & Pai, S.-Y. (1992). The current status of treatment for anorexia nervosa and bulimia nervosa. International Journal of Eating Disorders, 12, 215±220. Johnson, C., Tobin, D., & Dennis, A. (1990). Dierence in treatment outcome between borderline and non-borderline bulimics at one-year follow-up. International Journal of Eating Disorders, 9, 617±627. McKisack, C., & Waller, G. (1997). Factors in¯uencing the outcome of group psychotherapy for bulimia nervosa. International Journal of Eating Disorders, 22, 1±13. McManus, F., Waller, G., & Chadwick, P. (1996). Biases in the processing of dierent forms of threat in bulimic and comparison women. Journal of Nervous and Mental Disease, 184, 547±554. Mitchell, J. E., Pyle, R. L., Eckert, E. D., Hatsukami, D., Pomeroy, C., & Zimmerman, K. (1990). A comparison of antidepressants and structured group therapy in the treatment of bulimia nervosa. Archives of General Psychiatry, 47, 149±157. Mizes, J. S., & Klesges, R. C. (1989). Validity, reliability and factor structure of the anorectic cognitions questionnaire. Addictive Behaviors, 14, 589±594. Newton, J. R., Freeman, C. P., & Munro, J. (1993). Impulsivity and dyscontrol in bulimia nervosa: is impulsivity an independent phenomenon or a marker of severity? Acta Psychiatrica Scandinavica, 87, 389±394. Padesky, C. A. (1994). Schema change processes in cognitive therapy. Clinical Psychology and Psychotherapy, 1, 267±278. Phillips, L., Tiggemann, M., & Wade, T. (1997). Comparison of cognitive style in bulimia nervosa and depression. Behavior Research and Therapy, 35, 939±948.
156
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
Pitts, C., & Waller, G. (1993). Self-denigratory beliefs following sexual abuse: association with the symptomatology of bulimic disorders. International Journal of Eating Disorders, 13, 407±410. Root, M. P. P., & Fallon, P. (1989). Treating the victimized bulimic. Journal of Interpersonal Violence, 4, 90±100. Sansone, R. A., & Fine, M. A. (1992). Borderline personality disorder as a predictor of outcome in women with eating disorders. Journal of Personality Disorders, 6, 176±186. Schmidt, N. B., Joiner, T. E., Young, J. E., & Telch, M. J. (1995). The Schema Questionnaire: investigation of psychometric properties and the hierarchical structure of a measure of maladaptive schemas. Cognitive Therapy and Research, 19, 295±321. Schupak-Neuberg, E., & Nemero, C. J. (1993). Disturbances in identity and self-regulation in bulimia nervosa: implications for a metaphorical perspective of `body as self'. International Journal of Eating Disorders, 13, 335±347. Skodol, A. E., Oldham, J. M., Hyler, S. E., Kellman, H. D., Doidge, N., & Davies, M. (1993). Comorbidity of DSM-III-R eating disorders and personality disorders. International Journal of Eating Disorders, 14, 403±416. Thelen, M. H., Farmer, J., Wonderlich, S., & Smith, M. (1991). A revision of the Bulimia Test: the BULIT-R. Psychological Assessment, 3, 119±124. Vitousek, K. M. (1996). The current status of cognitive-behavioral models of anorexia nervosa and bulimia nervosa. In P. M. Salkovskis, Frontiers of cognitive therapy. New York: Guilford. Waller, G., Ohanian, V. Meyer, C., & Osman, S. (in press). Cognitive content among bulimic women: the role of core beliefs. International Journal of Eating Disorders. Waller, G., & Ruddock, A. (1995). Information-processing correlates of reported sexual abuse in eating-disordered and comparison women. Child Abuse and Neglect, 19, 745±759. Wilson, G. T., & Fairburn, C. G. (1993). Cognitive treatment for the eating disorders. Journal of Consulting and Clinical Psychology, 61, 261±269. Wonderlich, S. A. (1995). Personality and eating disorders. In K. D. Brownell, & C. G. Fairburn, Eating disorders and obesity: a comprehensive handbook. London: Guilford. Young, J. E. (1994). Cognitive therapy for personality disorders: a schema-focused approach (2nd ed.). Sarasota: Professional Resource Press.
www.elsevier.com/locate/brat
Outcome of group cognitive-behavior therapy for bulimia nervosa: the role of core beliefs Newman Leung a,b,*, Glenn Waller c, Glyn Thomas b a
Eating Disorders Service, Queen Elizabeth Psychiatric Hospital, Mindelsohn Way, Edgbaston, Birmingham B15 2QZ, UK b School of Psychology, University of Birmingham, Birmingham, UK c Department of Psychology, University of Southampton, Southampton, UK Received 3 December 1998; accepted 3 December 1998
Abstract It is hypothesized that individuals who bene®t less from CBT will be those who have more pathological core beliefs (unconditional beliefs, unrelated to food, shape and weight). Twenty bulimic women were treated using 12 sessions of conventional group CBT. Eating behavior and attitudes were assessed pre- and posttreatment. Core beliefs were assessed at the beginning of the programme, and were used as predictors of change across treatment (once any eect of pretreatment psychopathology was taken into account). Group CBT was eective, with reductions of over 50% in bulimic symptoms. Outcome on most indices was associated with pretreatment levels of pathological core beliefs. Possible reasons for these ®ndings are discussed. # 2000 Elsevier Science Ltd. All rights reserved. Keywords: Bulimia nervosa; Cognitive-behavioral therapy; Core beliefs
1. Introduction Cognitive behavioral therapy (CBT) is generally considered to be the treatment of choice for bulimia nervosa (e.g. Fairburn, 1988; Craighead & Agras, 1991; Wilson & Fairburn, 1993). For example, in a survey of clinicians, Herzog, Keller, Strober, Yeh and Pai (1992) found that 85± 94% of respondents indicated that they would consider using cognitive behavioral therapy alone or in conjunction with other approaches in the treatment of bulimia nervosa. This strong
* Corresponding author. 0005-7967/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 5 - 7 9 6 7 ( 9 9 ) 0 0 0 2 6 - 1
146
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
preference of clinicians is supported by empirical ®ndings, which indicate a high ecacy for CBT. In a review of its eectiveness, Vitousek (1996) concluded that individual CBT typically produces an 80% reduction of bingeing and purging symptoms, and a total elimination of bulimic episodes in 50% of treated clients. In addition to its use in individual treatment, CBT has been used in group settings. Group CBT has been demonstrated to be highly cost-eective (e.g. Garner, 1987; Mitchell et al., 1990; Agras et al., 1992; Fettes & Peters, 1992), although the degree of reduction in bulimic symptomatology may be more modest than with individual treatment (Freeman, 1995). Despite these suggestions that CBT should be considered as the treatment of choice for bulimia nervosa, there is a substantial amount of evidence in the literature that calls such claims into question. First, there is comparatively little evidence that CBT in its existing form is better than other therapies (rather than simply eective in its own right). Meta-analytic studies have consistently shown that the overall eects of dierent psychotherapeutic approaches to bulimia nervosa (including individual and group CBT) are very similar (e.g. Fettes & Peters, 1992; Hartmann, Herzog & Drinkmann, 1992). Few studies directly compare outcomes across psychotherapies. However, in the most rigorous of such studies, Fairburn et al. (1995) have shown that interpersonal therapy is at least as eective as CBT at long-term follow-up. Similarly, Cooper, Cooper and Hill (1989) conclude that adding a cognitive component to a behavioral intervention programme for bulimic individuals does not substantially improve outcome. The second reason for doubting the claim that CBT (as currently practiced) should be considered as the universal ®rst line of treatment is that it is known to have limited success with particular groups of bulimics, particularly those with comorbid borderline personality disorder (e.g. Johnson, Tobin & Dennis, 1990; Sansone & Fine, 1992). This limitation is an important one, given the high incidence of borderline personality disorder amongst bulimics (e.g. Skodol et al., 1993; Braun, Sunday & Halmi, 1994; Wonderlich, 1995; Carroll, Touyz & Beumont, 1996). The failure of CBT in some instances might be explained by its cognitive focus. The essence of existing cognitive behavioral models (e.g. Fairburn & Cooper, 1989; Vitousek, 1996; Cooper, 1997; Fairburn, 1997) is that bulimic symptoms are precipitated and maintained by a set of maladaptive thinking patterns (negative automatic thoughts and dysfunctional assumptions) regarding body weight, size and shape. Within this model, the individual typically overvalues slimness, usually as a means of restoring self-esteem. However, there is evidence to suggest the presence of a broader, more general dysfunctional thinking style in bulimics, rather than just maladaptive beliefs regarding food, weight and shape (e.g. Phillips, Tiggemann & Wade, 1997). Apart from their content, the key element of these other cognitions is that they can be seen as unconditional, re¯ecting core beliefs rather than negative automatic thoughts or dysfunctional assumptions. Relevant patterns of core belief include: insucient self-control (e.g. Newton, Freeman & Munro, 1993), vulnerability to threat and harm (Root & Fallon, 1989; Waller & Ruddock, 1995; McManus, Waller & Chadwick, 1996) and shame (Andrews, 1997). Such core beliefs, along with others, have been considered by Young (1994) in his work with individuals with personality disorders. Young's schema-focused model was developed to explore this deepest level of cognitive representation (which he terms `early maladaptive schemas'), as it was recognized that people with personality or characterological problems might not bene®t from conventional cognitive therapy, with its focus on primary and
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
147
secondary cognitions (i.e. automatic thoughts and underlying dysfunctional assumptions). Waller, Ohanian, Meyer and Osman (in press) have shown that the Schema Questionnaire (YSQ) of Young (1994) is a useful tool in understanding the psychopathology of bulimic disorders, at both the dimensional and the categorical level. Given CBT's failure to induce change in a substantial minority of bulimics (particularly those with personality disorder pathology), it is potentially valuable to explore the role of core beliefs in the ecacy of treatment for bulimic psychopathology. Such an understanding would demonstrate those core beliefs that might be addressed in order to enhance the value of existing models of CBT. Therefore, the aim of this study was to investigate the role of core beliefs (as conceptualized by Young, 1994) in the outcome of group CBT for bulimia nervosa. First, based on the existing literature (e.g. Fettes & Peters, 1992), it was hypothesized that group CBT would be eective in the treatment of bulimia nervosa. Second, it was hypothesized Table 1 Comparison of initial psychopathology of the women who completed group CBT (N = 20) and those who dropped out (N = 7) Group Completers mean Age Bulimic symptomatology Bulimia Test (BULIT-R) Mizes Anorexic Cognitions (MAC) Frequency of bingeing (per week) Frequency of vomiting YSQ scales Abandonment Functional dependence/incompetence Defectiveness/shame Emotional deprivation Emotion inhibition Enmeshment Entitlement Failure to achieve Insucient self-control/discipline Mistrust/abuse Subjugation Social isolation Self sacri®ce Social undesirability Unrelenting standards Vulnerability to harm
Drop-outs S.D.
mean
Mann±Whitney S.D.
z
p
26.2
5.57
23.0
3.34
0.92
NS
117.7 115.6 5.6 12.6
8.71 19.3 4.8 7.2
116.3 116.9 7.2 8.6
7.65 13.9 6.4 5.5
0.50 0.28 0.73 1.42
NS NS NS NS
1.04 0.57 1.10 1.11 0.67 0.40 0.45 1.27 0.55 0.60 0.92 0.99 0.65 1.03 0.89 0.54
0.75 1.55 0.44 0.78 1.80 1.50 0.17 1.47 1.16 1.00 1.66 0.06 1.55 1.55 0.30 1.28
NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS
3.41 2.91 3.45 3.87 3.29 2.38 2.96 3.33 3.49 3.97 3.34 3.43 3.78 3.44 4.30 3.12
1.12 1.05 1.12 1.33 1.03 0.88 0.88 1.41 0.95 1.08 1.19 1.25 0.76 1.24 0.93 0.95
3.75 3.54 3.78 3.46 3.78 3.00 2.86 4.06 3.17 3.55 4.09 3.50 4.24 4.29 4.48 3.53
148
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
that less healthy levels of core beliefs would be associated with a poorer outcome from such treatment.
2. Method 2.1. Design This study used a pre±post design, where a sample of 20 bulimics completed measures of eating pathology before and after treatment (group CBT for bulimia nervosa). Pretreatment levels of core beliefs were used to predict the degree of change of eating attitudes and behaviors. 2.2. Participants The participants in the study were 20 female outpatients, who were consecutive referrals to an eating disorders service. Each had a diagnosis of bulimia nervosa (DSM-IV; American Psychiatric Association, 1994) and agreed to take part in group treatment. Their mean age was 26.2 years (range=18±39 years; S.D.=5.57). Four CBT groups were run consecutively, starting with a total of 27 patients across the groups. Thus, there were a further seven patients who failed to complete the treatment groups, divided relatively evenly across those groups (two from the seven patients in the ®rst group, one each from the seven patients in the second and fourth groups and three from the six patients in the third group). Preliminary analysis showed that there were no signi®cant dierences on any measure between those who completed treatment and those who dropped out (see Table 1). However, it is worthy of note that the drop-outs tended to have higher levels of vomiting than those who completed the group, possibly supporting the need for a close `®t' between group aims and patients' goals (e.g. McKisack & Waller, 1997). Kruskal±Wallis tests showed no dierences between the four groups' pretreatment levels of bulimic symptoms or core beliefs when the whole group (27 women) were considered ( p > 0.05 in all cases). When comparing the 20 women who remained in treatment throughout, an almost identical pattern emerged. The one exception was that the fourth group had lower levels of subjugation belief than the ®rst group ( p = 0.04). This overall similarity across groups suggests that any role of core beliefs in predicting change is a general one (i.e. not a product of a particular subset of the sample). The dierence in subjugation beliefs is of minimal importance, as this variable did not prove to be important in the multivariate analyses. 2.3. Measures and procedure Each patient completed three self-report questionnaires, as follows. The ®rst two were completed at the beginning of treatment and immediately after the ®nal group session. The third was completed only at the beginning of treatment. Each patients also recorded the frequency of bingeing and vomiting before and after the group CBT.
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
149
2.3.1. Bulimia Test Ð Revised (BULIT-R; Thelen, Farmer, Wonderlich & Smith, 1991) The BULIT-R consists of 36 items designed to measure bulimic symptoms, including estimated frequency of binge-eating and purging behaviors, negative aect and weight ¯uctuations. Each item is answered on a ®ve-point Likert scale. Higher scores (range=28±140) indicate greater levels of bulimic psychopathology. 2.3.2. Mizes Anorectic Cognition scale (MAC; Mizes & Klesges, 1989) The MAC is a 33-item questionnaire, measuring cognitions associated with anorexic and bulimic behaviors. Each item is answered on a ®ve-point Likert scale. Higher scores (range=33±165) indicate greater levels of cognitions regarding weight and eating regulation, and their connection to self-esteem, self-control and approval from others. 2.3.3. Young Schema Questionnaire (YSQ; Young, 1994) The YSQ is a 205-item questionnaire, developed to measure 16 core beliefs. Schmidt, Joiner, Young and Telch (1995) have demonstrated that the YSQ has good levels of psychometric and clinical utility. The items are answered on a six-point Likert scale. Higher item mean scores (range=1±6) re¯ect a more unhealthy level of core beliefs. The 16 schemas assessed by the YSQ are: abandonment (the belief that close relationships will end imminently), functional dependence/incompetence (the belief that one is not competent and cannot be independent), defectiveness/shame (the belief that one is internally ¯awed), emotional deprivation (the belief that one's emotional needs will never be met), emotional inhibition (the belief that emotions must be inhibited to avoid adverse consequences), enmeshment (the lack of individual identity, due to emotional overinvolvement with others), entitlement (the belief that one can act without consideration for others), failure to achieve (the belief that one is incapable of performing well), insucient self-control/self discipline (the belief that one cannot control one's impulses or feelings), mistrust/abuse (the belief that one will be taken advantage of by others), subjugation (the belief that one must submit to the control of others to avoid negative consequences), social isolation (the belief that one is dierent and isolated from the world), self-sacri®ce (the belief that one must sacri®ce one's own needs to help to satisfy others' needs), social undesirability (the belief that one is unattractive to and disliked by others), unrelenting standards (the belief that one should strive for unrealistic standards) and vulnerability to harm and illness (the belief that one has no control over the threat of disasters). 2.3.4. Frequency of bulimic behaviors Frequency of bingeing and self-induced vomiting were also recorded before and after the group CBT, using diary measures. A binge was de®ned in terms of both quantity of food and the experience of loss of control (American Psychiatric Association, 1994). 2.4. Treatment Each of the four CBT groups consisted of 12 weekly sessions, and was run by the same two therapists. The treatment was based on an existing cognitive behavioral model of the aetiology and treatment of bulimia nervosa (Fairburn, Cooper & Cooper, 1986; Fairburn, Marcus & Wilson, 1993). The main aims of the group were: to increase patients' knowledge about
150
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
bulimia nervosa; to teach them behavioral techniques to overcome the bingeing and purging behaviors; to equip them with cognitive skills to challenge their maladaptive thoughts surrounding food, weight and size and to provide an opportunity for them to gain support and mutual understanding from one another. The weekly themes for the group were: principles of normal eating and breaking the binge-purge cycle (session 1), behavioral techniques to overcome bingeing and purging (session 2), physical and psychological eects of bulimia (session 3), thinking errors, changing the way you think and looking for rational answers (sessions 4 and 5), dietary advice (session 6), relaxation training (sessions 7 and 8), body image and its distortion (session 9), assertiveness (sessions 10 and 11) and forward planning and relapse prevention (session 12). 2.5. Data analysis Due to the non-normal distribution of some of the scores, nonparametric tests were used in the early analyses. The results were essentially identical to those obtained when using parametric tests. Where no nonparametric equivalent was available (i.e. when multivariate methods were used), parametric analyses were used, but the results of those analyses will be presented with the appropriate caution. Wilcoxon tests were used to measure the change in eating symptomatology across treatment. Spearman's correlation coecients were calculated to test for associations of pretreatment core beliefs with changes in eating psychopathology. Multiple regression analyses were then used to determine whether these associations were a product of the pretreatment core beliefs or of the pretreatment level of the relevant symptom. Finally, multiple regression analyses were used to partial out the intercorrelation of the YSQ scales, in order to determine the most parsimonious models of associations between pretreatment core beliefs (and symptoms) and changes in symptomatology after CBT. Where there were directional hypotheses, one-tailed tests were used. 3. Results 3.1. Eectiveness of group CBT in reducing eating psychopathology Table 2 shows the women's levels of eating psychopathology before and after group CBT. Table 2 Changes in eating psychopathology following 12-session group CBT (N = 20)
Bulimia Test (BULIT-R) Mizes Anorexic Cognitions (MAC) Frequency of bingeing (per week) Frequency of vomiting (per week)
Pretreatment
Posttreatment
Wilcoxon test
mean
S.D.
mean
S.D.
z
p
117.7 115.6 5.6 12.6
8.7 19.3 4.8 7.2
93.5 100.3 2.3 5.9
20.4 21.1 2.8 6.1
3.58 3.45 3.36 3.82
0.001 0.001 0.001 0.001
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
151
The mean pretreatment scores on the BULIT-R and MAC are consistent with the published norms for bulimic individuals (Mizes & Klesges, 1989; Thelen et al., 1991). There was a signi®cant decrease in all aspects of eating psychopathology over the course of the group CBT. As might be expected from a group treatment (Freeman, 1995), the reduction in symptoms was not as great as if the CBT had been delivered individually (Vitousek, 1996). However, the reduction in binge frequency (59%) and vomiting (53%) was high, suggesting that the treatment had been relatively eective. 3.2. Association of core beliefs with changes in eating behaviors and cognitions Table 3 shows the association of changes in eating psychopathology with pretreatment core beliefs (YSQ scores). Change on the BULIT-R (re¯ecting more bulimic attitudes) was negatively correlated with 10 of the 16 YSQ scales (unhealthy pretreatment core beliefs predicted less change in bulimic attitudes). In contrast, only one of the YSQ scales (`emotional deprivation') was reliably associated with change on the MAC, and the correlation was in an unexpected direction (i.e. higher levels of pretreatment `emotional deprivation' belief were linked with greater changes in anorectic cognitions). Change in the frequency of vomiting was negatively correlated with scores on four YSQ scales (less healthy `abandonment', `defectiveness/shame', `social isolation' and `social undesirability' beliefs predicted less of a
Table 3 Associations (Spearman's rho) between levels of core belief (YSQ scores) and changes in eating psychopathology following group CBT. (p < 0.05; p < 0.01) Changes in eating psychopathology
YSQ scale Abandonment Dependence/incompetence Defectiveness/shame Emotional deprivation Emotional inhibition Enmeshment Entitlement Failure to achieve Insucient self control Mistrust/abuse Subjugation Social isolation Self sacri®ce Social undesirability Unrelenting standards Vulnerability to harm
BULIT-R
MAC
Bingeing
Vomiting
ÿ0.53 ÿ0.73 ÿ0.45 0.18 ÿ0.38 ÿ0.54 ÿ0.11 ÿ0.70 ÿ0.19 0.22 ÿ0.58 ÿ0.28 ÿ0.53 ÿ0.64 ÿ0.55 ÿ0.60
0.15 ÿ0.20 0.07 0.86 0.10 0.12 ÿ0.06 0.02 0.09 0.17 0.20 0.39 0.16 ÿ0.03 ÿ0.07 ÿ0.21
ÿ0.22 ÿ0.16 ÿ0.32 ÿ0.10 ÿ0.27 ÿ0.06 ÿ0.15 ÿ0.17 ÿ0.22 ÿ0.17 ÿ0.13 ÿ0.34 ÿ0.09 ÿ0.46 ÿ0.04 ÿ0.19
ÿ0.46 ÿ0.30 ÿ0.64 ÿ0.10 ÿ0.43 ÿ0.07 ÿ0.23 ÿ0.40 ÿ0.42 ÿ0.31 ÿ0.35 ÿ0.49 ÿ0.22 ÿ0.60 ÿ0.11 ÿ0.15
152
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
reduction in vomiting). Finally, the women's reduction in frequency of bingeing was lower if they had high pretreatment `social undesirability' beliefs. While these results appear to indicate that pretreatment core beliefs are associated with the outcome of treatment on all four measures, it is important to account for the impact of pretreatment symptomatology, since such pathology is also likely to be related to core beliefs. In other words, do the associations between core beliefs and treatment change hold up when one also considers the predictive power of pretreatment symptomatology? Multiple regression analyses were used to test this possibility. For each case where there was a signi®cant correlation between a pretreatment core belief and the level of change in psychopathology Table 3, the pretreatment level of that symptom and the core belief were entered simultaneously as independent variables, predicting levels of change in the symptom. First, associations of core beliefs with changes in bulimic behaviors (bingeing and vomiting) were considered. In the case of the link between Social undesirability beliefs and binge frequency, there was a signi®cant overall eect (F = 70.7; p < 0.0001) of the two independent variables, but this was a product of a positive association of greater change in bingeing with pretreatment binge frequency (t = 10.6, p < 0.001), rather than any eect of the Social undesirability belief (t = 0.59; NS). When considering the impact of core beliefs and pretreatment symptom levels upon changes in vomiting, all of the overall eects were signi®cant (F > 11.2, p < 0.001 in all cases). However, the patterns of signi®cant predictors diered across analyses. In the case of the association with Abandonment beliefs, a higher pretreatment level of vomiting was a signi®cant predictor of greater reduction in symptoms (t = 10.6, p < 0.001), but there was no signi®cant eect of Abandonment beliefs (t = 1.33; NS). In contrast, a greater degree of change in vomiting was predicted by both high pretreatment levels of vomiting (t = 4.81, p < 0.001) and low levels of Defectiveness/shame beliefs (t = 3.29, p < 0.004). Similarly, a greater reduction in vomiting was predicted by both high pretreatment levels of vomiting (t = 4.87, p < 0.001) and low levels of Social isolation beliefs (t = 2.54, p < 0.02). Such a change was also predicted by both high pretreatment levels of vomiting (t = 4.50, p < 0.001) and low levels of Social undesirability beliefs (t = 2.57, p < 0.02). Next, the associations of core beliefs with changes in eating attitudes (MAC and BULIT-R) were examined. The overall association of Emotional deprivation beliefs and pretreatment MAC scores with changes in MAC scores was signi®cant (F = 16.1; p < 0.0001), but was due to a relationship between greater change and high Emotional deprivation beliefs (t = 5.59, p < 0.001) rather than pretreatment MAC scores (t = 0.60, NS). The overall association of the pairs of variables with BULIT-R scores was signi®cant in all cases (F > 2.54, p R 0.05, in all cases). In each individual analysis, the eect of pretreatment BULIT-R scores was nonsigni®cant (t < 1.40, NS, in all cases), whereas the eect of the YSQ scale was signi®cant (t > 2.20, p < 0.05, in all cases). To summarize, more unhealthy YSQ Defectiveness/shame, Social isolation and Social undesirability core beliefs were robust predictors of a subsequent failure to reduce frequency of vomiting. In the case of eating attitudes (MAC and BULIT-R scores), all of the associations with pretreatment core beliefs (shown in Table 3) were robust. However, the apparent impact of YSQ Social undesirability beliefs upon change in binge frequency was not robust when the eect of pretreatment pathology was taken into account.
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
153
3.3. Developing a parsimonious model of predictors of bulimic behaviors and attitudes Multiple regression analyses were used to determine the most parsimonious set of the core beliefs that predicted change in bulimic attitudes (BULIT-R) and frequency of vomiting. (No such analysis was conducted for the other two indices of therapeutic gain, since each was predicted by a single YSQ scale). Due to the number of variables relative to the number of patients, YSQ scales were used only if they correlated strongly (r r0.60) with changes in symptomatology. As before, the impact of pretreatment level of symptomatology was also entered into the analysis as an independent variable. The ®rst regression analysis showed that changes in BULIT-R were reliably predicted by the YSQ scale scores and pretreatment pathology (overall F = 4.62; p < 0.02; explained variance=43%). However, the only signi®cant individual predictor was a negative eect of Functional dependence/ incompetence beliefs (t = 2.00; p < 0.05). Change in frequency of vomiting was also reliably predicted by the YSQ scale scores and pretreatment levels of vomiting (overall F = 13.1; p < 0.01; explained variance=66%). This association was a product of signi®cant individual eects of Defectiveness/shame (t = 1.75; p < 0.05) and pretreatment vomiting (t = 4.64; p < 0.01).
4. Discussion Each of the two hypotheses was supported. First, group cognitive behavioral therapy was eective in treating bulimia nervosa, and the size of the treatment eect was comparable with that obtained in other controlled and uncontrolled CBT trials (e.g. Garner, 1987; Mitchell et al., 1990; Agras et al., 1992; Fettes & Peters, 1992). Second, even after the impact of pretreatment symptoms was taken into account, pretreatment core beliefs were associated with the degree of change in bulimic psychopathology (vomiting, bulimic attitudes, restrictive attitudes). This con®rms that such cognitive representations constitute a further factor that may need to be taken into account when considering the ecacy of group or individual CBT in such cases (e.g. Baell & Wertheim, 1992; McKisack & Waller, 1997). With one exception, the associations between core beliefs and the change in eating psychopathology were in the expected direction Ð less healthy pretreatment core beliefs were followed by lower gains. A failure to show an improvement in general bulimic attitudes (BULIT-R scores) was related to a high level of unconditional `functional dependence/ incompetence' beliefs, as suggested by previous researchers (e.g. Garner et al., 1990; Baell & Wertheim, 1992). These ®ndings are broadly in line with earlier suggestions (e.g. Fairburn, Kirk, O'Connor, Anastasiades & Cooper, 1987) that pretreatment level of self-esteem can reliably predict treatment outcome. In contrast, the strongest cognitive predictor of maintained vomiting was a high level of unconditional `defectiveness/shame' beliefs Ð the YSQ variable that has the strongest association with levels of purging (Waller et al., in press). This ®nding is compatible with earlier models, which suggest that purging reduces awareness of intolerable cognitions about oneself (Pitts & Waller, 1993) or removes the problem from within oneself (Schupak-Neuberg & Nemero, 1993).
154
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
The one ®nding that ran counter to prediction was that unconditional `emotional deprivation' beliefs were strongly correlated with a positive change in anorectic cognitions. In other words, restrictive attitudes and cognitions were likely to be more reduced if the women believed pretreatment that their emotional needs would never be met. It can be hypothesized that this bene®t was a function of the group format, challenging the emotional deprivation belief in vivo by providing an emotionally supportive environment. Such an environment might have the value of encouraging the individual with such beliefs to participate more fully in the therapy, thus leading to greater gains. In contrast, the individual with lower emotional deprivation beliefs at the outset might ®nd the group format less supportive (and hence a less safe environment for change). In order to test this hypothesis, it would be necessary to determine whether this positive association still holds in individual therapy. At a general level, these ®ndings support the clinical suggestion that existing models of CBT (with their emphasis on negative automatic thoughts and dysfunctional assumptions) are more appropriate for some bulimic patients than for others. It could be argued that CBT for bulimia nervosa would be more eective if it included a schema-focused component (e.g. Padesky, 1994; Young, 1994), addressing core beliefs that are unrelated to food, shape and weight. Alternatively, it might be more ecient if dierent therapeutic `packages' were designed for dierent bulimic groups, where the level of unhealthy core beliefs played a role in determining the targeting of therapy. Since schema-focused CBT is usually more consuming of time and resources than routine CBT (e.g. Fairburn & Cooper, 1989; Young, 1994; Fairburn, 1997), it will be important to develop appropriate heuristics for making such decisions. At the diagnostic level, the presence of a comorbid personality disorder or a history of trauma might be a useful initial determinant of who should receive schema-focused CBT, but a more re®ned use of psychometrics (such as the YSQ) might assist in the task of assigning bulimic patients to the most appropriate treatment. These preliminary ®ndings suggest that the clinician might consider addressing the core beliefs of defectiveness/shame and functional dependence/ incompetence in order to enhance CBT's eectiveness at reducing vomiting behaviors and bulimic attitudes.
References Agras, W. S., Rossiter, E. M., Arnow, B., Schneider, J., Telch, C. F., Raeburn, S. D., Bruce, B., Perl, M., & Koran, L. M. (1992). Pharmacological and cognitive-behavioral treatment for bulimia nervosa: a controlled comparison. American Journal of Psychiatry, 149, 82±87. American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: American Psychiatric Association. Andrews, B. (1997). Bodily shame in relation to abuse in childhood and bulimia: a preliminary investigation. British Journal of Clinical Psychology, 36, 41±49. Baell, W. K., & Wertheim, E. H. (1992). Predictors of outcome in the treatment of bulimia nervosa. British Journal of Clinical Psychology, 31, 330±332. Braun, D. L., Sunday, S. R., & Halmi, K. A. (1994). Psychiatric comorbidity in patients with eating disorders. Psychological Medicine, 24, 859±867. Carroll, J. M., Touyz, S. W., & Beumont, P. J. V. (1996). Speci®c comorbidity between bulimia nervosa and personality disorders. International Journal of Eating Disorders, 19, 159±170.
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
155
Cooper, M. (1997). Cognitive theory in anorexia nervosa and bulimia nervosa: a review. Behavioral and Cognitive Psychotherapy, 25, 113±145. Cooper, P. J., Cooper, Z., & Hill, C. (1989). Behavioral treatment of bulimia nervosa. International Journal of Eating Disorders, 8, 87±92. Craighead, L. W., & Agras, W. S. (1991). Mechanisms of action in cognitive-behavioral and pharmacological interventions of obesity and bulimia nervosa. Journal of Consulting and Clinical Psychology, 59, 115±125. Fairburn, C. G. (1988). The current status of psychological treatments for bulimia nervosa. Journal of Psychosomatic Research, 32, 635±645. Fairburn, C. G. (1997). Eating disorders. In D. M. Clark, & C. G. Fairburn, Science and practice of cognitive behavior therapy. Oxford: Oxford University Press. Fairburn, C. G., & Cooper, P. (1989). Eating disorders. In K. Hawton, P. M. Salkovskis, J. Kirk, & D. M. Clark, Cognitive behavior therapy for psychiatric problems. Oxford: Oxford University Press. Fairburn, C. G., Cooper, Z., & Cooper, P. J. (1986). In K. D. Brownell, & J. P. Foreyt, Handbook of eating disorders: physiology, psychology and treatment of obesity, anorexia and bulimia (pp. 389±404). New York: Basic Books. Fairburn, C. G., Kirk, J., O'Connor, M., Anastasiades, P., & Cooper, P. J. (1987). Prognostic factors in bulimia nervosa. British Journal of Clinical Psychology, 26, 223±224. Fairburn, C. G., Marcus, M. D., & Wilson, G. T. (1993). Cognitive-behavioral therapy for binge eating and bulimia nervosa: a comprehensive treatment manual. In C. G. Fairburn, & G. T. Wilson, Binge eating: nature, assessment and treatment (pp. 361±404). New York: Guilford. Fairburn, C. G., Norman, P. A., Welch, S. L., O'Connor, M. E., Doll, H. A., & Peveler, R. C. (1995). A prospective study of outcome in bulimia nervosa and the long term eects of three psychological treatments. Archives of General Psychiatry, 52, 304±312. Fettes, P. A., & Peters, J. M. (1992). A meta-analysis of group treatments for bulimia nervosa. International Journal of Eating Disorders, 11, 97±110. Freeman, C. (1995). Cognitive therapy. In G. Szmukler, C. Dare, & J. Treasure, Handbook of eating disorders: theory, treatment and research. New York: Wiley. Garner, D. M. (1987). Psychotherapy outcome research with bulimia nervosa. Psychotherapy and Psychosomatics, 48, 129±140. Garner, D. M., Olmsted, M. P., Davies, R., Rockert, W., Goldbloom, D., & Eagle, M. (1990). The association between bulimic symptoms and reported psychopathology. International Journal of Eating Disorders, 9, 1±15. Hartmann, A., Herzog, T., & Drinkmann, D. (1992). Psychotherapy for bulimic nervosa: what is eective? A meta analysis. Journal of Psychosomatic Research, 36, 159±167. Herzog, D. B., Keller, M. B., Strober, M., Yeh, C., & Pai, S.-Y. (1992). The current status of treatment for anorexia nervosa and bulimia nervosa. International Journal of Eating Disorders, 12, 215±220. Johnson, C., Tobin, D., & Dennis, A. (1990). Dierence in treatment outcome between borderline and non-borderline bulimics at one-year follow-up. International Journal of Eating Disorders, 9, 617±627. McKisack, C., & Waller, G. (1997). Factors in¯uencing the outcome of group psychotherapy for bulimia nervosa. International Journal of Eating Disorders, 22, 1±13. McManus, F., Waller, G., & Chadwick, P. (1996). Biases in the processing of dierent forms of threat in bulimic and comparison women. Journal of Nervous and Mental Disease, 184, 547±554. Mitchell, J. E., Pyle, R. L., Eckert, E. D., Hatsukami, D., Pomeroy, C., & Zimmerman, K. (1990). A comparison of antidepressants and structured group therapy in the treatment of bulimia nervosa. Archives of General Psychiatry, 47, 149±157. Mizes, J. S., & Klesges, R. C. (1989). Validity, reliability and factor structure of the anorectic cognitions questionnaire. Addictive Behaviors, 14, 589±594. Newton, J. R., Freeman, C. P., & Munro, J. (1993). Impulsivity and dyscontrol in bulimia nervosa: is impulsivity an independent phenomenon or a marker of severity? Acta Psychiatrica Scandinavica, 87, 389±394. Padesky, C. A. (1994). Schema change processes in cognitive therapy. Clinical Psychology and Psychotherapy, 1, 267±278. Phillips, L., Tiggemann, M., & Wade, T. (1997). Comparison of cognitive style in bulimia nervosa and depression. Behavior Research and Therapy, 35, 939±948.
156
N. Leung et al. / Behaviour Research and Therapy 38 (2000) 145±156
Pitts, C., & Waller, G. (1993). Self-denigratory beliefs following sexual abuse: association with the symptomatology of bulimic disorders. International Journal of Eating Disorders, 13, 407±410. Root, M. P. P., & Fallon, P. (1989). Treating the victimized bulimic. Journal of Interpersonal Violence, 4, 90±100. Sansone, R. A., & Fine, M. A. (1992). Borderline personality disorder as a predictor of outcome in women with eating disorders. Journal of Personality Disorders, 6, 176±186. Schmidt, N. B., Joiner, T. E., Young, J. E., & Telch, M. J. (1995). The Schema Questionnaire: investigation of psychometric properties and the hierarchical structure of a measure of maladaptive schemas. Cognitive Therapy and Research, 19, 295±321. Schupak-Neuberg, E., & Nemero, C. J. (1993). Disturbances in identity and self-regulation in bulimia nervosa: implications for a metaphorical perspective of `body as self'. International Journal of Eating Disorders, 13, 335±347. Skodol, A. E., Oldham, J. M., Hyler, S. E., Kellman, H. D., Doidge, N., & Davies, M. (1993). Comorbidity of DSM-III-R eating disorders and personality disorders. International Journal of Eating Disorders, 14, 403±416. Thelen, M. H., Farmer, J., Wonderlich, S., & Smith, M. (1991). A revision of the Bulimia Test: the BULIT-R. Psychological Assessment, 3, 119±124. Vitousek, K. M. (1996). The current status of cognitive-behavioral models of anorexia nervosa and bulimia nervosa. In P. M. Salkovskis, Frontiers of cognitive therapy. New York: Guilford. Waller, G., Ohanian, V. Meyer, C., & Osman, S. (in press). Cognitive content among bulimic women: the role of core beliefs. International Journal of Eating Disorders. Waller, G., & Ruddock, A. (1995). Information-processing correlates of reported sexual abuse in eating-disordered and comparison women. Child Abuse and Neglect, 19, 745±759. Wilson, G. T., & Fairburn, C. G. (1993). Cognitive treatment for the eating disorders. Journal of Consulting and Clinical Psychology, 61, 261±269. Wonderlich, S. A. (1995). Personality and eating disorders. In K. D. Brownell, & C. G. Fairburn, Eating disorders and obesity: a comprehensive handbook. London: Guilford. Young, J. E. (1994). Cognitive therapy for personality disorders: a schema-focused approach (2nd ed.). Sarasota: Professional Resource Press.