052 23-Turnercoopercogorigins2002

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Clinical Psychology and Psychotherapy Clin. Psychol. Psychother.

9, 242252 (2002)

Cognitions and their Origins in Women with Anorexia Nervosa, Normal Dieters and Female Controls
Hannah Turner* and Myra Cooper
Isis Education Centre, Warneford Hospital, Headington, Oxford, UK The objective of the study was to develop a semi-structured interview with which to investigate cognitions and their origins in female patients with anorexia nervosa, normal female dieters and female nondieting controls and to explore group differences in cognitions and, where appropriate, the role of early experience in their development. Following piloting, all participants completed the semi-structured interview and ve self-report questionnaires to assess reliability and validity of the interview. When discussing eating-related concerns clinical participants reported more eating-related thoughts than nonclinical participants. These differences were also evident, although to a slightly lesser extent, when discussing concerns about weight and shape. Clinical participants reported more assumptions relating to eating, and weight and shape as a means to acceptance by self and others, and to control over-eating than non-clinical participants. A similar pattern of group difference was reported in degree of belief and associated distress. Clinical participants identied more negative self-beliefs than non-clinical participants. A similar pattern of group difference was reported in degree of rational and emotional belief, and associated distress. All clinical participants identied an association between negative early experiences and negative self-beliefs, and all reported a link between negative self-beliefs and dieting. There are clear differences in cognitive characteristics between women with anorexia nervosa, normal dieters and non-dieting female controls. The meaning attached to dieting may also distinguish dieters from those with a clinical eating disorder. Copyright 2002 John Wiley & Sons, Ltd.

INTRODUCTION
In the rst systematic cognitive-behavioural account of anorexia nervosa Garner and Bemis (1982) highlighted the role of automatic thoughts, underlying assumptions and disturbed information
* Correspondence to: Dr Hannah Turner, Isis Education Centre, Warneford Hospital, Headington, Oxford, OX3 7JX, UK. Tel: (01865)226431. Fax:(01865)226364. E-mail: hannah.turner@hmc.ox.ac.uk

processing. In particular they emphasized the role of weight- and shape-related self-schemata. Recently attention has also been drawn to the importance of negative self-beliefs (Cooper, 1997), similar to the concept of self-schemata identied by Vitousek and Hollon (1990). However, despite growing interest, cognitive theory lacks detail regarding the nature of negative self-beliefs in eating disorders. It also lacks detail about the content and form of the underlying assumptions

Copyright 2002 John Wiley & Sons, Ltd.

Published online 1 March 2002 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/cpp.317

Cognitions and their Origins in AN and Dieting linking these beliefs to disturbed eating patterns (Cooper, 1997). Finally, there is little detail on the role of early experiences in the development of both negative self-beliefs and underlying assumptions (Cooper, 1997). Recently, a multidimensional measure designed to assess the negative self-beliefs and underlying assumptions associated with eating disorders has been developed. The measure is the Eating Disorder Belief Questionnaire (EDBQ; Cooper, Cohen-Tov e, e Todd, Wells, & Tov e, 1997). However, despite e possessing favourable psychometric properties, the EDBQ has disadvantages when compared with other types of data collection, for example, semistructured interviews. Unlike self-report questionnaires, interviews can assess idiosyncratic cognitive content, different dimensions of cognitions and associated emotional distress/affect. In eating disorders, a semi-structured interview would also allow early experiences associated with negative self-beliefs and underlying assumptions to be explored. Links between negative self-beliefs and disturbed behaviour, such as extreme dieting could also be investigated. The latter providing information on a possible mechanism. One such semi-structured interview already exists (Cooper, Todd, & Wells, 1998). In a preliminary study using this interview it was found that the self-beliefs of patients with eating disorders, unlike those of non-eating disordered women, were negative, unconditional and concerned with themes such as abandonment and uselessness. They were also invariably associated with negative early experiences. Underlying assumptions were of two kinds; weight and shape as a means to self-acceptance, and weight and shape as a means to acceptance by others. There were also assumptions about eating. The interview also indicated that underlying assumptions provided a link between negative self-beliefs and dieting behaviour. Assumptions appeared to be compensatory beliefs, providing a way for the individual to overcome negative self-beliefs (Young, 1990). Although these ndings provide detailed information about beliefs, assumptions and early experiences in eating disorders, the study had a number of limitations. The data were purely qualitative, no information was provided on the validity and reliability of the interview and no systematic ratings of assumptions and beliefs were taken. The sample size was small and the interview failed to distinguish between eating-, and weight- and shape-related situations, despite suggestions that the core psychopathology of anorexia nervosa may lie in the personal meaning attached to weight and
Copyright 2002 John Wiley & Sons, Ltd.

243 shape (Vitousek & Hollon, 1990). It also remains unclear whether, and how, the features identied are unique to eating disordered patients as opposed to relevant comparison groups, such as dieters. This paper therefore describes the development and use of a revised version of the interview rst developed by Cooper et al. (1998). Unlike the original interview, it investigates eating-, and weight- and shape-related situations separately. Systematic ratings were taken, both of cognitions and associated affect, and a group of dieters was included. It also sought to provide information on the interviews reliability and validity. Additionally it investigated levels of rational and emotional belief (see for example Beck, 1995).

Aim
The rst aim of the study was to pilot the semistructured interview and gather relevant reliability and validity data, including data on testre-test reliability. The second aim was to investigate potential group differences in cognitions and, where relevant, associated early experiences.

METHOD
Participants Patients with Anorexia Nervosa These were 18 female patients with a DSM-IV diagnosis of anorexia nervosa (American Psychiatric Association, 1994). All were recruited through their primary therapist. Normal Dieters These subjects were 18 female volunteers who fullled a strict denition of dieting. They were recruited by requesting volunteers from amongst university students and hospital staff. Dieting was dened as following a standard weight reducing diet and/or setting rigid rules about what should be eaten for the past month or more (Cooper & Fairburn, 1992). Women who had engaged in selfinduced vomiting, bingeing1 or laxative/diuretic abuse were excluded by interview, as were those with a history of an eating disorder or psychiatric illness.
1 Dened as a period during which the individual experiences a sense of lack of control over their eating and consumes a quantity of food that is greater than most individuals would normally eat in a similar period of time.

Clin. Psychol. Psychother. 9, 242252 (2002)

244

H. Turner and M. Cooper help participants identify any automatic thoughts related to eating and/or to weight and shape, and to clarify the duration of each. Participants were also asked to rate the extent to which they believed each thought at the time, as well as to identify the most distressing eating-related thought.3 The downward arrow technique described by Burns (1980) was then used to pursue the most distressing eatingrelated thought and to identify any underlying assumptions. Specic probe questions included suppose that were true, whats the worst that would say about you? what do you think other people would think about you? Participants were asked to rate their level of belief in each assumption, as well as the degree of distress associated with it. Both assumptions related to self-acceptance (e.g. If I eat this it means Im a bad person) and acceptance by others (e.g. If I eat this others will think Im no good) were identied and subsequently explored separately. The origins of any assumptions were then identied. Participants were asked whether these assumptions reected their current beliefs. Assumptions related to self-acceptance were then pursued again using the downward arrow technique to elicit any negative self-beliefs. Participants were asked to rate their rational and emotional belief in each negative self-belief and the degree of distress caused by each belief. Information about the origins of any negative selfbeliefs was collected. Prompt questions included what is your rst memory of these beliefs? how old were you? and what was happening in your life at the time? Participants were asked whether these beliefs reected their current beliefs about themselves and whether they did anything to change the way they thought or felt. This led into additional questions designed to shed further light upon the manner in which negative self-beliefs might impact upon behaviour (e.g. do you feel that the beliefs.. . .and the distress that goes with them are connected in some way to your attempts to diet/prevent weight gain?).

Non-clinical Controls These were 18 female volunteers recruited in the same way as the normal dieters. Women with a history of an eating disorder or psychiatric illness were excluded by interview. Women who were currently dieting or had a history of self-induced vomiting, bingeing or laxative/diuretic abuse were also excluded from this group. Measures Demographic Information Information was collected on age, number of years in full time education, weight and height (used to calculate Body Mass Index, kg/m2 ). Self-report Questionnaires To enable reliability and validity information to be calculated participants completed the following: the Eating Attitudes Test (EAT: Garner & Garnkel, 1979), the Eating Disorder Belief Questionnaire (EDBQ: Cooper et al., 1997), the Rosenberg SelfEsteem Scale (RSE: Rosenberg, 1965), the Beck Anxiety Inventory (BAI: Beck, Epstein, Brown, & Steer, 1988) and the Beck Depression Inventory (BDI: Beck, Ward, Mendelson, Mock, & Erbaugh, 1961). These measures were chosen as they have established reliability and validity, and as such were regarded as reliable markers from which to assess the interviews psychometric properties. Semi-Structured Interview
The semi-structured interview was based on that developed by Cooper et al. (1998). However, it was modied and extended to allow more detailed and systematic assessment of cognitions, affect and early experiences. Eating, and weight and shape concerns were also investigated separately. The interview was piloted on six volunteers (two participants with anorexia nervosa, two normal dieters and two female controls).

Section 1: Eating-related Concerns Participants were asked to focus on the most recent situation in which they had felt anxious, bad or worried about their eating. Prompt questions included where were you? and what were you doing? They were asked to identify the feelings they had experienced in the situation and to rate the strength of each.2 Questions were then asked to
All ratings were made on a 100-point Likert scale (e.g. 0 D not at all anxious; 100 D as anxious as I have ever felt).
2

Section 2: Weight- and Shape-related Concerns The second part of the interview asked participants to identify the last situation in which they had felt anxious, worried or bad about their weight and/or shape. As for the eating-related situation, they were asked to describe the situation
3 All ratings of belief and distress were also given on a 100-point Likert scale (e.g. 0 D I did not believe this at all; 100 D I was completely convinced that this was true).

Copyright 2002 John Wiley & Sons, Ltd.

Clin. Psychol. Psychother. 9, 242252 (2002)

Cognitions and their Origins in AN and Dieting and identify the duration of their concerns. The probe questions used in the rst section were again employed, as participants were asked to describe how they had felt and what thoughts were running through their mind at the time. As before, the duration and the extent to which they believed each thought was also documented, and the most distressing weight- and/or shape-related thought was identied. This was subsequently pursued in a manner synonymous to that described earlier, with assumptions relating to self and others being explored independently, and any additional negative self-beliefs also being identied and explored.

245 Where the data failed to meet the criteria for parametric tests, non-parametric equivalents were applied.

Participant Characteristics
Information on age, years in full time education, body mass index (kg/m2 ) can be seen in Table 1. One-way analyses of variance with post hoc Tukey tests were used to assess differences between groups on demographic variables. Results indicated that the controls were signicantly younger than the dieters (p < 0.05) and had spent signicantly longer in education than the clinical participants (p < 0.05). Differences between groups were also found for BMI (F 2, 50 D 38.13, p < 0.001). The clinical participants had a signicantly lower BMI than the dieters and the controls, and the dieters had a signicantly higher BMI than the controls (all comparisons p < 0.05).

Procedure
All participants were screened with the Eating Disorder modules of the SCID-IV (Spitzer, Williams, & Gibbons, 1996) to conrm or exclude a DSM-IV (American Psychiatric Association, 1994) diagnosis of anorexia nervosa. All participants were seen individually. All the interviews were conducted by a clinician trained in the use of the interview. Interviews lasted between 15 and 45 min depending upon group membership, with clinical participants typically taking longer. Each interview was audiotaped with the participants permission. To assess testre-test reliability of the interview 27 participants (nine from the anorexia nervosa group, nine from the dieting group and nine from the female control group) completed the interview a second time, 12 weeks later.

Inter-rater Reliability
Inter-rater reliability of the interview was checked by comparing the frequency of negative automatic thoughts, assumptions4 and core beliefs identied by two independent raters in each section of the interview. Given that the data was ordinal as opposed to categorical, Wilcoxon matched pairs sign rank tests were used instead of the more traditional Cohns kappa coefcient. The two raters scored none of the 17 items signicantly differently.
4 The interview distinguishes between rst order and second order assumptions. Second order assumptions link automatic thoughts directly with core beliefs e.g. If I get fat Ill have failed. First order assumptions do not, rather they are at the same level as automatic thoughts e.g. If I eat this biscuit Ill get fat. These are particularly common in relation to eating concerns.

RESULTS
Data was analysed using SPSS (SPSS, Inc., 1997). Checks for normality of distribution were made by group, using the KolmogorovSmirnov test.

Table 1. Demographic data by group Controls (N D 18) Mean age in years (SD) Mean years in education (SD) Mean BMI (SD)

Dieters (N D 18) 29.9 (9.1) 11.0 (2.3) 25.2 (4.9)

Anorexia nervosa patients (N D 18) 24.7 (7.1) 9.8 (1.5) 16.1 (1.6)

23.9 (4.1) 12.2 (1.8) 21.0 (1.8)

SD, standard deviation. Mean number of years in full-time education D from the age of 11 years upwards. BMI, kg/m2 .

Copyright 2002 John Wiley & Sons, Ltd.

Clin. Psychol. Psychother. 9, 242252 (2002)

246

H. Turner and M. Cooper and EDBQ control over eating subscale, were signicant. Concurrent validity was further assessed by correlating the mean emotion expressed at different points in each section with scores on the Beck Depression Inventory (BDI), the Beck Anxiety Inventory (BAI), and the Rosenberg Self-Esteem Scale (RSE). Within the eating section all correlations were signicant. Within the weight and shape section, all correlations, with the exception of that between mean emotion and BAI, were signicant. The EDBQ negative self-beliefs subscale was also correlated with mean belief in negative selfbeliefs and mean associated distress (for rational and emotional belief separately). In both sections all correlations were signicant (see Table 3).

Test Re-test Reliability


The test re-test reliability of the interview was assessed by comparing the frequency and belief in second order assumptions, and core beliefs identied in each section of the interview by group. Wilcoxon matched pairs sign rank tests were again used as the data was ordinal rather than categorical. Only four of the 45 items were signicantly different at p < 0.05.

Concurrent Validity
Global scores on the EAT and the EDBQ assumption subscales, which might be expected to correlate positively with the number of eating-, and weightand shape-related automatic thoughts and assumptions elicited by the interview, were used to check the concurrent validity of the interview. Spearman coefcient correlations were used. Table 2 shows the correlations between the EAT and these variables. Eating, and weight and shape sections are considered separately. Within the eating section all the correlations were signicant. Within the weight and shape section all the correlations, with the exception of those between the number of self- and other referent rst order eating-related assumptions and the EAT

Semi-structured Interview Eating and Weight- and Shape-related Negative Automatic Thoughts Eating Section. KruskalWallis analysis of variance tests with post hoc MannWhitney U analyses indicated a signicant difference in the number of eating-related NATs (e.g. I shouldnt be eating this) found in the three groups ( 2 2 D 42.0, p < 0.001). The clinical participants reported significantly more of these thoughts than the dieters and

Table 2. Correlations between the EAT, the EDBQ subscales, and the number of eating-, and weightand shape-related thoughts and assumptions EAT Eating section Eating-related automatic thoughts Weight- and shape-related automatic thoughts First order eating-related assumption (SR) First order eating-related assumption (OR) Second order eating-related assumptions (SR) Second order eating-related assumptions (OR) Weight and shape section Eating-related automatic thoughts Weight- and shape-related automatic thoughts First order eating-related assumption (SR) First order eating-related assumption (OR) Second order eating-related assumptions (SR) Second order eating-related assumptions (OR) Second order weight- and shape-related assumptions (SR) Second order weight- and shape-related assumption (OR) 0.73 0.23 0.74 0.40 SA A C 0.71 0.71 0.34

0.22 0.50 0.17 70 0.73

0.21 0.11

N D 54. p < 0.05; p < 0.01; p < 0.001 (one-tailed). SR, self-referent; OR, other-referent; EAT, Eating Attitudes Test. Eating Disorder Belief Questionnaire Subscales: SA, self acceptance; A, acceptance by others; C, control over eating. Gaps indicate that no analysis was conducted and indicates that a correlation could not be computed due to a lack of data in the corresponding cells.

Copyright 2002 John Wiley & Sons, Ltd.

Clin. Psychol. Psychother. 9, 242252 (2002)

Cognitions and their Origins in AN and Dieting


Table 3. Correlations between the BDI, BAI, RSE, EDBQ-NSB and mean ratings of emotion, rational and emotional belief, and associated distress BDI Eating section Mean emotion Weight and shape section Mean emotion Negative self-beliefs Rational belief Emotional belief Distress rating BAI RSE EDBQNSB

247

0.72 0.38

0.72 0.20

0.69 0.27 0.76 0.75 0.76

N D 54. p < 0.05; p < 0.01; p < 0.001 (one-tailed). BDI, Beck Depression Inventory; BAI, Beck Anxiety Inventory; RSE, Rosenberg Self-Esteem Scale; EDBQ-NSB, Negative Self-Beliefs Scale. Summed across situations.

the controls (both comparisons, p < 0.001). Signicant group differences were also found in relation to the duration of eating-related NATs ( 2 2 D 44.3, p < 0.001). The clinical group reported these thoughts as being signicantly longer in duration than the dieters and the controls (both comparisons, p < 0.001). A signicant difference in the degree to which participants believed their eatingrelated NATs was also found between the groups ( 2 2 D 38.9, p < 0.001). The clinical participants believed these thoughts signicantly more than the dieters and the controls (both comparisons, p < 0.001). Signicant group differences were also found in relation to the distress caused by the most distressing eating ( 2 2 D 47.4, p < 0.001) and the most distressing weight- and shape-related (e.g. Ill get fat, 2 2 D 8.1, p < 0.05) NAT identied. The clinical participants found their most distressing eating- (p < 0.001) and most distressing weightand shape-related (p < 0.01) thoughts signicantly more distressing than the dieters. Similarly, the clinical participants were signicantly more distressed by their most distressing eating- (p < 0.001) and weight- and shape-related (p < 0.05) thought than the controls. There were no signicant differences between the dieters and the controls.

Weight and Shape Section. The three groups differed signicantly in relation to the number of eating- ( 2 2 D 8.0, p < 0.05) and the number of weight- and shape-related NATs ( 2 2 D 29.8, p < 0.001) elicited. Although the clinical participants did not report a signicantly greater number of eating- or weight- and shape-related NATs than the dieters, they did report signicantly
Copyright 2002 John Wiley & Sons, Ltd.

more eating- (p < 0.01) and weight- and shaperelated (p < 0.001) NATs than the controls. The dieters also reported signicantly more eating(p < 0.01) and weight- and shape-related (p < 0.001) NATs than the controls. Signicant group differences were found in relation to the duration of eating ( 2 2 D 7.7, p < 0.05) and the duration of weight- and shape-related ( 2 2 D 31.1, p < 0.001) NATs identied. More specically, the clinical participants reported signicantly longer duration of weight- and shape-related NATs than the dieters (p < 0.01). They also reported signicantly longer duration of eating- (p < 0.01) and longer duration of weight- and shape-related (p < 0.001) NATs than the controls. The dieters reported signicantly longer duration of eating- (p < 0.01) and longer duration of weight- and shape-related (p < 0.001) thoughts than the controls. Signicant group differences were found in relation to the extent to which eating- ( 2 2 D 8.3, p < 0.05) and weight- and shape-related ( 2 2 D 22.7, p < 0.001) thoughts were believed. The clinical participants believed these eating- (p < 0.01) and weight- and shape-related (p < 0.001) NATs signicantly more than the controls. There was no signicant difference in the extent to which the clinical participants and the dieters believed these eating-, and weight- and shape-related thoughts. The dieters believed these eating- (p < 0.01) and weight- and shape-related (p < 0.001) NATs signicantly more than the controls. Signicant group differences were also found in relation to the degree of distress caused by the most distressing weightand shape-related NAT ( 2 2 D 34.1, p < 0.001). The clinical participants found their most distressing weight- and shape-related NAT signicantly
Clin. Psychol. Psychother. 9, 242252 (2002)

248 more distressing than the dieters and the controls, and the dieters found their most distressing weightand shape-related NAT signicantly more distressing than the controls (all comparisons, p < 0.001).

H. Turner and M. Cooper signicantly more distressing than did the controls (p < 0.01).

Assumptions Related to Eating as a Means to Self-acceptance Signicant differences between groups were found in relation to the number of second order assumptions related to eating as a means to selfacceptance (e.g. If I eat this it means Im a bad person, 2 2 D 24.5, p < 0.001). The clinical group reported signicantly more of these assumptions than the dieters and the controls (both comparisons, p < 0.001). Signicant group differences were found in relation to the extent to which these assumptions were believed ( 2 2 D 24.5, p < 0.001). The clinical group believed these assumptions signicantly more than the dieters and the controls (both comparisons, p < 0.001). Signicant group differences were also found in relation to the degree to which these assumptions caused distress ( 2 2 D 24.5, p < 0.001). The clinical participants found these assumptions signicantly more distressing than did the dieters and the controls (both comparisons, p < 0.001). There were no signicant differences between the dieters and the female controls. Assumptions Related to Weight and Shape as a Means to Self-acceptance There was a signicant difference between groups in the number of assumptions related to weight and shape as a means to self-acceptance (e.g. If Im fat it means Im a disgrace, 2 2 D 27.1, p < 0.001). The clinical group reported signicantly more of these assumptions than the dieters (p < 0.01) and the controls (p < 0.001). The dieters reported signicantly more of these assumptions than the controls (p < 0.01). Signicant group differences were found in relation to the extent to which these assumptions were believed ( 2 2 D 30.4, p < 0.001). The clinical participants believed these assumptions signicantly more than the dieters (p < 0.01) and the controls (p < 0.001). The dieters believed these assumptions signicantly more than the controls (p < 0.01). Signicant group differences were also found in relation to the degree to which these assumptions caused distress ( 2 2 D 33.4, p < 0.001). The clinical participants found these assumptions signicantly more distressing than did the dieters and the controls (both comparisons, p < 0.001). The dieters found these assumptions
Copyright 2002 John Wiley & Sons, Ltd.

Assumptions Related to Eating as a Means to Acceptance by Others Signicant differences between groups were found in relation to the number of second order assumptions related to eating as a means to acceptance by others (e.g. If I eat this others will think Im greedy, 2 2 D 18.4, p < 0.001). The clinical group reported signicantly more of these assumptions than the dieters and the controls (both comparisons, p < 0.001). Signicant group differences were found in relation to the extent to which these were believed ( 2 2 D 18.4, p < 0.001). The clinical participants believed them signicantly more than the dieters and the controls (both comparisons, p < 0.001). Signicant group differences were found in relation to the degree to which assumptions related to eating as a means to acceptance by others caused distress ( 2 2 D 18.4, p < 0.001). The clinical participants found these assumptions signicantly more distressing than did the dieters and the controls (both comparisons, p < 0.001). There were no signicant differences between the dieters and the controls. Assumptions Related to Weight and Shape as a Means to Acceptance by Others There was a signicant difference between groups in the number of second order assumptions related to weight and shape as a means to acceptance by others elicited (e.g. If Im fat others will think Im useless, 2 2 D 31.8, p < 0.001). The clinical group reported signicantly more of these assumptions than the dieters (p < 0.01) and the controls (p < 0.001). The dieters reported signicantly more of these assumptions than the controls (p < 0.001). There was a signicant difference between groups in the extent to which these assumptions were believed ( 2 2 D 37.4, p < 0.001). The clinical participants believed these assumptions signicantly more than the dieters and the controls, and the dieters believed these assumptions signicantly more than the controls (all comparisons, p < 0.001). Signicant differences between groups were also found in relation to the degree to which these assumptions caused distress ( 2 2 D 39.2, p < 0.001). The clinical participants found these assumptions signicantly more distressing than did the dieters and the controls, and the dieters found these assumptions signicantly more distressing than did the controls (all comparisons, p < 0.001).
Clin. Psychol. Psychother. 9, 242252 (2002)

Cognitions and their Origins in AN and Dieting

249 clinical group also found these assumptions significantly more distressing than did the dieters and the controls (both comparisons, p < 0.01).

Control Over Eating Signicant differences between groups were found in relation to the number of self-referent rst order eating-related assumptions, (e.g. If I eat this Ill get fat, 2 2 D 43.1, p < 0.001) and number of other-referent rst order eating-related assumptions (e.g. If I eat this others will think Ive put on weight, 2 2 D 10.8, p < 0.01). The clinical group reported signicantly more self-referent rst order eating-related assumptions than the dieters and the controls (both comparisons, p < 0.001). No signicant differences were found between the dieters and the controls. The clinical participants also reported signicantly more self-referent rst order eating-related assumptions than the dieters and the controls (both comparisons, p < 0.01). There was a signicant difference between groups in relation to the extent to which selfreferent ( 2 2 D 43.0, p < 0.001) and other-referent ( 2 2 D 10.8, p < 0.01) rst order eating-related assumptions were believed. The clinical participants believed their self-referent rst order eatingrelated assumptions signicantly more than the dieters and the controls (both comparisons, p < 0.0001). The clinical group also believed their otherreferent rst order eating-related assumptions signicantly more than the dieters and the controls (both comparisons, p < 0.01). Signicant group differences were found in relation to the degree to which self-referent ( 2 2 D 36.1, p < 0.001) other-referent ( 2 2 D 10.8, p < 0.01) rst order eating-related assumptions caused distress. The clinical participants found their selfreferent rst order eating-related assumptions signicantly more distressing than did the dieters and the controls (both comparisons, p < 0.001). The

Negative Self-beliefs Negative self-beliefs (e.g. Im bad, Im lazy) were summed across the two sections. KruskalWallis analyses of variance tests for between group differences with post hoc MannWhitney U tests indicated signicant differences between groups in relation to the number of negative self-beliefs ( 2 2 D 44.8, p < 0.001); the degree of rational belief ( 2 2 D 35.7, p < 0.001); the degree of emotional belief ( 2 2 D 38.0, p < 0.001); and associated distress ( 2 2 D 42.5, p < 0.001). The clinical group reported signicantly more negative selfbeliefs than the dieters and the controls (both comparisons, p < 0.001). They also reported a signicantly higher degree of rational and emotional belief, and associated distress, than the dieters and the controls (all comparisons, p < 0.001). The dieters reported signicantly more negative self-beliefs than the controls. They also reported signicantly higher levels of rational and emotional belief, and associated distress, than the controls (all comparisons p < 0.05). Origins of Assumptions and Negative Self-beliefs The number of clinical participants who associated underlying assumptions and negative selfbeliefs with negative early experiences is reported in Table 4 below. The results indicated that all the clinical participants identied an association between negative self-beliefs and negative early experiences. Furthermore, a large percentage (72100%) also identied

Table 4. The number of clinical participants who identied an association between negative self-beliefs and second order assumptions, and negative early experiences Association identied No (%) Yes (%) Eating section Second order eating-related assumption (SR) Second order eating-related assumptions (OR) Weight and shape section Second order eating-related assumptions (SR) Second order eating-related assumptions (OR) Second order weight- and shape-related assumptions (SR) Second order weight- and shape-related assumptions (OR) Summed across groups Negative self-beliefs
N D 18. SR, Self-referent; OR, other-referent.

4 (22) 5(28) 0(0) 0(0) 2(11) 0(0) 0(0)

14(78) 13(72) 18(100) 18(100) 16(89) 18(100) 18(100)

Copyright 2002 John Wiley & Sons, Ltd.

Clin. Psychol. Psychother. 9, 242252 (2002)

250

H. Turner and M. Cooper


Table 5. The number of participants within each group who identied a link between negative self-beliefs and dieting Controls (N D 18) Dieters (N D 18) Anorexia nervosa patients (N D 18) 18 0

Link identied Yes No

0 18

4 14

an association between second order assumptions and negative early experiences.

Links Between Dieting and Negative Self-beliefs A chi-squared test5 for between group differences indicated a signicant difference between the three groups ( 2 2 D 41.1, p < 0.001). As indicated in Table 5, all the clinical participants, and four of the dieters identied a link between dieting and negative self-beliefs. None of the controls identied such a link.

DISCUSSION
This study describes the development of a semistructured interview designed to assess cognitions and their origins in anorexia nervosa. Preliminary ndings suggest that the interview has promising psychometric properties. It has good inter-rater and test re-test reliability, as well as promising concurrent validity. The study also investigated differences between patients with anorexia nervosa, non-symptomatic dieters and female controls. The results suggest that the interview may be a useful addition to the research tools available to investigate cognitions and their origins in eating disorders. It may also be of value in clinical work, particularly in developing a cognitive formulation. Findings suggest that when asked to discuss eating-related concerns, the clinical participants experience more eating-related negative automatic thoughts than dieters and controls, with similar differences being found in duration, degree of belief and associated distress. When asked to discuss concerns about weight and shape, clinical participants and dieters report more eating-, and weight- and shape-related thoughts than controls,
5 None of the conditions of the test, including independence of observations and expected cell size, were broken.

with similar differences being found in degree of belief. Clinical participants also report longer duration of weight- and shape-related thoughts than dieters, and longer duration of eating-, and weight- and shape-related thoughts than controls. However, dieters found their most distressing weight- and shape-related thought less distressing than clinical participants and more distressing than did the controls. Clinical participants have more underlying assumptions related to eating, and weight and shape as a means to acceptance by self and others, and more underlying assumptions related to control over eating than dieters and controls. Similar differences were found in degree of belief and associated distress. Dieters were also found to have more weight- and shape-related assumptions concerned with acceptance by self and others than controls. Again, similar differences were found in degree of belief and associated distress. Clinical participants were found to have more negative self-beliefs than both groups of non-clinical participants, with similar differences being reported in degree of rational and emotional belief, and associated distress. Negative self-beliefs and second order assumptions were invariably associated with negative early experiences by the clinical participants, who also identied a link between dieting and negative self-beliefs. These results give prominence to the potential importance, in anorexia nervosa, of assumptions linking eating, weight and shape to acceptance by self and others, and to the role of negative automatic thoughts related to eating and to weight and shape. In addition, they draw attention to the role of negative self-beliefs. Recently it has been argued (Cooper et al., 1998) that it is the presence of both negative self-beliefs and underlying assumptions about weight, shape and eating that are required for an eating disorder to develop. The present study is consistent with this suggestion. It also suggests that, while underlying assumptions about weight and shape are important, it may be the presence of
Clin. Psychol. Psychother. 9, 242252 (2002)

Copyright 2002 John Wiley & Sons, Ltd.

Cognitions and their Origins in AN and Dieting negative self-beliefs and underlying assumptions about eating that is particularly crucial in anorexia nervosa. These cognitions distinguished patients most clearly from dieters in the current study. Such ndings parallel the earlier work of Cooper and Fairburn (1992), who found that concerns about eating, rather than weight and shape, distinguished patients with anorexia nervosa most clearly from dieters. The ndings are also consistent with other theoretical accounts of anorexia nervosa (Vitousek & Hollon, 1990), in which weight- and shape-related self-schemata and self-schemata are highlighted and, in anorexia nervosa, particular emphasis is placed on control over eating. Weight- and shaperelated self-schemata appear to map onto the underlying assumptions related to weight and shape identied here; whilst self-schemata map on to negative self-beliefs. In relation to the causal link, the data suggests that clinical participants regard successful dieting as a means of reducing the distress caused by their negative self-beliefs. This then results in enhanced feeling of success and self-esteem. This is consistent with the suggestion by Cooper et al. (1998) that these beliefs may function as schema compensation beliefs (Young, 1990). In the current study, clinical participants also report that the enhanced self-esteem produced is quickly negated by thoughts of inadequate dieting, which in turn reinforces negative self-beliefs. We suggest that this process may be synonymous with the schema maintenance processes identied by Young (1990). The ndings of the present study raise a number of implications for the psychological treatment of anorexia nervosa. In relation to formulation, they highlight key cognitive issues that require careful exploration during the early stages of therapy. For example, rather than identifying assumptions as a group per se, it may be benecial to differentiate between rst and second order eating-, and weight- and shape-related assumptions. A common difculty in treatment is failure to distinguish the two levels of assumptions, resulting in unsuccessful attempts to challenge rst order assumptions (which are at the automatic thought level) with techniques more suitable for underlying assumptions. In previous models of anorexia nervosa, much emphasis has been placed on NATs and, in particular, on underlying assumptions related to weight and shape. However, patients with anorexia nervosa and dieters appear to experience somewhat similar weightand shape-related automatic thoughts, suggesting
Copyright 2002 John Wiley & Sons, Ltd.

251 that these cognitions should not form the sole focus of treatment. The results presented here suggest that automatic thoughts related to eating should also be carefully considered. Of particular importance, the current ndings also suggest that second order eating-related assumptions, in addition to those concerned with weight and shape, may constitute an additional and important focal area for treatment. The degree of belief and distress associated with assumptions may also be useful areas to investigate and monitor in treatment. The nding that patients with anorexia nervosa had more negative self-beliefs than dieters supports the suggestion that therapy should address negative self-beliefs (Cooper et al., 1998). Schema Focused Therapy (Young, 1990) may constitute an effective form of treatment when working at this level. In addition, given that these beliefs were typically associated with negative early experiences and contain high levels of emotional belief, techniques such as Guided Visual Imagery (Edwards, 1990) may also be useful. The link between negative self-beliefs and dieting is also an important area for therapeutic input, given that such links may be fundamental in driving behaviour (Cooper et al., 1998). Therapeutic time may usefully be spent helping clients challenge the underlying assumptions that link negative self-beliefs and dieting, and in dealing with the schema compensation processes that these underlying assumptions represent. Further investigation is needed into the content of assumptions related to eating, and weight and shape as a means to acceptance by self and others. Their developmental history and their links with dieting behaviour also need to be explored in more detail. Similar studies are required in relation to negative self-beliefs. Although the present study identied an association between negative early experiences and negative self-beliefs, this does not necessarily imply causality. This mechanism continues to require more detailed qualitative and quantitative investigation. Whilst the current study does not disagree with the emphasis on weight- and shape-related assumptions, presented as weight- and shape-related selfschemata in Vitousek and Hollons (1990) model, it is suggested that it may be the presence of eatingrelated assumptions and negative self-beliefs that distinguishes those with anorexia nervosa most clearly from female controls and, in particular, from dieters. As such these cognitions require further investigation.
Clin. Psychol. Psychother. 9, 242252 (2002)

252

H. Turner and M. Cooper


Cooper, M.J., Todd, G., & Wells, A. (1998). Content, origin and consequences of dysfunctional beliefs in anorexia nervosa and bulimia nervosa. Journal of Cognitive Psychotherapy, 12, 213230. Edwards, D.J.A. (1990). Cognitive therapy and the restructuring of early memories through guided imagery. Journal of Cognitive Psychotherapy, 8, 862886. Garner, D.M., & Bemis, K.M. (1982). A cognitivebehavioural approach to anorexia nervosa. Cognitive Therapy and Research, 6, 123150. Garner, D.M., & Garnkel, P.E. (1979). The Eating Attitudes Test: An index of the symptoms of anorexia nervosa. Psychological Medicine, 9, 273279. Rosenberg, M. (1965). Society and adolescent self image. Princeton NJ: Princeton University Press. Spitzer, R., Williams, J., & Gibbons, M. (1996). Instruction manual for the structured clinical interview for DSM-IV. New York: New York State Psychiatric Institute. SPSS Inc. (1997). SPSS advanced statistics 7.5. Chicago: SPSS Inc. Vitousek, K.B., & Hollon, S.D. (1990). The investigation of schematic content and processing in the eating disorders. Cognitive Theory and Research, 14, 191214. Young, J.E. (1990). Cognitive therapy for personality disorders: a schema-focused approach. Sarasota: Professional Resource Exchange.

REFERENCES
American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (4th edn). Washington DC: American Psychiatric Association. Beck, J.S. (1995). Cognitive therapy: basics and beyond. New York: Guilford. Beck, A.T., Epstein, N., Brown, G., & Steer, R.A. (1988). An inventory for measuring clinical anxiety: psychometric properties. Journal of Consulting and Clinical Psychology, 56, 893897. Beck, A., Ward, C., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561571. Burns, D.D. (1980). Feeling good: the mood therapy book. New York: New American Library. Cooper, M.J. (1997). Cognitive theory in anorexia nervosa: a review. Behavioural and Cognitive Psychotherapy, 25, 113145. Cooper, M.J., & Fairburn, C.G. (1992). Thoughts about eating, weight and shape in anorexia nervosa and bulimia nervosa. Behaviour, Research and Therapy, 30, 501511. Cooper, M.J., Cohen-Tov e, E., Todd, G., Wells, A., e & Tov e, M. (1997). The Eating Disorder Belief e Questionnaire: preliminary development. Behaviour, Research and Therapy, 35, 381388.

Copyright 2002 John Wiley & Sons, Ltd.

Clin. Psychol. Psychother. 9, 242252 (2002)

You might also like