Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Eating in anorexia nervosa and bulimia : an application of the tri-partite model of anxiety Buree, Barbara Ursula 1988

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-UBC_1988_A1 B87.pdf [ 7.54MB ]
JSON: 831-1.0097956.json
JSON-LD: 831-1.0097956-ld.json
RDF/XML (Pretty): 831-1.0097956-rdf.xml
RDF/JSON: 831-1.0097956-rdf.json
Turtle: 831-1.0097956-turtle.txt
N-Triples: 831-1.0097956-rdf-ntriples.txt
Original Record: 831-1.0097956-source.json
Full Text

Full Text

EATING IN ANOREXIA NERVOSA AND BULIMIA: AN APPLICATION OF THE TRI-PARTITE MODEL OF ANXIETY by B a r b a r a U r s u l a Buree B.A. hon., The U n i v e r s i t y o f B r i t i s h Columbia, 1977 M.A., The U n i v e r s i t y o f B r i t i s h Columbia, 1981 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY i n THE FACULTY OF GRADUATE STUDIES (Department o f P s y c h o l o g y ) We a c c e p t t h i s t h e s i s as c o n f o r m i n g t o t he r e q u i r e d s t a n d a r d The U n i v e r s i t y o f B r i t i s h Columbia F e b r u a r y 1988 © B a r b a r a U r s u l a Buree, 1988 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 Date H^ryt g Iol DE-6(3/81) Abstract Although many factors have been Implicated in the etiology and maintenance of anorexia nervosa and bulimia, anxiety, particularly in the context of eating, may be cr i t ical . Applying the tri-partite model of anxiety, this study was designed to assess anxiety before, during, and after eating in eating disorder and normal control subjects. The experimental eating procedure was preceded by a neutral task. Four groups of ten female subjects each participated: normal-weight females, restricting-anorexics, bulimic-anorexics, and bulimics. Anxiety was assessed by self-report (ratings of pleasure, arousal, and anxiety), psychophysiological (heart rate and skin conductance) and behavioural (food consumption) measures. Controls reported themselves to be non-anxious throughout the study arid ate almost all of the small test meal. Somewhat surprisingly, physiological arousal (especially heart rate) was high during eating. During the neutral task, heart rate declined slightly in all groups. The eating disorder groups indicated a high level of anxiety throughout the study which showed a trend to increase further during eating. In addition, anorexics and bulimics described dysfunctional beliefs regarding the effects of eating on body shape and weight. Similar to controls, physiological arousal was high during eating. Overall, heart rate proved to be a more useful measure of arousal than skin conductance because many anorexics were hyporesponsive. Restricting-anorexics ate the least amount, bulimic-anorexics ate slightly more, and bulimics ate similar amounts to controls. Thus, food consumption was probably associated with weight status. Several conclusions were drawn. Women with eating disorders have a high level of general anxiety probably because of a conflict between biological i i pressures to eat and fears of weight gain. The high physiological arousal during eating appears to reinforce perceptions of anxiety. The eating disorder groups showed different degrees of concordance among measures of anxiety. Borrowing from the research on phobias, therapeutic procedures such as cognitive therapy, progressive relaxation, and exposure treatments can address the different components of anxiety anorexics and bulimics experience during eating. i i i TABLE OF CONTENTS PAGE ABSTRACT i i LIST OF TABLES v i i i LIST OF FIGURES ix ACKNOWLEDGEMENTS x OVERVIEW 1 Chapter 1 ANOREXIA NERVOSA 3 Historical Overview 3 Demographics and Prevalence 7 Course and Outcome 8 Clinical Features 8 Weight Loss 9 Eating Behaviours 9 Physical Symptoms 10 Personality Factors 10 Cognitive Factors 11 Body Image Disturbances 11 Subgroups of Anorexia Nervosa 12 Contribution of Starvation Symptoms 13 Relationship to Other Disorders 14 Diagnostic Criteria 14 Theories of Anorexia Nervosa 20 Summary 24 BULIMIA 26 Historical Overview 26 Demographics and Prevalence 29 Course 30 i v Clinical Features 30 Binge-Eating 30 Purging 32 Weight History 32 Body Image Disturbances 33 Cognitive Features 33 Personality 34 Physical Sequelae 35 Diagnostic Criteria 35 Theories 39 Set-Point Theory 39 Restraint Theory 42 Sociocultural Factors 43 Feminist Theory 43 The Role of Depression 44 Neurological Factors 45 Energy Balance Model 45 Anxiety Reduction Model 45 Summary 46 ASSESSMENT OF ANXIETY 47 Anxiety and Eating 50 Psychophysiology 50 Eating and Anxiety 52 Summary 54 THE PRESENT STUDY 55 Hypotheses 56 Self-Report of Anxiety 56 Psychophysiological Measures 56 Behavioural Measures 57 Chapter 2 METHOD 58 Subjects 58 Material s 62 Apparatus 63 Procedure 64 v C h a p t e r 3 RESULTS 68 S e l f - R e p o r t Measures 68 S e l f - R e p o r t I n v e n t o r y 1 68 S e l f - R e p o r t I n v e n t o r y 2 ...77 Most Prominent Thought a t S e l f - R e p o r t 1 78 P s y c h o p h y s i o l o g i c a l Measures 78 H e a r t Rate 78 S c o r i n g C r i t e r i a 78 I n t e r r a t e r R e l i a b i l i t y 81 He a r t Rate V a r i a n c e s 81 T o n i c H e a r t Rate 82 H e a r t Rate d u r i n g Task 1 84 He a r t Rate d u r i n g S e l f - R e p o r t Measures 85 S k i n Conductance 87 S c o r i n g C r i t e r i a 87 I n t e r r a t e r R e l i a b i l i t y 88 V a r i a n c e s o f T o n i c S k i n Conductance 88 T o n i c S k i n Conductance 88 S k i n Conductance d u r i n g Task 1 91 S k i n Conductance d u r i n g S e l f - R e p o r t Measures 91 S k i n Conductance Amplitude 94 N o n - S p e c i f i c F l u c t u a t i o n s 96 Recovery H a l f Time 102 B e h a v i o u r a l Measures 106 Q u a n t i t y Eaten 106 P r e f e r e n c e R a t i n g o f Task 2 107 P r e f e r e n c e R a t i n g o f Task 1 107 Chapt e r 4 DISCUSSION 109 S e l f - R e p o r t Measures 110 P s y c h o p h y s i o l o g i c a l Measures 115 He a r t Rate 115 S k i n Conductance 119 B e h a v i o u r a l Measures 123 D i s c o r d a n c e o f Measures 124 v i D i f f e r e n c e s between E a t i n g D i s o r d e r Groups 125 C o n c l u s i o n 126 REFERENCES 135 APPENDICES Appendix 1: D i a g n o s t i c C r i t e r i a f o r A n o r e x i a Nervosa (DSM-III-R)..144 Appendix 2: D i a g n o s t i c C r i t e r i a f o r B u l i m i a Nervosa (DSM-III-R)... 145 Appendix 3: C o n d i t i o n 2 146 Appendix 4: S e l f - R e p o r t I n v e n t o r y 1 148 Appendix 5: S e l f - R e p o r t I n v e n t o r y 2 ( V e r s i o n A) 149 Appendix 6: S e l f - R e p o r t I n v e n t o r y 2 ( V e r s i o n B) 150 Appendix 7: Informed Consent Form 151 Appendix 8: Means o f S e l f - R e p o r t e d P l e a s u r e , A r o u s a l , A n x i e t y 152 Appendix 9: R e s u l t s o f S e l f - R e p o r t I n v e n t o r y 2 155 Appendix 10: Means o f H e a r t Rate P a t t e r n 160 Appendix 11: Means o f S k i n Conductance P a t t e r n 161 v i i LIST OF TABLES PAGE Table 1: Feighner et al . Criteria for Anorexia Nervosa 16 Table 2: Diagnostic Criteria for Anorexia Nervosa (DSM-III) 17 Table 3: Diagnostic Criteria for Bulimia (DSM-III) 36 Table 4: Subject Characteristics 60 Table 5: Summary of Procedure 66 Table 6: Prominent Thoughts during Task 1 79 Table 7: Heart Rate Change Scores for the Self-Report Measures 86 Table 8: Skin Conductance Change Scores for Self-Report Measures 93 Table 9: Skin Conductance Amplitudes for Task 1 and Task 2 95 Table 10: Skin Conductance Fluctuations for Rest Periods 99 Table 11: Skin Conductance Fluctuations during Self-Reports 101 Table 12: Half-Time Recovery Rates during Rest Periods 104 Table 13: Half-Time Recovery Rates during Self-Reports 105 Table 14: Quantity Eaten 106 Table 15: Self-Report Inventory 2: Mean Ratings 156 vi 1 i LIST OF FIGURES PAGE Figure 1: Self-Report: Pleasure 71 Figure 2: Self-Report: Arousal 73 Figure 3: Self-Report: Anxiety 75 Figure 4: Heart Rate Pattern 83 Figure 5: Skin Conductance Pattern 90 Figure 6: Skin Conductance Amplitude 97 Figure 7: Positions of all Groups on Axes of Arousal and Pleasure.... I l l ix ACKNOWLEDGEMENTS I like to thank Dr. Demetrios Papageorgis for his guidance throughout the preparation of this dissertation. Further, I like to extend my thanks to Dr. Leslie Solyom for letting me see his anorexic and bulimic patients. Dr. Robert Hare generously let me use his laboratory and made helpful suggestions. Dr. Ralph Hakstian helped with the statistical analysis. Finally, I thank my husband Peter for his continual support throughout the duration of this study. x OVERVIEW Anorexia nervosa and bulimia are two eating disorders that appear to have become more prevalent in the past decade. Research interest in anorexia nervosa increased in the 1970s and, with the inclusion of diagnostic criteria for bulimia in the DSM-III (American Psychiatric Association, 1980), a considerable body of research now addresses both of these eating disorders. On the surface, anorexia nervosa and bulimia appear to be dissimilar: anorexia nervosa can be characterized by severe weight loss and a refusal to maintain average body weight, whereas in bulimia average body weight is usually maintained, but binge eating episodes followed by various behaviours to counteract the effects of eating are prominent. The two disorders also share many features: indeed, some anorexics develop bulimic symptoms when they gain weight, and, in turn, onset of bulimic episodes is usually preceded by a period of dieting. As the name 'eating disorders' suggests, eating - the process of food intake - is grossly disordered. Anorexics attempt to abstain from eating, placing themselves on a strict diet, and their struggle to limit food intake permeates their lives. Bulimics at times eat normally, but often alternate between strict diets and binge eating episodes where large amounts of food are consumed. Again, normal eating is largely absent, and eating per se leads to anxiety about loss of control over eating. Anorexics also report themselves to be anxious while eating. Food intake is seen to lead to excessive weight gain and feelings of fatness; loss of control over eating is also feared. An analysis of the eating behaviour and especially its associated anxiety in anorexia nervosa and bulimia may further our theoretical understanding of these two disorders; in addition, it is hoped that such 1 knowledge will contribute to the clinical management of such patients. A comparison of the two disorders will add some data to the current debate on diagnostic criteria. Borrowing from research on phobias, the t r i -partite model is used to study anxiety during eating: cognitive, physiological, and behavioural measures are obtained while subjects anticipate and then eat a test meal. 2 Chapter 1 This chapter consists of four main sections. The first section presents an overview of anorexia nervosa. This is followed by a similar, but necessarily sketchier and more speculative, overview of bulimia. Since the present research explores the role of anxiety in eating disorders, the third major section presents the tri-partite model of anxiety and some of the immediately relevant research. Finally, the concluding section of the chapter describes the purpose and goals of the present study, its design, and its specific hypotheses. ANOREXIA NERVOSA Research interest in anorexia nervosa during the past two decades has led to the availability of data on various aspects of this disorder and an increased understanding of the symptoms the anorexic presents with. Such symptoms include disturbed eating behaviours, fear of obesity, body image disturbance, endocrine and metabolic abnormalities, variations in mood, and cognitive distortions. A detailed review of all research on anorexia is beyond the scope of this thesis; instead, only those features directly relevant to the present study will be discussed. This section will give a historical overview of the syndrome, describe the important features of anorexia nervosa and the role of starvation, discuss current diagnostic criteria and problems, and review etiological theories. Historical Overview Although there are accounts of women such as St. Catherine of Siena in the fourteenth century living an ascetic Ideal which resemble anorexic behaviour (Rampling, 1985), the f irst medical account of anorexia nervosa 3 is credited to Richard Morton who in 1689 described a "Nervous Consumption" in the following way: "Mr. Duke's Daughter in St. Mary Axe, in the year 1684, and in the Eighteenth Year of her Age, in the month of July fell into a total suppression of her Monthly Courses from a multitude of Cares and Passions of her Mind, but without any Symptom of the Green-Sickness following upon it. From which time her Appetite began to abate, and her Digestion to be bad... for that she was wont by her studying at Night, and continual poring upon books... she wholly neglected the care of her self for two full Years, t i l l at last being brought to the last degree of a Marasmus, or Consumption... I do not remember that I did ever in all my Practice see one, that was conversant with the Living so much wasted with the greatest degree of a Consumption, (like a Skeleton only clad with skin) yet there was no fever... " (quoted in Silverman, 1985). After administering a variety of medicines, Morton continues: "Upon the use of which (the medication]} she seemed to be much better, but being quickly tired with medicines, she beg'd that the whole Affair might be committed again to Nature, whereupon consuming every day more and more, she was after three Months taken with a fainting Fit and dyed." Morton aptly described many features of anorexia nervosa: the emaciation, the activity despite of i t , and amenorrhea. The next clear descriptions of anorexia nervosa were made almost simultaneously by Gull in England (1874) and Lasegue in France (1873). Gull already had mentioned briefly this condition in 1868 but described it more fully in 1874 when he used the term anorexia nervosa (Strober, 1986). Again, the symptoms are easily recognizable today. For example, Gull (reprinted in Kaufman, 1964) wrote: " . . . I referred to a peculiar form of disease occurring mostly in young women, and characterized by extreme emaciation... The want of appetite is , I believe, due to a morbid mental state... I believe, therefore, that its origin is central and not peripheral... I prefer, however, the more general term, "nervosa", since 4 the disease occurs in males as well as females... The treatment required is obviously that which is fitted for persons of unsound mind. The patients should be fed at regular intervals and surrounded by persons who have moral control over them, relations and friends being generally the worst attendants". Gull also observed the over-activity characteristic of many anorexics: "The patient complained of no pain, but was restless and active. This was in fact a striking expression of the nervous state, for it seemed hardly possible that a body so wasted could undergo the exercise which seemed agreeable". He recommended that attempts should be made to control this activity. Finally, Gull also observed the presence of eating binges: "Occasionally for a day or two the appetite was voracious, but this was very rare and exceptional". A year earlier, Lasegue (1873) had independently given an account of anorexie hvsterique based on his observations of eight cases. He saw as the central feature the patient's belief that food must be avoided, but also recognized other symptoms. Lasegue felt that the course was sufficiently similar in all cases to warrant the description of a typical patient. The following quotes highlight some of Lasegue1s observations (reprinted in Kaufman, 1964): "A young g i r l , between fifteen and twenty years of age, suffers from some emotion which she avows or conceals. . . . At f i rst , she feels uneasiness after food, vague sensations of fullness, suffering... Neither she nor those about her attach any importance to this. The same sensations are repeated during several days, but i f they are slight they are tenacious. The patient thinks to herself that the best remedy for this indefinite and painful uneasiness will be to diminish her food... Gradually she reduces her food, furnishing pretexts, sometimes in a headache, sometimes in temporary distaste, and sometimes in the fear of a recurrence of pain after eating. At the end of some weeks there is no longer a supposed temporary repugnance, but a refusal of food that may be 5 indefinitely prolonged. The disease is now declared... Another ascertained fact is , that so far from muscular power being diminished, this abstinence tends to increase the aptitude for movement. The patient feels more light and active... and is able to pursue a fatiguing l i fe in the world without perceiving the lassitude she would at other times have complained of. . . The patient, when told that she cannot live upon an amount of food that would not support a young infant, replies that it furnishes sufficient nourishment for her, adding that she is neither changed nor thinner... What dominates in the mental condition of the hysterical patient is , above a l l , the state of quietude - I might almost say a condition of contentment truly pathological. Not only does she not sigh for recovery, but she is not ill-pleased with her condition... " Lasegue went on to say that, as weight loss continues, physical symptoms including amenorrhea appear, and that then the patient recognizes the anxiety of the people around regarding her condition and becomes amenable to treatment. Regarding the outcome, Lasegue was positive; he did not encounter any deaths and he writes: "I have,never known the disease to relapse, and once established, the relative or complete cure is maintained." At another place he notes that disturbed eating often persists:"I know patients who ten years after the origin of the affection have not yet recovered the aptitude of eating like other people. Their health is not deeply affected, but their amendment is very far from representing a cure." Notable in both descriptions are the similarities of symptoms observed independently. Both Gull and Lasegue recognized the importance of psychological factors in the development and maintenance of anorexia. Diagnosis of anorexia nervosa as a psychological disorder became more diff icult when Simmonds published in 1914 a case of cachexia associated with pituitary atrophy (see Strober, 1986). In the decades that followed anorexia nervosa was perceived as a physical disorder until 6 Simmond's reports were re-examined and differential diagnostic criteria were established. A variety of psychoanalytic writings appeared but the most influential work on the thinking about anorexia nervosa was contributed by HiIde Bruch (1973) who observed and treated anorexics for over three decades. Demographics and Prevalence Anorexia nervosa appears to occur in industrialized western nations where food is abundant and readily available; with few exceptions, data are unavailable from other parts of the world. Most patients are Caucasians; in the United States and Canada the diagnosis is rarely made in black or hispanic women. Most patients (95%) diagnosed with anorexia nervosa are female and, consequently, only few studies deal with male anorexics. In this study, too, only women participated. Because of the preponderance of women with this diagnosis, female pronouns are used when describing patients. Anorexia nervosa usually develops during adolescence (sometimes earlier) and early adulthood although cases of late onset anorexia nervosa (after age 25) are reported. Patients seem to come more frequently from families with higher socio-economic status, although there may now be a downward trend. In the f irst five decades of this century anorexia nervosa was considered to be a rare disorder, and only few cases were reported. This was partly due to diagnostic difficulties when possible cases were diagnosed as Simmond's disease (see historical overview). The increase in the number of cases since 1960 appears to be due not only to a better diagnostic understanding (weight loss can, for example, also be a symptom of depression), but to reflect a genuine increase in the number of cases, probably due to sociocultural factors. Bruch (1973) described a group of 45 patients. Nineteen of those patients reported onset of anorexia nervosa between 1942 and 1959, and the remainder between 1960 and 1971. 7 Crisp (1976) estimated that for girls over the age of 16 one in 100 in private schools and one in 550 in state schools will be diagnosed as anorexic. Button and Whitehouse (1981) administered an Eating Attitudes Test to 578 students. Their findings showed a one-year prevalence of 1 in 220 in females aged 16 to early 20s. A further 5% of the female students reported anorexic symptoms. Course and Outcome The sequence of the appearance of various symptoms of anorexia nervosa was studied by Beumont, Booth, Abraham, Griffiths, and Turner (1985). Their patients described a progression of the disorder from vague concerns about weight to adoption of dieting leading to an increasing preoccupation with food. Bizarre eating behaviours appeared. Some symptoms appeared more randomly than others; among these were amenorrhea and bulimic episodes. These two latter symptoms do not, therefore, necessarily result from dieting. The outcome of anorexia nervosa is variable. Reviewing the literature, Hsu (1980) found that generally 75% of patients had recovered or improved at follow-up. Improvement is most notable in body weight, whereas results for other indicators of recovery such as menstrual status and social adjustment are less encouraging. More recently, Toner, Garfinkel, and Garner (1986) reported on the long-term outcome (more than 5 years): two-thirds had recovered or improved. Interestingly, restricting and bulimic anorexics (see below, p.12) did not differ regarding their outcome except that bulimic anorexics were more likely to report substance abuse disorders at follow-up. Both groups had an increased lifetime prevalence for anxiety and mood disorders compared with a control group. In summary, although many (up to 75%) anorexics do improve, they are more likely to develop a mood or anxiety di sorder. Clinical Features The following paragraphs will sketch the prominent clinical features 8 of anorexia nervosa. Weight Loss Central to anorexia nervosa is the severe loss of body weight: most sets of diagnostic criteria require a loss of 25% or more. The loss of weight is achieved mainly through dieting; occasionally, anorexics also vomit after eating. Usually, anorexics are of average weight or only slightly overweight when they begin to diet. The diet may be in response to some remark that they have been gaining weight or it may be undertaken in conjunction with friends. Initially, the anorexic loses some weight, feels good about i t , and, unlike her peers, continues to lose more weight. This weight loss is gradual and often unnoticed by the people around her. Anorexics often cannot give plausible explanations why they diet, only that it makes them feel better. As more weight is lost, dieting becomes more important in their daily l i fe . They now desire to be st i l l thinner and start to worry or fear that they will regain the weight. People around the anorexic become concerned, and may start to pressure her to eat. These efforts are resisted and the refusal to eat becomes central to daily living. The loss of weight is egosyntonic: the anorexic is proud of it and denies any problems. Whereas in normal dieting the successful dieter enjoys compliments regarding the weight loss and experiences more self-esteem in relation to peers, the anorexic becomes increasingly more isolated. She had started to diet to be more accepted and admired, but eventually dieting becomes all-pervasive and prevents her from engaging in more social contacts. First of a l l , she is not satisfied with her weight loss and wants it to continue; furthermore, social activities often include eating which has become very problematic for her. Eating Behaviours As the diet becomes more established, unusual eating behaviours and habits are frequently observed. Anorexics will choose only a few foods 9 and exclude all others. They mix foods in unusual ways and spice those mixtures; some will eat discarded food. They prefer to eat alone, taking a long time to eat, often cutting up the food into small pieces. Sometimes eating is followed by vomiting in order to prevent absorption of the food. Laxative abuse is at times present. Many anorexics report being uncomfortable or anxious while eating, as each bite is a struggle. Yet, they often enjoy cooking for others; some are well informed about nutrition, and are often concerned that others may not eat enough. Anorexia is actually a misnomer as appetite is not lost until a large degree of emaciation is present. In fact, anorexics are constantly hungry and need a lot of self-control in order to continue with their dieting. Occasionally, this control is lost and eating binges result, which reinforce the anorexic's fear of further loss of control and weight gain. Physical Symptoms Since menstruation depends partly on the percentage of body fat, a cessation of menses may result from the dieting. Interestingly, amenorrhea is frequently present before significant weight loss occurs and, conversely, menstruation does not always resume right after weight restoration. Much research has dealt with metabolic and endocrine factors (for example, see Garfinkel & Garner, 1982 or Wakeling, 1985). Other observed physical symptoms include bradycardia (resting pulse of less than 60 beats per minute) and lanugo. Anorexics may feel cold and have bruises from fal ls. Sleep is often disturbed. Periods of overactivity are frequently present even after great loss of weight. Personalitv Factors Anorexics are usually described as having been compliant, eager to please, no-problem children. Many are also perfectionist1c, not tolerating failure so that when they diet they want to do it well. Their fear of fatness has an obsessive quality. Other behaviours such as 10 counting calories while eating or doing a prescribed set of exercises appear to be compulsive. Other anorexics, particularly those that experience episodes of bulimia, report impulsive behaviours such as stealing or drug addiction. Cognitive Features Anorexics seem to be of average intelligence or above, many doing well in school. However, cognitive distortions are often present. For example, anorexics often show dichotomous thinking such as 'I am either thin or fat' . Similarly, they often insist on a large margin of safety regarding their weight fearing that i f they, for example, cannot stay under 80 pounds they will quickly reach 130 pounds. Irrational beliefs as described by El l is (1962) often underlie such thinking. Garfinkel and Garner (1982) described such cognitive distortions and Garner and Bemis (1985) developed a therapeutic approach based on Beck's (1976) cognitive therapy but adapted to anorexic thinking. Body Image Pi sturbances Distortions in body image perception are often observed (e.g., Garfinkel & Garner, 1982). Anorexics often overestimate their actual body shape and are unable to see how thin they are. They may be extremely dissatisfied with some aspect of their body. When they gain weight they often feel very uncomfortable (eg., bloated), and become very anxious. Some anorexics have an extremely thin body ideal and enjoy their emaciated look. Body image disturbances have been investigated empirically (for a review see Slade, 1985), but the results are often conflicting. One difficulty derives from measurement problems, i .e. , how to assess body image. However, such body image disturbances must be addressed in therapy for those anorexics where they contribute significantly to the disorder. For example, an anorexic who consistently overestimates her size might learn to rely on more objective indicators ('even though I feel fat I know that my weight at 120 pounds is average 11 for my height1). Bruch (1973) also described anorexics' lack of interoceptive awareness: they find it diff icult to identify feelings of hunger, satiety, or fatigue. They attempt to control bodily functions and fail to recognize required nutritional needs. For example, they might believe that a diet of say 800 calories is sufficient for survival. Subgroups of Anorexia Nervosa As anorexics often appear to be heterogeneous, attempts have been made to identify subgroups. Janet (1919, cited in Garfinkel & Garner, 1982) distinguished two groups; obsessional patients who experienced hunger and hysterical patients who had lost all appetite. Later, Dally (1969, cited in Garfinkel & Garner, 1982) differentiated three groups; group 0, where bulimia and vomiting are often present; group H, with a loss of hunger; and group M with mixed symptoms. More recently, Beumont, George, and Smart (1976) differentiated two groups of anorexics based upon the presence of vomiting and found the groups to differ on several other variables. For example, they found that vomiters were more likely to have been overweight prior to the onset of anorexia and did less well in treatment. Subsequently, researchers have focussed on the presence of bulimia as a criterion for establishing subgroups. Generally, i t has been found (Casper, Eckert, Halmi, Goldberg, & Davis, 1980; Garfinkel & Garner, 1982; Garfinkel, Moldofsky, & Garner, 1980) that bulimic anorexics weigh more, are more impulsive and extroverted, and have a poorer prognosis. They are also more likely to vomit and abuse laxatives, and labile mood is more frequently present. Most developed bulimia early, i .e. , within two years, and in some 1t was already present at the beginning of the weight loss. In an influential paper, G. Russell (1979) described a group of 30 patients who like restricting anorexics showed a morbid fear of fatness but who also engaged in binge eating episodes followed by self-induced vomiting and purging. He coined the term 12 'bulimia nervosa1 to describe such patients. Most had been anorexic previously. Their weight was kept below their healthy weight although weight loss was not as severe as in true anorexia nervosa. Subsequently, research interest was directed at the presence of bulimia in women of normal weight. This syndrome will be described more fully in a later section. The relationship between anorexia nervosa, bulimic subtype and bulimia or bulimia nervosa is a matter of debate: diagnostic difficulties have not been resolved. More comments on these problems will be made later. At the present, there is sufficient evidence to warrant separating restricting and bulimic anorexics at least for research purposes. Contribution of Starvation Symptoms Observations of starving people have shown that many symptoms associated with anorexia nervosa appear to be the result of the low dietary intake. In the classic Minnesota study of human starvation (Keys, Brozek, Henschel, Mickelsen, & Taylor, 1950), 36 healthy male volunteers agreed to undergo a six-month period of semi-starvation and thus furnish data on the effects of starvation and refeeding on physiology and behaviour. After the subjects had achieved a weight loss similar to the one seen in anorexia, many so-to-speak anorexic symptoms were observed. Food was anticipated and thoughts about it took up increasingly more time. Preoccupation with food and eating was accompanied by reduced social and sexual interests. Meals were planned. The men collected recipes and some planned to work in the food industry after the study. Eating habits changed. Most ate slowly, enjoying each bite. Many preferred to eat alone. Unusual food concoctions were prepared, and spices were used excessively. Food had to be hot. Physical symptoms such as bradycardia and low skin temperature appeared. One striking difference was observed regarding physical activity. Whereas anorexics frequently exercise and attempt to be very active, the Minnesota subjects became more passive, attempting to conserve energy. Psychological changes were also observed. 13 For example, scores on the MMPI became elevated during starvation, particularly on the depression scale. It appears that starvation has widespread effects on the human body and accounts for many of the symptoms in anorexia nervosa. Part of any treatment must be the reversal of such symptoms. Interestingly, during refeeding many subjects experienced binge-eating episodes. After a meal they felt more hungry and continued to eat, often to the point of physical discomfort. One can, then, conceptualize the development of bulimic symptoms as a response to prolonged starvation. Relationship to Other Disorders Anorexics often report depression or obsessive thoughts, and the relationship of anorexia nervosa to other psychiatric disorders has been investigated. Bruch (1973) distinguished primary anorexia nervosa from an atypical type where the weight loss is caused by additional psychiatric problems such as schizophrenia (perhaps because of a delusional belief regarding food). In contrast, primary anorexia nervosa is characterized by a fear of fatness and a 'relentless pursuit of thinness'. Solyom, Thomas, Freeman, and Miles (1983) compared restricting and bulimic anorexics to obsessives, agoraphobics, social phobics, and specific phobics on a variety of psychometric measures. They found that both anorexic groups resembled more the obsessive than the phobic subjects. Piran, Kennedy, Garfinkel, and Owens, (1985) reported that 38% of a sample of anorexics and bulimics met criteria for a major mood disorder. It is diff icult to assess depressive symptoms during weight loss, but many therapists consider the use of antidepressant medication if the depression does not resolve after weight gain or i f the patient has a history of mood disorder. Diagnostic Criteria During the past decades several sets of diagnostic criteria for anorexia 14 nervosa have been devised. Such criteria are often closely related to etiological theories. Bruch (1962) described pursuit of thinness accompanied by denial of illness, a sense of ineffectiveness, and body image disturbances as core features of anorexia. Criteria formulated by G. Russell (1970) include self-induced weight loss, fear of fatness, and endocrine disturbances such as amenorrhea. The Feighner (1972) criteria were designed to further research by providing a more reliable diagnosis of anorexia nervosa. Many researchers adopted those criteria (see Table 1), but commonly made three modifications. First of a l l , although anorexia commonly has its onset in adolescence and early adulthood, late onset anorexia does occur and the age limit of 25 seems too restrictive. Anorexia is a misnomer, as real loss of appetite is rarely present at f irst . Finally, the required amount of weight loss necessary for a positive diagnosis remains a matter of debate. Although 25% is a figure commonly used, it excludes adolescent girls who lose less weight when they should have gained weight during a period of growth. S t i l l , the Feighner criteria were an attempt to unify diagnosis. After DSM-III (American Psychiatric Association, 1980) formulated new criteria, researchers in North America used those for diagnosis (see Table 2). As Halmi (1985) points out, problems remain. Again, a weight loss of 25% of body weight is required, although there is no strong research evidence to support this criterion. In addition, someone who is already of average weight at the onset of dieting will experience physical sequelae of starvation earlier (after less weight loss) than someone who is mildly overweight premorbidly. Halmi finds the last criterion 'no known physical illness that would account for the weight loss' unnecessary. The Feighner criteria excluded the presence of other psychiatric disorders for a positive diagnosis whereas the axial system of the DSM-III allows other diagnoses on Axis I and Axis II. DSM-III does not require presence of amenorrhea for a positive diagnosis, although other criteria such as 15 T a b l e 1 F e i g h n e r e t a l . (1972) C r i t e r i a f o r A n o r e x i a Nervosa A. Age a t o n s e t p r i o r t o 25. B. A n o r e x i a w i t h accompanying weight l o s s o f a t l e a s t 25 % o f o r i g i n a l body weight. C. A d i s t o r t e d , i m p l a c a b l e a t t i t u d e toward e a t i n g , f o o d , o r weight t h a t o v e r r i d e s hunger, a d m o n i t i o n s , r e a s s u r a n c e and t h r e a t s : e.g. ( 1 ) d e n i a l o f i l l n e s s w i t h a f a i l u r e t o r e c o g n i z e n u t r i t i o n a l needs; ( 2 ) a p p a r e n t enjoyment i n l o s i n g w e i g h t w i t h o v e r t m a n i f e s t a t i o n t h a t f o o d r e f u s a l i s a p l e a s u r a b l e i n d u l g e n c e ; (3) a d e s i r e d body image o f extreme t h i n n e s s w i t h o v e r t e v i d e n c e t h a t i t 1s re w a r d i n g t o the p a t i e n t t o a c h i e v e and m a i n t a i n t h i s s t a t e ; and ( 4 ) unusual h o a r d i n g o r h a n d l i n g o f food. D. No known me d i c a l I l l n e s s t h a t c o u l d a c c o u n t f o r the a n o r e x i a and weight l o s s . E. No o t h e r known p s y c h i a t r i c d i s o r d e r w i t h p a r t i c u l a r r e f e r e n c e t o p r i m a r y a f f e c t i v e d i s o r d e r s , s c h i z o p h r e n i a , o b s e s s i v e - c o m p u l s i v e and p h o b i c n e u r o s e s . (The assumption i s made t h a t even though i t may appear p h o b i c o r o b s e s s i o n a l , f o o d r e f u s a l a l o n e i s not s u f f i c i e n t t o q u a l i f y f o r o b s e s s i v e - c o m p u l s i v e o r p h o b i c d i s e a s e ) . F. A t l e a s t 2 o f the f o l l o w i n g m a n i f e s t a t i o n s : (1) amennorhea, ( 2 ) lanugo, ( 3 ) b r a d y c a r d i a ( p e r s i s t e n t r e s t i n g p u l s e o f 60 o r l e s s ) , ( 4) p e r i o d s o f o v e r a c t i v i t y , ( 5 ) e p i s o d e s o f b u l i m i a , ( 6 ) v o m i t i n g (may be s e l f -i n d u c e d ) . 16 T a b l e 2 D i a g n o s t i c C r i t e r i a f o r A n o r e x i a Nervosa DSM-III (1980) A. I n t e n s e f e a r o f becoming obese, which does not d i m i n i s h as weight l o s s p r o g r e s s e s . B. D i s t u r b a n c e o f body image, e.g., c l a i m i n g t o " f e e l f a t " even when emaciated. C. Weight l o s s o f a t l e a s t 25% o f o r i g i n a l body weight, o r i f under 18 y e a r s o f age, weight l o s s from o r i g i n a l body weight p l u s p r o j e c t e d weight g a i n e x p e c t e d from growth c h a r t s may be combined t o make the 25%. D. R e f u s a l t o m a i n t a i n body weight o v e r a minimal normal w e i g h t f o r age and h e i g h t . E. No known p h y s i c a l i l l n e s s t h a t would a c c o u n t f o r the weight l o s s . 17 those by G. Russell do and research has pointed to the importance of the role of amenorrhea in anorexia nervosa. Halmi concludes that the presence of amenorrhea should be included in the criteria. (In practice, it is at times diff icult to establish the presence of amenorrhea i f the patient is taking hormonal preparations). Finally, DSM-III does not describe any subgroups of anorexia nervosa, although research evidence indicates the usefulness of distinguishing restricting anorexics from those who also binge. The diagnostic criteria for bulimia (see Table 3, p.36 - to be discussed more fully later) exclude a diagnosis of anorexia nervosa. In practice, most bulimic anorexics easily meet the criteria for bulimia. In this study, diagnosis of anorexia nervosa was based on the modified Feighner criteria ( i .e, without the age restriction and requirement of the loss of appetite) and the DSM-III criteria. The revision of the DSM-III (DSM-III-R. 1987) addresses those diagnostic difficulties (see Appendix 1). The diagnosis of anorexia nervosa now requires the presence of four criteria. Required loss of weight is less than 85% of expected weight for height (although the difficulty of estimating appropriate weight remains). Intense fear of gaining weight must be present. As in the DSM-III, a third criterion refers to disturbance in body image. Finally, amenorrhea is now required (although there is no corresponding criterion for males). Another significant change occurs with the diagnostic criteria for bulimia. This disorder has been renamed bulimia nervosa (see Appendix 2) and, in addition, concurrent diagnoses of anorexia nervosa and bulimia nervosa are now possible. Therefore, a bulimic anorexic can be given both diagnoses i f the bulimia is severe enough to meet the criteria for bulimia nervosa. There is , however, no separate diagnostic category for those anorexics who experience only occasional bulimic episodes and, therefore, do not meet the frequency criterion of bulimia nervosa. 18 One further note regarding the use of various diagnostic criteria concerns early studies (before DSM-III) of 'bulimic-anorexics'. Investigators had been reporting this group to be different on several variables such as presenting weight, minimum weight, weight history, and menstrual status, and were thus viewing it as a subgroup of anorexia nervosa. Depending on the diagnostic criteria used, i f one were to re-evaluate these same subjects, some would now likely be diagnosed as bulimics and not as anorexics (particularly i f amenorrhea is a necessary criterion). Since it is not clear how bulimic anorexics differ from bulimics (a question to be considered more fully later), research results are probably also applicable to at least some bulimics. It also follows, of course, that the diagnostic criteria used in various studies must be taken into account when results are compared. A related question is whether anorexia nervosa is a disorder that occurs on a continuum of dieting and weight concerns. As part of their prevalence study, Button and Fransella (1981) examined high scorers on the Eating Attitudes Test. They found these subjects to be heterogeneous and assigned them to subgroups depending on their preoccupation with food, weight loss techniques, and weight history. Some subjects were classified as 'normal dieters', whereas others had been underweight and resembled more true anorexics. They concluded that many women show a varying degree of concern with weight and dieting without meeting diagnostic criteria for anorexia nervosa. The term 'subclinical anorexia nervosa' was used to describe these women. Excluding 'normal dieters' they found that 5% of the female students were abnormally preoccupied with weight. Their observations are consistent with the continuum hypothesis. Other investigators do not agree. Garner and associates (Garner, Olmsted, Pol ivy, & Garfinkel, 1984) compared anorexic women with weight-preoccupied and not-weight-preoccupied women. Weight-preoccupied women were selected on the basis of their scores on the Drive for 19 Thinness scale of the Eating Disorder Inventory. They were able to distinguish between the groups using the scores on the subscales measuring ineffectiveness, lack of interoceptive awareness, and interpersonal distrust. Examining further the weight-preoccupied women, two subgroups emerged. One group could best be described as normal dieters, whereas the other showed psychopathology similar to anorexia nervosa. The importance of multidimensional evaluation of anorexia nervosa was stressed. Garfinkel and Kaplan (1986) reviewed additional research and concluded that anorexia nervosa is a separate disorder that can be distinguished from other forms of dieting. Theories of Anorexia Nervosa Over the decades several theories have been formulated to explain the development of anorexia nervosa. Since the disorder frequently starts in puberty when sexual maturation occurs, psychoanalytic interpretations have been suggested. When the adolescent feels unable to accept mature sexuality, she regresses to an early stage of development where the sexual instinct is expressed in terms of oral gratification. Fear of oral impregnation leads to food refusal, while episodes of bulimia reflect strong sexual instincts. Psychoanalytic therapy has had, however, limited success as anorexics do not easily form transference relationships. More recently, psychoanalysts have focussed on the development of the self and its deficits (e.g., Goodsitt, 1985). Family interaction approaches, on the other hand, examine family dynamics and the role the anorexic plays in the family (e.g., Sargent, Liebman, & Silver, 1985). Bruch (1973) proposed the following key characteristics of primary anorexia: drive for thinness associated with a disturbance in body image perception which manifests Itself in denial of emaciation; a disturbance in interoceptive awareness, I.e., the inability to accurately perceive bodily sensations such as hunger and satiety, and needs such as 20 food and rest, and thus the anorexic's reliance on external indicators such as numbers on a weight scale; and finally, a strong sense of personal ineffectiveness. Because the anorexic feels so ineffective in many areas of her l i fe , she attempts to exert control by controlling her body; specifically, the weight and shape of her body. She is , init ial ly at least, successful. But this fight for control has to be pursued every day. Episodes of bulimia show the tenuous control the anorexic has. Bruch believes that these difficulties must be resolved for long-term improvement to occur. Crisp (1983) has conceptualized anorexia nervosa as a 'weight phobia1. The anorexics' fear of fatness is , actually, a fear of adult weight or, more precisely, a fear of adult body shape. A mature body shape represents the demands of adulthood such as career choice, adult relationships, and sexuality. The anorexic feels unable to face the new challenges and regresses to an earlier, safe state. Thus, anorexic behaviour can be described as avoidance behaviour. Extensive weight loss leads, biologically, to a prepubertal state. With a child-like body shape, the anorexic can avoid adult challenges. The pre-pubertal state is , however, very unstable as the body strives for growth. The need for food and the drive to eat remain very strong. Occasional binges lead to panic over loss of control. Crisp also discusses to role of anxiety in anorexia nervosa. Anxiety is not free-floating, but is linked to weight gain. Anxiety is described as phobic anxiety which leads to avoidance behaviours such as severe dieting, vomiting, and use of diuretics. It also leads to an avoidance of situations where eating might occur and thus results in social isolation. Any weight gain el icits extreme anxiety as pubertal processes are restarted. Crisp found that most anorexics like to stay just below 40 kg in order to avoid any pubertal development. Any weight loss below that 'magical' number only Increases the safety margin. This anxiety is often denied by the anorexic and is not easily recognized 21 by the therapist. Bemis (1983) discussed the functional relationships between anorexia nervosa and phobias. Although the fear of weight gain has obvious phobic qualities, i f one applies a more theoretical description of phobias to anorexia nervosa differences emerge. Phobics usually express distress about their fears and recognize that those fears are irrational, i .e. , much greater than the perceived danger of the feared object or situation warrants. In contrast, anorexics deny their anxieties regarding weight gain and do not recognize their fear of fatness to be irrational. Whereas the behaviour of phobics is characterized only by avoidance, anorexics also derive great satisfaction from pursuing the opposite of the phobic state. As an example, Bemis cites the elevator phobic who is relieved when leaving an elevator but who does not pursue being away from an elevator as a goal. His behaviour is negatively reinforced; the elevator is being avoided but otherwise forgotten. Anorexics not only avoid fatness, but also actively strive to be thin and experience much pleasure from the loss of weight. This pleasure 1s more than just relief experienced by being further away from the phobic situation. This positive reinforcement of the approach behaviour is unique to anorexia nervosa and not found in phobias. Bemis concludes that the psychopathology of anorexia nervosa 1s characterized by both the fear of fatness and the pleasure of thinness and that both components are powerful motivators. Thus, according to Bemis, anorexia nervosa is not a phobia. In a study referred to earlier, Solyom et al. (1983) had compared anorexics with phobics and obsesslves on several self-report measures and found that anorexics resembled obsesslves rather than phobics. For example, restricting and bulimic anorexics responded similarly on the IPAT Manifest Anxiety Scale, but only the scores of restrictors (not bulimics) were significantly lower than those of 22 obsessives and agoraphobics. Similarly, both anorexic groups reported significantly fewer fears (on the Fear Survey Schedule) than the obsessive and agoraphobic groups. Social and specific phobics did not differ from the anorexic groups on any of those measures. On psychiatrists' ratings of anxiety, anorexics scored lower than obsessives and agoraphobics, but these differences were not significant. In summary, anxiety about weight gain and fatness appears to play an important role in the psychopathology of anorexia nervosa. However, attempts to describe anorexia nervosa as a variant of another disorder have not been successful. Anorexia nervosa is, therefore, best described as a separate disorder with unique features, although resembling other psychiatric disorders on some symptoms. Behaviourists have not developed a theory regarding the etiology of anorexia nervosa, but have rather focussed on the treatment aspects. Using behavioural principles such as positive and negative reinforcement, weight gain is encouraged. Although anorexics often gain weight particularly in hospital settings, such weight gain is not always maintained. Some have criticized the behavioural approach (Bruch, 1974), because anorexics gain weight by responding to environmental constraints and not through self-motivation, thereby further increasing their sense of ineffectiveness. More recently, researchers have investigated the belief structure underlying anorexics' assumptions about themselves and the world and have found cognitive distortions similar to those in depression. As mentioned previously, Garner and Bemis (1985) have identified beliefs where the sense of self-worth and accomplishment is closely tied to body weight and shape. Cognitive distortions such as dichotomous thinking (e.g., 'If I am not thin, then I must be fat') further such beliefs. The goal of therapy then becomes to discover such underlying assumptions, challenge them, and in time replace them with more appropriate ones. Parenthetically, although many concepts are 23 borrowed from Beck's work on depression, the investigators found it necessary to adapt them to the unique features of anorexia nervosa. In an attempt to explain the increase in the number of anorexics, investigators have pointed to sociocultural factors (for example, Bruch, 1973; Garner, Garfinkel, & Olmsted, 1983; Garner, Rockert, Olmsted, Johnson, & Coscina, 1985). Since the 1960s the ideal shape of the female body has become thinner while, at the same time, the average body weight has increased. Increasingly fewer women, therefore, conform to an unrealistic standard. Thinness has become equated with beauty, success, and youth. At the same time, overweight -not just gross obesity- has become stigmatized. Above-average weight has been associated with increased morbidity, although scientific investigations have failed to provide clear evidence (see Bennett & Gurin, 1982). The popular perception that weight is controlled easily is widespread. Therefore, increased weight is perceived to be due to gluttony and lack of self-control, whereas dieting and exercise are seen to be effective means of weight loss. Whereas many men exercise to increase their strength, most women use exercise for weight control. The combination of an increased emphasis on slimness coupled with strong prejudice against being overweight has increased the pressure on women to control their weight. This pressure can precipitate an eating disorder in vulnerable women. Providing more accurate information regarding weight regulation and questioning unrealistic cultural standards can become an important focus in psychotherapy. Summary Although the main features of anorexia nervosa are severe loss of weight coupled with a pursuit of thinness, other psychopathological features such as body image disturbances and cognitive distortions are also present. Variables that lead to the initial loss of weight are not 24 necessarily the same as those that maintain it . Weight loss itself results in characteristic symptoms (due to starvation) including the development of bulimia in at least some cases. Anorexia nervosa is - as the t i t le of a recent book (Garfinkel & Garner, 1982) aptly suggests - a multidimensional disorder. For a variety of reasons the future anorexic embarks on a prolonged diet. When one assesses an individual patient, different antecedents can be hypothesized. For example, a particular girl may f i t Crisp's model of weight phobia, whereas for someone else family conflicts are apparent. Other patients may describe themselves in ways that suggest cognitive distortions. For s t i l l others, body image concerns might appear central to their disorder. Of course, an individual anorexic will rarely ' f i t ' a mold, but often her particular psychopathology can be understood along such theories. Treatment must always be two-fold. First, the starvation symptoms must be reversed and weight gain must occur. Then, in order to promote long-term weight maintenance, the original conflicts that led to the initial weight loss must be resolved so that the anorexic is able to accept herself at a normal weight. Without init ial weight gain meaningful psychotherapy is not possible. Because weight gain involves eating, particularly the normalization of eating, it is useful to study this behaviour more closely. In a way, the battle of anorexia is fought at the table. Eating leads to many conflicts: the body strives for growth, other people exert pressures to eat, but to eat also means to gain weight, to lose control, and to become 'fat ' . Denying oneself food gives a sense of accomplishment, but eventually weight loss results in physical discomfort. The struggle is continual. If one understands better the anxiety experienced by anorexics when confronted with food, one can perhaps apply effective therapeutic measures to make eating easier. 25 BULIMIA Bulimia is a newcomer to the psychiatric disorders: as a specific disorder, it appeared f irst in the DSM-III classification (1980). The category was included to describe an increasing number of women who reported frequent episodes of overeating followed by purging such as vomiting. Beginning with G. Russell (1979), many papers have appeared describing this syndrome. Like anorexia nervosa in the 1970s, bulimia has recently received media attention. Although researchers in many clinics are' undertaking studies to understand the important variables in bulimia, many results point to new questions. Outcome research, particularly long-term, is not available yet. The following section will summarize important aspects of bulimia. First a historical overview will be given. Demographic and prevalence data will be reported. Next, the features of bulimia will be described. The relationship of bulimia to other disorders, especially depression, will be summarized. Current diagnostic criteria and those being used 1n this study will be reviewed. The difficulties of distinguishing between anorexia nervosa and bulimia will be discussed. Finally, theories of the etiology and maintenance of bulimia will be presented. Historical Overview The term bulimia is derived from the greek 1imos = hunger and buos = ox, l i terally meaning hungry as an ox. The use of bulimia as describing great hunger goes back to the fourth century B.C. (for a review, see Ziolko, 1985). There were two different descriptions of bulimia. In the f i rst , bulimia was seen as having a voracious appetite (magna fames lat. = great hunger). The second referred to a condition where one fainted due to hunger. This description was applied to soldiers who when marching in cold climates became exhausted, experienced a voracious appetite, and then fainted. The recommended treatment was bread dipped in wine, 26 whereupon soldiers recovered. The version of bulimia as a fainting spell remained and is s t i l l found in patient descriptions of the 17th century. During the middle ages authors also described bulimia alternating with anorexia. Kynorexia (fames canina lat. = canine hunger, compare 'wolfing down food') was used to describe voracious appetite which led to the ingestion of large quantities of food followed by vomiting. During the Roman Period gluttony was fashionable, and many engaged in binge-eating and vomiting following Seneca's advice (see Ziolko, 1985) 'Edunt ut vomant. vomant ut edunt'. This eating behaviour was not connected to bulimia but seen as a voluntary activity. There was some discussion regarding the differences between kynorexia and bulimia and eventually (around the 17th century) many writers felt it to be the same condition. The term bulimia continued to be used, but the associated fainting spell was largely dropped from the definition although it survived as a subform of bulimia into the 18th century (bulimia syncopalis). Potton (1863, cited in Ziolko, 1985) mentioned amenorrhea in an 18-year old girl who ate eleven to twelve meals a day. Also worth mentioning is that until about the 17th century mostly men were said to be afflicted with bulimia whereas women appeared to be resistant to it . This changed during the 19th and, of course, during the 20th century. By that time the symptom of fainting was excluded from the definition of bulimia. Patients were often described as needing to carry food on their person at all times (e.g. Janet, 1903 cited in Ziolko, 1985). These provisions allowed them to eat frequently -say after a few steps of walking- and enabled them to engage in activity. This particular behaviour seems to have disappeared in the modern-day bulimic. In general, writers noted the distress caused to patients by their strong need to eat. The ability to eat large quantities of food and to enjoy i t was, however, already admired since antiquity and not viewed as bulimia. In the beginning of this century scientific 27 interest in this eating disturbance increased and more articles and descriptions were published. There was, however, a variety of terms being used such as polyphagia, hyperphagia, compulsive eating, addictive eating, bulimarexia, dietary chaos syndrome, hyperorexia nervosa, and so on. Stunkard in 1959 described binge-eating as occurring in obesity. In the 1970s researchers attempted to distinguish subgroups of anorexia nervosa and, as discussed earlier, used the presence of bulimia as an important criterion. One of their findings was the higher current and premorbid weight of those patients. At the same time, reports were published on groups of women with average weight who reported frequent binge-eating episodes followed by purging. For example, Boskind-White and White (1983) coined the term "bulimarexia" to describe such eating disturbances. As mentioned previously, G. Russell in 1979 described 30 patients who had an irresistible urge to overeat which was followed by purging and accompanied by a morbid fear of fatness. Most of these patients had experienced previously a true or cryptic form of anorexia nervosa. He called this disorder bulimia nervosa and conceptualized it as a variant of anorexia nervosa. DSM-III (1980) chose the term "bulimia" to describe binge-eating as a separate disorder from anorexia nervosa. This term will be used throughout this thesis although in the revised version of the DSM-III (DSM-III-R. 1987) the term "bulimia nervosa" is used. The preceding discussion hints at many of the current difficulties with the use of the concept of bulimia. Researchers who use different terms may not refer to similar cases. Its relationship to anorexia nervosa is unclear; indeed, 1t 1s not certain whether bulimia is a separate disorder. Some of these problems will be discussed later. One should keep in mind that the research on bulimia 1s at an early stage and results are tentative. Obviously, the following paragraphs reflect current knowledge. It is , however, possible and useful to describe the 'typical' patient who receives a diagnosis of bulimia although different 28 researchers will emphasize certain features more than others. Demographics and Prevalence Sex Although the proportion of males in patient series varies, it is thought that about 95% of bulimics are females. These figures are similar to anorexia nervosa. Age Bulimic patients are usually in late adolescence and early adulthood. Again, the age distribution is similar to the one in anorexia nervosa although bulimics tend to be a l i t t le older. For example, bulimia has been rarely described in early puberty. Socioeconomic Status Similar to anorexics, bulimics come from upper socioeconomic groups (Fairburn, Cooper, & Cooper, 1986). In most studies, all bulimic women were Caucasian (Schlesier-Stropp, 1984). Katzman and Wolchik (1984) reported that 90% of respondends who participated in a study of bulimia were Caucasians, although they did not provide data for subgroups (bulimics, bingers, and controls). The reported prevalence of bulimia varies in different studies and appears to depend on the population sampled and the diagnostic criteria used to identify cases. Prevalence figures are highest in college populations. For example, Halmi, Falk, and Schwartz (1981) found a prevalence of 13% in 355 college students. Eighty-seven percent of the bulimic students were female. This represented 19% of the females in their sample. Ten percent of the students reported self-induced vomiting after eating. Of these, however, 84.9 % vomited less than once per month, and only 8.5% did so daily. Halmi et al . concluded that binge-eating was a frequent problem among college students and that vomiting was not a necessary symptom for a positive diagnosis. Hart and Ollendick (1985) 29 established prevalence rates in working and university women. They found that although 41% of working women and 69 % of university students reported episodes of binge-eating, the rates dropped as other symptoms of bulimia were added, such as depressed thoughts after binges, and fears of not being able to stop eating voluntarily. Finally, when self-induced vomiting was added to the above symptoms, prevalence rates were 1% for working women and 5% for university students. The authors concluded that although binge-eating might be a common occurrence in young women, the prevalence of bulimia (binge-eating followed by purging) is much lower than popular reports in the media suggest. Further, bulimia appears to be more frequent in student populations than among working women. Similarly, Healy, Conroy, and Walsh (1985) determined prevalence in a sample of 1063 Dublin third-level students. Although 17.7% of males and 37% of females reported an episode of binge-eating, none of the males and only 2.8% of females met all DSM-III criteria. Course Little is known about the long-term course of bulimia. Binge-eating frequently starts in late adolescence and is usually preceded by a period of dietary restriction. Purging behaviours, usually vomiting, are sometimes present when binges f irst occur or even precede them, but commonly become established within two years of binge-eating. Bulimics who seek treatment do so about five years after onset of the binge-eating epi sodes. Clinical Features Binge-Eating Binge-eating refers to the ingestion of very large quantities of food within a short period of time. Bulimics usually binge on fattening, 'forbidden', high carbohydrate foods such as cookies, ice cream, cakes, etc. Episodes typically last one to two hours although they can be of 30 longer duration. Bulimics can usually distinguish between periods of overeating where a favorite food is enjoyed and binges where eating seems to be involuntary and out of control (Fairburn et a l . , 1984). This feeling of loss of control is important in the diagnosis of bulimia. Often large amounts of food are consumed (up to 5000 and 10,000, even 20,000 calories have been reported). The amount eaten seems less important than the experience of loss of control surrounding the eating. Anorexics also sometimes report out of control binges when they, in fact, eat very l i t t le (the 'binge1 might be just three cookies). In these latter cases the use of the term 'binge' is clearly inappropriate despite a subjective feeling of uncontrollable eating. Bulimics usually eat in secrecy and feel guilty and ashamed for having binged. They are often skilled in hiding their bulimic behaviour from family and friends, which explains the often long delay between onset of symptoms and treatment. Unlike anorexics, bulimics experience their symptoms as ego-dystonic. Binges often occur in the evening. They can be preceded by feelings of anxiety, depression, and boredom. Sometimes normal eating el icits binge eating, particularly i f a small amount of a favourite and usually denied food is consumed. Food is eaten rapidly and the urge to binge is so strong that eating is continued despite feelings of being fu l l . Initially, eating is enjoyed as it provides relief from strict dieting or other stressors, but as the binge progresses, anxiety increases due to a strong fear of weight gain. Binges are terminated by lack of food, exhaustion, interruption by others, and purging. Binge eating usually alternates with periods of strict dieting. It 1s, for example, not unusual for a bulimic to skip breakfast and lunch, have perhaps a small snack in the afternoon, and then binge following the evening meal. Some bulimics are unable to eat any normal meals. Sometimes binges may be planned, particularly 1f one is alone and will not be interrupted. 31 Purging Most bulimics use vomiting to prevent weight gain. Vomiting usually follows a binge but might also be present after every meal. Bulimics become very anxious if they are prevented from vomiting after a binge (see Fairburn et a l . , 1984, Johnson, Stuckey, Lewis, & Schwartz, 1982, G. Russell, 1979). Laxative abuse is also common (63% in the Johnson et al . sample) and some bulimics use both vomiting and laxatives for weight control. In the Halmi et al . (1981) study vomiting was correlated with laxative use. Diet pi l ls and diuretics may be used. Nonetheless, the most frequent method of purging is vomiting. Vomiting can be induced in a variety of ways to stimulate the gag reflex, although some bulimics can induce it voluntarily. Some bulimics use marker foods to ensure that all food has been vomited. Often large quantities of liquids are consumed. Weight Historv Most research on bulimia has been directed at normal-weight women. Binge-eating can also be present in obesity (for a review see Gormally, 1984) and will not be considered here. Of course, women who are significantly underweight while bulimic receive a diagnosis of anorexia nervosa, bulimic type. The weight of bulimic women not surprisingly, fluctuates since this is a diagnostic criterion in the DSM-III (1980). In some patient series bulimics are reported to be of slightly lower average weight (e.g. G. Russell, 1979), whereas others report that increased weight is associated with the presence of more diagnostic criteria (Hart & Ollendick, 1985; Ruderman, 1986). Participants in the last two samples were non-patients. Regarding premorbid weight history, investigators have reported varying proportions of subjects who had experienced anorexic episodes previously or, on the other hand, had been overweight prior to the onset of bulimia. No conclusions can be drawn. Several studies (see reviews by M1zes, 1985; Schlesler-Stropp, 1984) reported increased family history of weight problems (up to 68% of f irst 32 degree relatives were obese). Weight history data need to be obtained during assessment so that an optimal weight range can be estimated for a patient. Body Image Disturbances As for anorexia nervosa, body image variables have been investigated in bulimia. Results have confirmed observations similar to anorexia nervosa. Bulimics often report that they feel fat and tend to overestimate their body size. Although overestimation also occurs in normal controls, the amount of overestimation is greater in the bulimic groups (see Ruff, 1982, reviewed in Mizes, 1985). Halmi et al . (1981) reported that a distorted belief of weighing more than actual weight was correlated with the endorsement of bulimia criteria. Bulimics also attempt to lose weight and espouse a thin body ideal. Investigators report that bulimics usually want to weigh significantly less than standard weight tables suggest. G. Russell (1979) found that his patients weighed less than their healthy premorbid weight and were very reluctant to gain weight. He concluded that acceptance of a healthy body weight is important for recovery. Similarly, Freeman, Beach, Davis, and Solyom (1985) found that body image dissatisfaction (perceived size vs ideal size) best predicted relapse six months after treatment. In summary, research suggests that bulimics show body image distortions similar to anorexics and that such distortions must be addressed in therapy. Cognitive Features Researchers have described the bulimics' preoccupation with food, dieting, and weight (Pyle, Mitchell, & Eckert, 1981; G. Russell, 1979). Thoughts about eating and vomiting, and concerns with weight intrude in many activities. Cognitive distortions have also been noted 1n bulimia. Ruderman (1986) found that scores on a bulimia symptom test were negatively correlated with the Rational Beliefs Inventory: i .e. , subjects 33 who reported more bulimic symptoms endorsed rigid, irrational beliefs. Fairburn (1985) has described a cognitive-behavioural treatment program for bulimics. In the cognitive component, the patient is taught to identify and then challenge dysfunctional thoughts and underlying beliefs concerning body shape, weight, and food. Beliefs are examined and evaluated. Finally, cognitive restructuring is used to promote further change. Fairburn also noted that cognitive distortions such as dichotomous thinking may be present. Loro (1984) has also described irrational beliefs held by binge-eaters such as erroneous thinking regarding consumption of various foods and weight gain. Because of perfectionistic thinking, unattainable standards are set which lead to disappointments and often result in a binge. One of the treatment techniques used is cognitive restructuring. These approaches place more emphasis on cognitions in bulimia, although behavioural strategies are usually included. Most therapists recognize the importance of dysfunctional beliefs in bulimia and will address these issues at least implicitly. Personalitv Mizes (1985) reviewed studies on personality factors in bulimia. Many studies have methodological weaknesses such as different selection procedures, the use of global tests, and lack of proper control groups. There is , however, evidence of personality problems in addition to disordered eating. Bulimics are not only more anxious and depressed prior to the onset of a binge, but generally report higher levels of anxiety and depression. Test scores that measure these constructs, such as the Beck Depression Inventory and the MMPI, are often elevated. Low self-esteem and need for approval coupled with unrealistic high standards are common. Similar to bulimic anorexics, bulimics can have problems with impulse control such as suicidal behaviour and alcohol and drug abuse. 34 Physical Sequelae As a result of frequent binging and purging, bulimics need to be assessed for physical problems (see Mitchell, 1986). Gastric complications include gastric dilation after binges. Frequent vomiting can lead to electrolyte imbalances such as hypokalemia and result in cardiac arrythmias. Dental problems are frequent. Fluid retention can follow laxative abuse. Bulimics also complain of fatigue, which is likely caused by inadequate diet. Although bulimics often remain sexually active and ferti le, menstrual disturbances are common. Most of the serious physical symptoms are caused by purging and it is important that patients be assessed for the likelihood of physical risks. Diagnostic Criteria G. Russell (1979) proposed the following set of criteria for bulimia nervosa: an irresistible urge to overeat; use of vomiting or purging to avoid the effects of foods; and a morbid fear of fatness. The last criterion is shared with anorexia nervosa. In North America, DSM-III (1980) formulated diagnostic criteria for bulimia (see Table 3). These criteria were used in the present study to establish a diagnosis of bulimia. The criteria reflect the state of the field in 1980; characteristics of binge eating are described, but aside from the recognition that depressed moods follow binges, other psychopathological features such as possible body image disturbances or a fear of fatness are not included. No weight criterion 1s given, I.e., someone who is very overweight can also receive this diagnosis. Someone who is underweight is more likely to receive a diagnosis of anorexia nervosa. Criterion E, however, precludes concurrent diagnoses of anorexia nervosa and bulimia, making diagnosis of a bulimic Individual who is also underweight diff icult . The relationship between anorexia nervosa and bulimia is not well defined. Regarding the description of 35 Table 3 Diagnostic Criteria for Bulimia DSM-III (1980) A. Recurrent episodes of binge eating (rapid consumption of a large amount of food in a discrete period of time, usually less than two hours). B. At least three of the following: (1) consumption of high-caloric, easily ingested food during a binge (2) inconspicuous eating during a binge (3) termination of such eating episodes by abdominal pain, sleep, social interruption, or self-induced vomiting (4) repeated attempts to lose weight by severely restrictive diets, self-induced vomiting, or use of cathartics or diuretics (5) frequent weight fluctuations greater than ten pounds due to alternating binges and fasts C. Awareness that the eating pattern is abnormal and fear of not being able to stop eating voluntarily. D. Depressed mood and self-deprecating thoughts following eating binges. E. The bulimic episodes are not due to Anorexia Nervosa or any known physical disorder. 36 binge-eating , 'recurrent episodes' are not defined in terms of frequency and individuals with varying clinical severity could be included. Investigators have, therefore, often included a frequency criterion for binges such as twice a month or weekly. In the present study bulimic subjects had to have at least twice monthly binge-eating episodes, although most subjects actually reported more frequent i .e . , weekly and daily episodes. The DSM-III criteria do not require vomiting or purging as a necessary criterion for a positive diagnosis. In the present study 75% of the bulimics vomited following a binge and/or regular meals in order to prevent weight gain. Despite some of the difficulties outlined above, the criteria were the f irst attempt to unify the diagnosis of bulimia. In the revision of the DSM-III (DSM-III-R. 1987; see Appendix 2) some of the above concerns were answered. As mentioned earlier, the disorder has now been renamed bulimia nervosa. This change will reduce confusion about the use of terms since 'bulimia' will now describe only the symptom and not the syndrome. A frequency criterion now excludes occasional (and apparently widespread) binge-eating. An additional criterion recognizes the importance of concerns bulimics have regarding body shape and weight. It is now possible to give concurrent diagnoses of bulimia nervosa and anorexia nervosa. The relationship of bulimia (or now bulimia nervosa) to anorexia nervosa must remain a topic of investigation. Of course, an emaciated restricting anorexic presents quite differently from a normal-weight bulimic (although they share similar concerns regarding body shape and weight), but the distinction becomes increasingly diff icult for bulimics of varying weights. Presently, the underweight bulimic also receives a diagnosis of anorexia nervosa. It is , however, not uncommon for bulimics to move along the diagnostic spectrum: weight may be gained or lost. Garner, Olmsted, and Garfinkel (1985) compared bulimics with different weight histories and weight status on eating and psychological measures. 37 They found the groups to be similar in their attitudes to eating, food, and body shape concerns.The use of weight history failed to distinguish subgroups. The authors suggest the examination of a patient's profile on various test scores to obtain clinical and theoretical information. Longitudinal outcome studies may Identify meaningful subgroups of bulimic patients. In a study of 35 bulimic patients (selected using the G. Russell criteria), Fairburn and Cooper (1984) compared those with prior history of anorexia nervosa (25% of the sample) to those who had never been severely underweight. Aside from weight variables (e.g,. lower current weight), no differences in psychopathology were found. Katzman and Wolchik (1984) compared bulimics (who met all DSM-III criteria with an additional criterion regarding frequency of binges), binge-eaters (who binged eight or more times per month but did not meet all DSM-III criteria) and normal-weight controls on several psychological measures. Binge-eaters scored higher than controls only on binge-eating and restraint measures, whereas bulimics differed, in addition, from the other groups on several measures indicating more depression, poorer body image, higher self-expectations and need for approval, and lower self-esteem. It was concluded that binge-eaters differ from bulimics on measures of psychological functioning. These studies reflect the attempts to delineate meaningful subgroups in bulimia. A frequency criterion appears to be necessary since more frequent binge eating increases disruption in personal l i fe and likely has other sequelae. It should also be noted that G. Russell's criteria include vomiting or purging and possibly describe a more disturbed group of bulimics. The DSM-III (and DSM-III-R) criteria make no such distinction: a bulimic who alternates binging and fasting meets the criteria. Comparative studies of various groups of bulimics along hypothesized variables may provide important information on etiology, 38 psychopathology, and possible subgroups. Theories Although bulimia has only been studied recently, several investigators have formulated theories to describe the etiology and maintenance of this disorder. Concepts from the study of weight regulation and eating behaviour in normals contribute to an understanding of bulimia. In the following sections such concepts will be reviewed before etiological theories of bulimia are presented. Accordingly, set point theory and its role in weight regulation will be summarized. Research on eating behaviour as a function of restraint (dieting) is relevant to onset of binge-eating. As in anorexia nervosa, sociocultural pressures on women to be thin contribute to the seemingly widespread binge-eating. Citing their success with antidepressant medication, some researchers have described bulimia as a variant of a mood disorder. Others have reported successes with anti-convulsive medication. Psychological models of bulimia have been proposed, such as the energy balance model and the anxiety reduction model. The latter model will be described in some detail as 1t 1s directly relevant to the study at hand. Set-Point Theory Several observations about weight regulation have led to the hypothesis of a set-point around which weight is defended (for a review see Bennett & Gurin, 1982). Adult body weight appears to be stable over decades, particularly i f people do not consciously watch their weight and food intake. During a period of reduced food intake (through dieting or food shortages), bodily processes change to slow the loss of weight. After an init ial loss of weight through fluids, basal metabolic rate (BMR) slows to conserve energy. The slowing of BMR explains the plateau many dieters experience. Weight is regained quickly, i f food intake is increased, and often stabilizes at a somewhat higher level than before. If one understands weight regulation in evolutionary terms, one can 39 understand that a higher weight provides protection against the next food shortage. Until recently in the Western world, famines were common (as they indeed continue to be in many parts of the world). Conversely, i f food intake is increased, excess calories are disposed off through thermogenesis or "luxury consumption". If food Intake remains at a higher level over a longer period of time (as studies of overfeeding have shown; e.g. , Sims, 1967, quoted in Bennett & Gurin, 1982), body weight does increase but a large amount of calories is required to maintain the higher level. Subjects typically feel uncomfortable and have to force themselves to eat the large amounts of food. This excess weight is lost quite easily (except in subjects with a family history of diabetes or obesity). Animal models show that some rats with hypothalamic lesions change their average body weight and then defend this new level. Exactly how such a setpoint operates and where i t might be located is not clear; one hypothesis looks at the function of different fat cells (white fat for fat storage and brown fat for thermogenesis). Similarly, the processes which lead to perceptions of hunger and satiety are not well understood. Stomach contractions and fullness and secretions by the liver seem to influence appetite. For the short term, probably the stomach plays an important role in hunger. Changes in metabolic rate follow prolonged periods of changed food intake. Despite the lack of knowledge about the precise physiological components of weight regulation, the set-point theory permits some integration of research findings. The set-point (or, better, set-range since weight is stable within a 2-3 kg range) is not immutable throughout Hfe, but can change. One obvious example is pregnancy and lactation when weight increases dramatically. Set-point also appears to increase somewhat with age (perhaps to provide energy reserves during anticipated sickness) and with a sedentary l i fe style. Overabundance of food might also lead to a higher body weight. 40 Exercise seems to reduce set-point and apparently so does nicotine by increasing metabolic rate (this may partially explain why many smokers gain weight after they quit smoking). Set-point appears to be determined genetically and through early feeding experiences; i .e. , the range of adult weight is given through genetics, whereas the eventual weight within that range reflects environmental influences. Height is normally distributed in the population and there is no reason why weight should not be so as well. Therefore, some people appear to be naturally thin and others fat. Studies on obesity have found that many overweight people appear to behave like average weight people regarding weight regulation: they too defend their weight, only at a higher (and currently unfashionable) level (Keesey, 1986). Although normal weight bulimics are of average weight, they might be underweight with respect to their individual set point. G. Russell (1979), for example, observed that bulimics are invariably thinner than at their premorbid healthy weight. Bruch (1973) described the 'thin-fat' people in similar terms. If bulimics are indeed below their set-point, then biological pressures operate on them to restore their premorbid weight. The body cannot distinguish between dieting -a voluntary starvation- and famine and will react to regain the weight. Studies on starvation -as reviewed in the section on anorexia nervosa- demonstrate some of the mechanisms the body uses to regain weight. During refeeding many volunteers started to binge-eat and reported that eating increased their hunger. Like those volunteers and anorexics, bulimics are preoccupied with food and eating. Furthermore, onset of binge-eating is usually preceded by a period of dieting. All these observations suggest that bulimics indeed behave as if their current body weight was below setpolnt. Explaining these principles (Garner et a l . , 1985) to patients 1s expected to further their understanding of why they experience powerful urges to binge. Important to recovery will be acceptance of healthy body weight at the individual 41 set-point. It is not always easy to estimate set-point, although researchers have suggested (see Garner et a l . , 1985) 90% of highest adult weight. But as long as one remains below setpoint, biological pressures to gain weight will probably predispose the individual to eating problems. Restraint Theory Herman and Pol ivy and associates (Pol ivy, Herman, Olmsted, & Jazwinski, 1984) have studied the eating behaviour of dieters and non-dieters under a variety of conditions. Subjects are classified as high restraint or low restraint eaters on the basis of a Restraint Scale which measures chronic dieting. Generally, they have found that dieters tend to eat less than non-dieters, but that dieters counterregulate, i .e. , eat more under a variety of conditions that seem to disinhibit dieting. For example, following no or a small preload, dieters eat less than non-dieters, but after a large preload dieters consume significantly more. (The usual design of these studies allows the subject unlimited, except for time constraints, access to food under the guise of taste testing). Disinhibition of eating was also demonstrated when subjects were anxious or had consumed alcohol. Recently, the role of hunger was examined (Herman, Polivy, Lank, & Heatherton, 1987). Anxiety reduced hunger in hungry nondieters; and only dieters who were init ial ly hungry experienced increased anxiety. In a theoretical paper, Polivy and Herman (1985) discuss the role of dieting 1n overeating. Reviewing studies, they conclude that binging is normally preceded by dieting. Although acknowledging the presence of physiological factors as potentiating overeating when it occurs (dieters are not necessarily below set-point), they believe the onset of overeating to be mediated cognitively. Overeating in dieters can also be predicted by the belief the dieter has: e.g. , i f a preload was high-caloric or contained alcohol. Dieting itself 42 is mediated cognitively; the dieter restricts food intake because of certain beliefs and must overcome biological pressures. They also found that counterregulatory eating can be prevented through cognitive means such as increasing self-consciousness. They conclude that restrained eaters or dieters impose a cognitive control over food intake (good vs. bad foods) and generally eat less. If this control is disrupted, counterregulation or overeating results. Over-eating in the laboratory is usually considerably less than a binge by a bulimic, but some similarities between restraint eaters and bulimics exist. Bulimics often report that feelings of anxiety or the eating of a small amount of 'forbidden1, usually high caloric food, will trigger a binge. They usually restrict their food intake between binges. It may, therefore, be helpful to apply results from restraint research to bulimia. Sociocultural Factors Sociocultural factors have been proposed to play an important role in the etiology of bulimia (Garner et a l . , 1985; Striegel-Moore, Silberstein, & Rodin, 1986). The same arguments advanced to explain the onset of dieting in anorexia nervosa are applied to bulimia. Briefly, society's emphasis on thinness with a thin body shape representing beauty, 'goodness', and success, coupled with prejudices against obesity results in body image dissatisfaction for many women. Attempting to achieve an unrealistic ideal, women embark on a diet. Unlike restricting anorexics, many are not able to maintain rigid dietary control and do not become severely underweight. Loss of control, however, leads to binge-eating episodes. Issues of cultural ideals must be addressed in therapy. Femini st Theory Elaborating on sociocultural factors, some investigators (see Boskind-White & White, 1983) have described bulimia as resulting from adopting feminine ideals. The bulimic attempts to become the 'perfect' female by being passive, unassertive, and beautiful, i .e. , controlling 43 body shape and weight. These demands are frequently unrealistic. Binge-eating provides a temporary release from such pressures. Purging becomes a 'purification rite' and gives feelings of self-control. Binge-eating becomes a coping mechanism to avoid dealing with a variety of problems. When problems (such as stress or sexual relationships) are dealt with directly, binge-eating 1s no longer necessary. Sociocultural and biological factors interact to increase the probability of binge-eating. A thin body ideal leads to widespread dieting. Reduced food intake might result in a body weight below set point. Disinhibition in a chronic dieter results in counterregulation and overeating; physiological pressures to gain weight increase the tendency to binge. The Role of Depression Bulimics not only feel depressed following binge-eating, but also report higher levels of depression than normals during other times (for example, Hatsukami, Eckert, Mitchell, & Pyle, 1984). Mood disorder in relatives of bulimics appears to occur more frequently than in the normal population. Consequently, some researchers (see Hudson, Pope, & Jonas, 1985) have described bulimia as a variant of a mood disorder and have reported successes with antidepressant medication. Recently, Hinz and Williamson (1987) reviewed the literature and concluded that bulimia is a disorder with unique features, though depressive symptoms are frequently reported (which is not uncommon in psychiatric disorders). Their review showed that many studies lacked a control group of depressed patients. Although antidepressant medication has been shown to be effective in some well-controlled studies, not all antidepressants reduce binge-eating, and a positive treatment response does not necessarily imply common etiology. Treatment of depression might well be important for those bulimics with significant depressive symptomatology or with a coexisting mood disorder, 44 but for other bulimics depression might be secondary to the eating disorder. Neurological Factors Rau and Green (1984) have identified a subgroup of bulimics who report neurological symptoms during binge-eating. Some describe an aura prior to the binge and usually binge-eating is felt to be not only ego-dystonic but also ego-alien. Neurological signs such as dizziness, headaches, and an abnormal EEG are often present. Such patients respond well to anticonvulsive medication. It is therefore suggested, that the assessment of bulimia include an inquiry regarding neurological signs, and that a trial of phenytoin might be warranted, particularly for weight deviant (under- or over-weight) bulimics. Energy Balance Model With respect to the two major components of bulimia - binge-eating and purging - , the energy balance model focusses on the binge-eating. In response to weight gain and fear of obesity, the bulimic attempts to lose weight by adopting a strict diet. When her efforts are not successful, dietary restrictions increase but become impossible to maintain (Johnson, Schlundt, Kelley, & Ruggiero, 1984). Binge-eating results. Treatment focusses on issues of weight control through appropriate dieting and the use of exercise. An initial treatment evaluation found this approach to be promising (Johnson et a l . , 1984). Anxiety Reduction Model Alternatively, purging and, in particular, vomiting is focussed on. Clinicians have often observed that eating and especially binge-eating increases levels of anxiety, whereas vomiting appears to decrease anxiety (for a review see Rosen & Leitenberg, 1985). Vomiting becomes a reinforcer for binge-eating and strengthens the behaviour. Rosen and Leitenberg (1985) cite a study by Abraham and Beumont (1982) where it was found that binge-eating increases after vomiting is learned. Comparing 45 bulimics who vomited with those who used laxatives, Lacey and Gibson (1985) found that the f irst group ate more but weighed less, whereas bulimics who used laxatives ate less but weighed more. Vomiting appears to be an effective weight control mechanism, but at the same time it exacerbates the problem. Vomiting thus becomes the target behaviour for treatment. Similarly to obsessives-compulsives, an exposure plus response prevention approach is used. Bulimics eat progressively more diff icult meals (ordered on a hierarchy) until they have a strong urge to vomit, but are prevented from vomiting while focussing on feelings of anxiety provoked by, e.g. , fullness in the stomach and irrational beliefs regarding weight gain. Eventually, anxiety subsides. Binge-eating episodes are often preceded by feelings of depression, loneliness, or guilt, but become more likely i f vomiting is possible. Comparing bulimics with normal controls using standardized test meals, Rosen, Leitenberg, Gross, and Willmuth (1985) found that bulimics when unable to vomit ate much less than controls and reported greater anxiety. It is suggested that a test meal be part of an assessment of bulimia. In an init ial treatment study using exposure plus response prevention (Leitenberg, Gross, Peterson, & Rosen, 1984) four of five subjects had improved significantly by reducing or stopping binge-eating and vomiting, although only the latter behaviour was treated. This model is applicable only to those bulimics who also vomit (and who possibly constitute a distinct subgroup). Summary Factors that contribute to the etiology and maintenance of anorexia nervosa are also important 1n bulimia. Theories address different aspects of the disorder and therapies vary accordingly. As with anorexia nervosa, i t may prove diff icult to understand bulimia using a single approach; more likely, biological, sociocultural, and psychological factors 46 Interact to determine a particular patient's disorder. It is possible that some bulimics f i t one model of bulimia better than others; eventually, valid subgroups may be identified. Most treatment approaches are, however, two-fold. They deal with the disordered eating and the important factors that e l ic i t and maintain such eating behaviours. The control of disordered eating is particularly important for those bulimics who purge and vomit because of the potentially dangerous consequences. Many clinicians have noted the anxiety experienced by bulimics when eating and binging. As in anorexia nervosa, further investigation of the anxiety during eating appears to be warranted. An understanding of the parameters of the anxiety not only furthers theoretical formulations, but may also have treatment implications. The purpose of the present study is to assess the anxiety experienced by anorexics and bulimics during eating. Anxiety is also part of other disorders, especially the anxiety disorders such as phobias. Adoption of assessment strategies developed when measuring anxiety in phobics provides a helpful framework to measure anxiety during eating experienced by eating disorder subjects. Consequently, a brief review will now summarize measures used in other disorders to assess anxiety. Finally, studies that assessed anxiety during eating will be reviewed; these studies have largely been conducted in the context of formulating the anxiety reduction model of bulimia. ASSESSMENT OF ANXIETY Anxiety occurring in various anxiety disorders such as phobias and obsessive compulsive disorders has been assessed in terms of a three-response system. A behavioural test usually measures avoidance of the feared object or situation; for example, degree of approach to the feared object is measured. Self-reports assess the subjective level of fear; common measures are various fear surveys or a 'fear thermometer'. Finally, physiological reactions such as heart rate, skin conductance, 47 muscle tension, and sweating measure physical arousal in the feared situation (see Lang, 1985; Taylor & Agras, 1980). These three response systems covary imperfectly, and have been described as 'discordant' (Rachman & Hodgson, 1974). A phobic might report strong fear and avoid the situation while showing only moderate physiological arousal. Furthermore, treatments do not necessarily affect all three response systems at the same time: avoidance behaviour may be weakened whereas physical arousal remains high. This phenomenon has been called 'desynchrony' and remains an object of study in fear reduction and the return of fear. It is possible that in simple phobics the responses in the three systems covary more than, for example, in agoraphobics where the phobic situation is less narrowly defined. For example, snake phobics reported greater fear and showed greater physiological arousal when confronted with a snake than socially anxious subjects did (Lang, Levin, Miller, & Kozak, 1983). The socially anxious subjects reported, as expected, greater verbal and physical arousal when delivering a speech; however, the snake phobics also showed increased physiological arousal. Treatment studies have also explored the effect of different treatments on the three response systems. For example, interpersonally anxious subjects who had received behavioural training showed more improvement on behavioural measures, whereas rational restructuring affected verbal reports of fear (McCann, Woolfolk, & Lehrer, 1987). Similarly, treatment for agoraphobics was matched with their individual characteristics. Generally, the matched group improved more than the unmatched group, although cognitive therapy proved to be more successful overall (Mackay & Liddell, 1986). Furthermore, synchronous changes have been associated with treatment response. The role of heart rate appears to be important in fear reduction. For example, Grey, Rachman, and Sartory (1981) found that high heart rate accompanied by report of no fear predicted a return of fear. 48 Experimental studies on heart rate have found that anxious individuals respond with a heart rate increase (acceleration) when the phobic stimulus is presented. Presentation of a neutral or novel stimulus usually el icits heart rate decrease (deceleration) or no change (for example, see Forth & Hare; Hare, 1973; Hare & Blevings, 1975; Lacey, 1967). These heart rate response patterns have been called, respectively, "defensive responses" (when heart rate accelerates after phobic stimuli are presented) and "orienting responses" (when deceleration occurs). Although other physiological measures distinguish the responses (for example, cephalic vasodilation has been associated with the orienting response and cephalic vasoconstriction with the defensive response), heart rate usually discriminates best between the groups and is relatively easy to measure. The usual design of such studies has involved classical conditioning paradigms. Fear-inducing stimuli are usually presented via slides. In such studies the subject sits quietly, only responding with minimal physical activity (perhaps by pushing a switch). However, when heart rate is mediated vagally, heart rate appears to covary with somatic activity. This cardiac-somatic coupling has been demonstrated experimentally by Obrist and associates (Obrist, 1981). Somatic activity during experiments must therefore be taken into account when studying heart rate response to experimental stimuli. As mentioned previously, the subjective component of fear is usually assessed by way of self-report. Developing scales to assess mood and emotion, researchers have included several dimensions such as anger, anxiety, depression, tension and so on (see Lang, 1985). J . Russell (1980) has evaluated this research and concluded that many of these components are intercorrelated. He proposed a circumplex model where emotions can be described satisfactorily with two independent axes: one measuring valence (pleasure-displeasure), the other activation (high-low 49 arousal). In this model, anxiety is characterized by high arousal and low valence. Earlier, J . Russell and Mehrablan (1974) had included a third factor (dominance-submission), which distinguished anxiety from anger (anxiety was associated with low dominance). This third factor accounts, however, for relatively l i t t le variance and, as J . Russell (1980) discussed, does not relate to the emotion itself. Assessment of the two other factors appears to be sufficient to assess emotional states. In summary, research evidence supports the merit of assessing anxiety through the three response systems. In the case of eating disorder subjects, anxiety while eating has been reported frequently. This anxiety appears to interfere with normal eating behaviour whose resumption is important for lasting recovery. Therefore, careful assessment of anxiety during eating can contribute to the understanding of the eating disorders and aid in treatment selection. In addition, i f the anxiety reported by anorexics can be quantified (for example, on self-rating scales), then such scores can perhaps be correlated with other eating disorder symptoms and used to predict outcome and other critical variables. The assessment of anxiety 1n the three response systems will provide information regarding the merit of such assessment procedures. By including three groups of eating disorder patients -restricting and bulimic anorexics, and bulimics- results will point to differences and similarities among groups. A control group of average-weight females will provide information about eating behaviour in normals. Such a study has, therefore, theoretical and practical (I.e., treatment) implications. However, before describing the design of the study and the specific predictions, prior research on anxiety and eating will be reviewed. Anxiety and Eating Psvchophvsioloqv Two studies have investigated the psychophysiology of eating 50 disorders. Calloway, Fonagy, and Wakeling (1983) measured autonomic arousal ( i .e . , skin conductance level and responses) to tones over repeated trials in restricting and bulimic anorexics, bulimics, and control subjects. No significant differences between patient and control groups were found regarding the skin conductance measures. Scores within the patient group were heterogeneous, however: bulimic anorexics and bulimics needed significantly fewer trials to habituate to the tones and showed fewer spontaneous fluctuations. Restricting anorexics showed patterns similar to controls. Subjects had also rated anxiety on visual analog scales. Patients reported higher levels of anxiety than controls. No associations between ratings of anxiety and skin conductance measures were found. The authors concluded that the presence of binging and vomiting is associated with reduced autonomic responding. A study of the effect of imaged food, weight, and body configuration stimuli on skin conductance level and response (Salkind, Fincham, & Silverstone, 1980) found that such stimuli were followed by few skin conductance changes. Two of the subjects were also phobic. In vivo presentation of fear-related stimuli was associated with a large skin conductance response. The authors questioned the description of anorexia nervosa as a phobic disorder. One more interesting point concerns the heterogeneity of the skin conductance resting level. Some patients had a low, others a high resting level. The patient group (beside being small with n = 9) was also heterogeneous: some subjects appeared to be anorexic, while others were of normal weight with eating disturbances. Formal diagnostic criteria were not reported. Skin conductance resting level did not appear to be associated with weight status or presence of binge-eating. Both studies measured only skin conductance and the results were inconclusive. 51 Eating and Anxiety In order to develop a treatment of bulimia by exposure with response prevention, anxiety during eating was measured in five patients (Leitenberg, Gross, Peterson, & Rosen, 1984). Subjects were required to eat various meals in the laboratory and were then prevented from vomiting. In four of the subjects treatment was successful. Throughout the session (while eating and afterwards) subjects rated repeatedly their level of anxiety and urge to vomit; heart rate and amount eaten were also recorded. The study yielded several results. During eating self-reported anxiety and urge to vomit increased. After eating anxiety and urge to vomit declined. Heart rate did not increase and decrease in synchrony with anxiety ratings except in one subject. During treatment sessions two subjects showed increased heart rate (compared with resting level). Heart rate levels during eating declined across treatment sessions. Amount eaten increased over treatment sessions, but decreased anxiety was not necessarily associated with increased food consumption. Also, eating behaviour did not covary with heart rate: two subjects showed a significant negative and one subject a positive correlation. Thought sampling data showed that over treatment negative food thoughts decreased and positive ones increased. There were no consistent correlations between food thoughts and the other measures. The authors concluded that the anxiety reduction model seems to be useful for at least some subjects. Another study (Williamson, Kelley, Davis, Ruggiero, & Veitia, 1985) measured anxiety as assessed by physiological measures in bulimic, obese, and normal-weight control subjects. Baseline recordings of heart rate, skin resistance, EMG, and peripheral vasomotor responses were obtained. Subjects then ate a standardized meal. Following the meal, psychophysiological measures were again recorded. The bulimic subjects showed higher heart rate and other signs of anxiety; however, only for 52 the severe bulimics was heart rate significantly higher after eating. Controls showed greater vasoconstriction and lower skin temperature. The authors concluded that the data do not support the anxiety model very well except perhaps for severe cases of bulimia. Unfortunately, heart rate was not obtained during eating. In a study evaluating gastric motility 1n anorexia nervosa (Holzl & Lautenbacher, 1984) heart rate measures were also obtained during and after a test meal. The groups Included anorexics and two controls. One of the control groups consisted of subjects who were of low body weight, while the other consisted of dieters who were losing weight to an anorexic level. Results showed that gastric motility after a meal is reduced in anorexic subjects. Preliminary data on two subjects showed that heart rate Increased at onset of eating and later decreased. One anorexic reported a high level of tension which declined after eating. By contrast, one control subject reported a lower level of tension which increased briefly during eating. These data suggest that a higher heart rate might be associated with eating, although detailed data of heart rate during eating were not reported. Luck and Wakeling (1980) investigated blood flow in response to a raise of body temperature in anorexic and control subjects. Vasodilation occurred at lower temperatures in anorexic subjects. Following a meal, core temperature Increased 1n anorexics (as did blood flow in most cases). Controls did not show any temperature changes. Heart rate was also reported for the 60th to 90th minute period after eating. Both groups showed a rise 1n heart rate. The authors concluded that a lower threshold of vasodilation and the heat-produdng effect of food are consistent with reports by anorexics that they feel warm after eating (particularly i f they are in warm surroundings). In order to develop a standardized test meal assessment procedure 53 for bulimics, the amount eaten during three different meals was compared in bulimics and normal controls (Rosen, Leitenberg, Gross, & Willmuth, 1985). Subjects were instructed to eat as much as they could comfortably. Bulimics were asked not to vomit for 2 1/2 hours after eating. Under those conditions bulimics ate significantly less than controls. They also reported significantly more anxiety after eating. Food-related thoughts were negative in the bulimic group. Amount eaten also correlated with severity of binging and vomiting and other measures of psychological disturbance. The authors concluded that standardized test meals are useful in the assessment of bulimia. In a preliminary study investigating hunger and satiety in anorexia nervosa, anorexics and controls were given test meals and rated their levels of hunger and satiety (Owen, Halmi, Gibbs, & Smith, 1985). The patient group was small (n = 6) and heterogeneous: four subjects were bulimic anorexics, the other two restrlctors. Results showed that the restricting anorexics (tested at 60% of body weight) consumed the most: more than bulimic anorexics and controls. The authors speculate that they responded to starvation hunger. Hunger and satiety ratings were also disturbed in the anorexic subjects. In a study investigating response to a caloric stimulus (Robinson et a l . , 1983), restricting anorexic and bulimic (including bulimic anorexics and bulimics) patients were asked to complete analogue scales describing their psychologic state prior to and after eating. Bulimics reported significantly more anxiety than control or anorexic subjects on both occasions. The scores of restricting anorexics fell between the two other groups. Summary Only a few studies have investigated anxiety during eating and the results are inconsistent. Lack of controls, patient groups selected by unclear diagnostic criteria, small number of subjects, and presentation 54 of data by averages over long time periods make it diff icult to draw conclusions from these studies. The value of psychophysiological measures is uncertain and some researchers no longer include such measurements. It is , however, apparent that anxiety is an important factor in eating for both anorexics and bulimics. THE PRESENT STUDY The purpose of the present study was to assess anxiety, in terms of the tri-partite model, before, during, and after eating in subjects with clinically diagnosed eating disorders. Four groups of female subjects participated: restricting-anorexics, bulimic-anorexics, bulimics, and normal-weight controls. This allowed for the assessment of the effects of weight status and the presence of bulimia on eating behaviour and the anxiety associated with it . All subjects knew in advance that they would be asked to eat, but the presentation of the food occurred after a neutral task (viewing of slides of landscape paintings and rating them according to preference). Anxiety was assessed for all three aspects of the anxiety model. There were continuous recordings of heart rate and skin conductance. A series of self-report measures of anxiety was obtained. The amount of food actually consumed was also recorded. Additionally, there was an attempt to assess some of the cognitive distortions frequently noted in eating disorders. Finally, a second condition (called condition 2) was included where a few subjects from each group were exposed to the stimuli but were not required to react to them; this condition was intended to aid in the interpretation of the results of the study. The procedure and results of this condition are summarized in Appendix 3. The study was conceived and carried out as an exploratory project. Few specific predictions could be made with any degree of confidence, given the paucity of existing research, especially 1n the case of 55 bulimia. In the case of anxiety, for example, both bulimic and anorexic subjects report that they are anxious when eating, but it is not clear i f this anxiety is experienced in a similar manner in the two groups. For example, anorexics may be more anxious prior to eating whereas bulimics may report increased anxiety after some food is eaten fearing loss of control. Similarly, one might expect anorexics to eat very l i t t le , although at least one study found that the restricting-anorexics ate the most, probably responding to starvation hunger. The results of the psychophysiological studies are also conflicting. It is hoped that this study (with its repeated measures design) will answer some of these questions and, thus, further our understanding of the role anxiety plays in eating disorders. Hypotheses Self-Report of Anxiety Subjects in the eating disorder groups were expected to respond with low levels of pleasure and high levels of arousal and anxiety during the experimental task. In contrast, controls were expected to respond with higher levels of pleasure, moderate arousal (due to task demand), arid low anxiety during the eating and control tasks. No specific predictions were made regarding levels of anxiety reported by anorexics and bulimics during the control task. If anxiety is pervasive, these groups were expected to respond in the same fashion in the two tasks. On the other hand, a neutral task may possibly reduce the level of anxiety. Secondarily, on the specific questionnaire, the eating disorder subjects were expected to show irrational beliefs in their responses to the questions (such as how the food will affect their weight). Psychophysiological Measures During the experimental task 1t was expected that heart rate and skin conductance levels would increase before and during eating and decrease after eating for the eating disorder subjects. An increase in 56 heart rate during eating was also expected for the control subjects because of increased somatic activity; however, this increase was expected to be smaller than in the other groups. During the control task, subjects should show l i t t le change in heart rate. After presentation of the neutral slides, heart rate should remain the same or decrease slightly, indicating an orienting response. Previous studies suggest that the eating disorder subjects may be heterogeneous in resting levels of heart rate and skin conductance. The continuous recording of skin conductance also allowed the measurement of other indices of the response: amplitude, number of skin conductance fluctuations, and recovery half-time. A large amplitude, increased number of fluctuations, and shorter recovery time have been associated with increased arousal. However, since previous studies found bulimic-anorexics to be hyporesponsive, it was unclear how these measures would covary with other measures of arousal. Behavioural Measures Amount of food eaten was the primary behavioural measure. Because the meal was relatively small, i t was expected that controls would eat most or all of it . Restricting-anorexics were expected to eat significantly less since the study was conducted away from a hospital setting and with far fewer pressures to eat. It was diff icult to predict the performance of bulimic-anorexics and bulimics, since no provision to prevent vomiting could be made (although subjects were probably unsure if they could vomit quickly after the study if they wished to do so). Food consumption for subjects characterized by bulimia was, therefore, expected to fall in between the two other groups (restricting-anorexics and controls). As an aside, preference ratings of the slides were not expected to differ among the subjects. However, controls were expected to like the food better than restricting-anorexics. 57 Chapter 2 METHOD Subjects A total of 48 subjects participated in the research, with 40 of these in the main study. Subjects were assigned to one of four groups on the basis of diagnosis. Three groups were formed with subjects with eating disorders; a fourth group consisted of control normal-weight females without history of an eating disorder. The control subjects were recruited through university classes and acquaintances. Normal-weight females were included in the study if they did not report any eating problems such as strong concern with dieting or overeating either currently or in the past. In addition, weight had to be stable within a reasonable range during late adolescence and adulthood, i .e. , no losses or gains greater than 15 % were reported. Subjects with eating disorders were recruited through various treatment faci l i t ies (three large hospitals in Vancouver, Canada). The treating psychiatrist presented a summary of the study to potential subjects. Those who showed interest then received a more detailed description of the study from the investigator. It is therefore impossible to estimate what percentage of eligible subjects actually agreed to participate in the study, and it is quite possible that the sample was in some way biased. Twenty-nine (80.6%) of the 36 eating disorder subjects came from a single treatment facil i ty. Those subjects may not be representative of all possible subjects and may be different in some way that reflects referral policies. The remaining subjects were either patients at one of two other hospitals in the city or were being seen in private practice. It was also observed that some subjects who after initial refusal eventually did 58 participate in the study appeared to do so after showing some weight gain. They explained that they had not 'felt up to it earlier 1 . Presumably, the experimental task - to eat in a laboratory - had appeared very threatening. If this interpretation is correct, then the results can only be generalized to eating disordered patients who are already showing some improvement in their condition. All experimental subjects were under treatment at the time of testing. The three eating disorder groups were described as restricting-anorexic, bulimic-anorexic, and bulimic. Diagnosis of anorexia nervosa was based on the DSM-III and modified Feighner criteria (as reviewed in the previous chapter). Subjects who met the DSM-III (1980) criteria for anorexia nervosa and were restrictors, i .e. , did not have bulimic episodes, formed the restricting-anorexic group. Subjects who met only the DSM-III criteria for bulimia and had never experienced any anorexic episodes, e.g. , lost much body weight, formed the bulimic group. Finally, subjects whose main symptoms were frequent episodes of binge-eating followed by behaviours to counteract the binges were assigned to the bulimic-anorexic group if they, in addition, also met the criteria for anorexia nervosa concurrently or i f they had experienced an anorexic episode in the past but had then become bulimic with a more normal weight. These subjects generally also met the DSM-III criteria for bulimia. This group was therefore a mixed group. A description of subject characteristics is given in Table 4. Mean ages of subjects in each group were similar: a one-way analysis of variance (ANOVA) failed to reach statistical significance, F (3,44)= 0.53, p_ > 0.05. Although the normal-weight subjects had an average of two more years of education than the restricting-anorexics, a one-way analysis of variance yielded a non-significant F-ratio, F (3,37) = 2.73 , p_ > 0.05. The lower number of years of education 1s not surprising in the restricting-anorexic group; several subjects reported that they became 59 Table 4 Subject Characteristics Means (Standard Deviations in Parentheses) Group Age Years of Height Weight Mean Weight BMI (Years) Education (m) (kg) for Height** * * * c* 24.58 14.73 1. .65 59.65 59. ,79 21.77 (5.56) (1.27) ( .05) (8.69) (2.51) RA* 26.33 12.58 1. .61 41.56 59. .41 16.00 (6.51) (1.97) ( .06) (4.44) (1.68) BA* 23.91 13.87 1. .63 50.44 60. .77 18.91 (3.73) (2.42) (• .04) (7.07) (2.63) B* 24.50 13.09 1. .66 61.53 59. ,86 22.26 (3.34) (1.58) (• .08) (9.26) (2.47) ** mean weight for medium frame for height, Metropolitan Life, 1983. * * * Body Mass Index: kg/m2". Group Menstrual Status Mean Duration of Eating Disorder (Years) Patient Status % Inpatient 100% reg. n/a n/a RA 82% amen. 18% irreg. 6.05 58% (6.16) BA 27% amen. 5.50 33% 36% irreg. (2.84) 36% reg. B 90% reg. 4.77 8% 10% irreg. (2.32) * C - Normal-Weight Controls RA - Restricting-Anorexics BA - Bulimic-Anorexics B - Bulimics 60 anorexic while s t i l l in high school and that they had been unable to resume their education. Further, the subjects in each group were also similar in terms of mean height. A one-way ANOVA yielded a non-significant F ratio, F (3,44) = 1.55, p. > 0.05. As expected, the groups differed significantly with respect to body weight, F (3,44) = 17.59, p_ < 0.01. Restricting-anorexic subjects had the lowest body weight. This finding had been expected since one of the diagnostic criteria of anorexia nervosa is substantial weight loss. Tukey comparisons of group means showed that the mean weight of the restricting-anorexics was significantly lower than the weight of the bulimic-anorexics (p_ < 0.05) and of bulimic and control subjects (p_ < 0.01). The weight of the bulimic-anorexics - though higher than the weight of the restricting-anorexics - was st i l l significantly lower than the weight of the control subjects (p. < 0.05) and bulimics (p. < 0.01). The bulimic subjects were similar to the control subjects in terms of weight (p_ > 0.05). The column headed 'Mean Weight for Height' in Table 4 refers to the weight at midpoint of the weight range for medium frame at the respective height and was derived from the Metropolitan Life Tables (1983). As can be seen, both the bulimic group and the control group were of average weight for their height, whereas the anorexic groups were underweight. In addition, the Body Mass Index (BMI) was calculated for each subject. Group means are shown in Table 4. The BMI (weight(kg)/height(m) ) correlates well with body fat and is used in obesity research as a measure of overweight. Generally, a BMI between 25 and 30 (kg/mz) indicates overweight, and a BMI over 30 obesity (Bray, 1986). For women, the lower limit of the acceptable range of the BMI is about 18.7 kg/m . As can be seen, the BMI is consistent with the other weight data. The body mass index of the restricting-anorexics fell well below the acceptable range, whereas the BMI of the bulimic-anorexics fell 61 in the lower end of the acceptable range. The body mass indices of the controls and bulimics were well within the acceptable range. Nine of the 12 restricting-anorexics were amenorrheic; another 2 subjects in this group had only occasional menstrual periods. (The menstrual status of the remaining subject could not be evaluated as she was taking contraceptive p i l ls ) . This difficulty in assessing menstrual status was also present in the other groups: some subjects had to be omitted from the calculations and the numbers shown in Table 4 are based on fewer than 12 subjects. Visual inspection of these data suggests that as average weight increases, menstrual status improves. Parenthetically, it should be noted that, in the bulimic-anorexic group, of those subjects who were experiencing menstrual periods (regular or irregular) all except one had been amenorrheic in the past. In the bulimic group 20 % of the subjects had been amenorrheic in the past. The longest duration of eating disorder occurred in the restricting anorexic group. The large standard deviation indicates a large variation in this group; in fact the range was from 1 to 20 years of illness. The high percentage of inpatients in this group is another reflection of the low weight of those patients. Subjects in the other groups were hospitalized mostly because of uncontrollable vomiting. Finally, it should be noted that the normal weight females reported no menstrual diff icult ies, were mostly weight conscious and dieted occasionally, and exercised moderately. Materials Subjects were asked to complete two tasks: a control task (Task 1) and an experimental task (Task 2). A series of ten slides depicting landscape paintings by various painters such as Van Gogh and Monet provided the stimuli for Task 1. It was hoped that the paintings were pleasant to look at and would thus induce some degree of relaxation in the subjects. When selecting the 62 slides, two criteria were used: paintings had to feature a landscape and, secondly, human figures had to be absent in order to avoid any reference to possible body image disturbance which could have been present in some subjects. For task 2, the stimulus was food. Subjects chose between a "Snickers" candy bar or 50 grams of unsalted "Planters" peanuts. Each snack contained about 275 calories. A few subjects brought their own snack, usually the one they were scheduled to eat in the hospital. In such cases, the caloric value of the food consumed was estimated using the guidelines of Kraus (1983). A Self-Report Inventory (called Self-Report 1) served to assess the emotional state (in terms of levels of pleasure, arousal, and anxiety) of the subjects (Appendix 4). Subjects rated themselves using 15 pairs of adjectives describing emotional states (e.g., pleased -annoyed, excited -calm) (see J . Russell & Mehrabian, 1974). This measure also asked subjects to record their most prominent thought. A second Self-Report Inventory (Self-Report 2, version A) was designed to ask specific questions regarding food and weight while the subjects were anticipating eating (Appendix 5). A third Self-Report Inventory (Self-Report 2, version B) was similar to the previous self-report inventory and contained specific questions regarding eating after consumption of food (Appendix 6). Subject background information was obtained by interview. Apparatus The study was conducted at a psychophysiology laboratory at the University of British Columbia. The laboratory contains a soundproof chamber where the subject remained throughout the recording session. A polygraph outside the chamber was connected to the chamber. In this manner, subject recordings were not influenced by extraneous stimuli, and the experimenter was able to monitor the recordings and make necessary adjustments. An audio system enabled communication with the subject at 63 all times. Part of the experiment was also videotaped in order to assist with the scoring. Psychophysiological recordings were obtained on a Beckman Type R-711 polygraph. Skin conductance was recorded by passing a constant . 5V current through Beckman electrodes covered with a .05M NaCl paste. The electrodes were attached with a double adhesive to the f irst and second digits (medial phalanges) of the non-dominant hand. Subjects were therefore able to write comfortably without disturbing the recording. Heart rate was recorded with a cardiotachometer coupler; Beckman electrodes were covered with Redux electrolyte paste and were attached to the upper and lower ribcage (sternum lead). A pneumometer was attached around the chest to obtain rate and relative size of respiration. These recordings were used as aids in interpreting heart rate and skin conductance responses. Procedure Since the purpose of the study was to assess subjects' responses before, during, and after a control task (viewing and preference rating of neutral slides) and an experimental task (eating and rating the taste of a snack), and to compare those responses, a repeated measures design was used. Psychophysiological recordings were continuous, and self-report measures to assess emotional states were obtained repeatedly. Both the control task (Task 1 - rating of slides) and the experimental task (Task 2 - eating) were completed during a single session. In order to reduce any influence anticipation of task 2 might have had on task 1, an attempt was made to separate the two tasks. Subjects f irst received a detailed description of task 1. After completion of this task, task 2 was described and the relevant materials were given to them. When a subject arrived at the laboratory, she was greeted by the experimenter. She was then shown the laboratory and the equipment and given an explanation of the procedure. The subject then signed an 64 informed consent form (Appendix 7). The subject was then seated in a large chair in the soundproof room and the electrodes were attached. From then on, continuous recordings (heart rate, skin conductance, and respiration) were made. After a stabilization period of approximately 20 minutes, the experiment began. The experimenter left the room and communicated with the subject via the audio system. A total of 40 subjects (10 in each group) participated. Table 5 gives a point summary of the procedure. Part 1 - Control Task: At the start of Part 1, the subject was asked to complete Self-Report 1, a measure of affective state. After completion of Self-Report 1, the subject was asked to relax. Seven minutes after the start of Part 1, the subject was asked to complete a second Self-Report 1. The ten landscape slides were then projected onto a screen. The subject rated each slide for preference. Subsequently, subjects completed a third Self-Report 1. Again, they were asked to relax. Seven minutes later, a fourth Self-Report 1 was completed. This concluded the f irst part of the study. In summary, subjects completed Self-Report 1 twice before the control task and twice after. Instructions to complete a self-report measure were given seven minutes apart. Since subjects took different amounts of time in completing the self-report measures, the length of time spent relaxing between the self-report measures also varied. This time period was usually 3-5 minutes. Part 2 - Experimental Task: The experimenter re-entered the sound-proof room, removed the completed self-report measures from Part 1, and gave the materials for Part 2. The subject was asked to choose the food (candy bar or peanuts) that she would eat. The food was brought into the room and placed beside the subject. Otherwise, the procedure was similar to Part 1. The experimenter left the room. The subject completed again 65 Table 5 Procedure subject arrives in lab procedure is explained informed consent is obtained subject is seated in soundproof chamber electrodes are attached stabilization period of 20 minutes begins subject receives self-report measure materials experimenter leaves room stabilization period ends PART 1: minute 1: subject is asked to complete self-report 1; after completion, subject is asked to relax minute 7: subject is asked to complete self-report 1; when completed, task 1 is presented Task 1: subject views and rates ten landscape slides after completion of task 1: minute 1: subject is asked to complete self-report 1; after completion, subject is asked to relax minute 7: subject is asked to complete self-report 1 experimenter enters room subject receives further self-report materials subject chooses food food is placed beside the subject experimenter leaves room PART 2: minute 1: subject is asked to complete self-report 1 and self-report 2 version A after completion, subject is asked to relax minute 7: subject is asked to complete self-report 1 and self-report 2 version A after completion, task 2 begins Task 2: subject is asked to eat the food. When finished eating, or afte 11 minutes, subject rates the food minute 1: subject is asked to complete self-report 1 and self-report 2 version B after completion, subject is asked to relax minute 7: subject is asked to complete self-report 1 and self-report 2 version B experimenter enters room, removes electrodes. Self-Report 1 and, additionally, Self-Report 2 (version A) which contained specific questions regarding eating. After a period of relaxation the subject was asked to complete the same self-report measures. Then the subject was asked to eat the food. Subjects ate either all or part of the food; i f they took more than 11 minutes to finish eating they could proceed with the rest of the experiment. After eating, they were asked to rate the taste of the food. Finally, the subject completed a third Self-Report 1 and Self-Report 2 (version B), again a questionnaire with specific questions regarding eating. Following a few minutes of relaxation, subjects completed a fourth Self-Report 1 and a second Self-Report 2. This concluded the experiment. As in Part 1, subjects completed self-report measures twice before and twice after the task; again, administration of the self-report measures was separated by seven minutes. After completion of the recording, the experimenter removed the electrodes and collected all materials. Finally, height and weight measurements were obtained. Background information was collected at this time if time permitted. Due to the length of the experiment, most subjects were seen at a later date for the background interview. 67 Chapter 3 RESULTS The results of the study will be presented in three sections. The three sections deal with the main experiment and present the results of the self-report, psychophysiological (heart rate and skin conductance), and behavioural measures. The results from the additional control condition (condition 2) are summarized in appendix 3. Self-Report Measures Self-Report Inventory 1 This self-report inventory asked subjects to describe their emotional state in terms of levels of pleasure, arousal, and anxiety. Subjects rated themselves using 15 word pairs, each of which defined the end-points of a nine point scale. Accordingly, each rating received a score between 1 and 9, with lower scores Indicating lower levels of the emotional state. Since an emotional state was described by several word pairs, the scores of word pairs describing the same emotional state were averaged so that the final rating for each emotional state fell between 1 and 9. Self-Report 1 was administered a total of eight times (twice before and twice after each task), so that there were eight scores for each affective state available for analysis. These scores were subjected to a 4 x 8 with three dependent variables multivariate analysis of variance (MANOVA) with repeated measures using the BMDP (Dixon, 1983) program. The four subject groups (the three eating disorder groups and the control group) served as independent variables and the three emotional states as dependent variables assessed on eight occasions. 68 The analysis showed that the overall effect for groups was significant by Wilks's lambda criterion, = 0.3757, F (9,82.9) = 4.56, P. < 0.0001. Further, the groups differed significantly from each other on each dependent variable. The overall effect for time was also significant: TSQ = 222.093, F (21,16) = 4.70, p. < 0.002. This effect for time was due to the combined dependent variables. Finally, the overall group by time interaction was not significant. For two of the dependent variables (pleasure and arousal), the time by group interaction was not significant; however, the time by group interaction for anxiety was significant, F (14.62,175.46) = 2.52; p. < 0.01. Since the overall multivariate analyses indicated significant results which were due to all dependent variables, univariate analyses were performed for each dependent variable. For these univariate analyses and, i f applicable, for the subsequent comparisons of marginal means (using the Tukey test), the type 1 error rate was adjusted to 0.05/3 or 0.017. First, the scores for pleasure were subjected to a 4 x 8 univariate analysis using the BMDP program. As expected, the results indicated that the four groups reported significantly different levels of pleasure, F (3,36) = 10.14, p_ < 0.0001), and that the effect over time was also significant. Since the analyses also showed that a sphericity test based on Anderson (1958, see Dixon, 1983) was significant (p_ < 0.01) indicating that the covariance matrices failed to meet compound symmetry assumptions, the Greenhouse-Geisser correction was used to report test results. (This procedure - the use of the Greenhouse-Geisser correction when symmetry assumptions are violated - was followed for all subsequent analyses. The BMDP program reduces the degrees of freedom used to 69 establish significance of the F-ratio. Invariably, the tests become more conservative). For the effect over time, a significant F ratio was obtained, F (4.69,168.89) = 3.65, p. < 0.005. As expected, the time by group interaction was not significant. Comparisons of all possible pairs of marginal means for each group were made using the Tukey method. (Generally, marginal means were compared using the Tukey method for the subsequent analyses as well). The highest level of pleasure was reported by the control group. The three eating disorder groups all reported significantly (p. < 0.01) lower levels of pleasure than the control group. However, none of the pairs of means of the eating disorder groups differed from each other (the means and standard deviations for each group are presented in Appendix 8). Next, comparisons of all possible pairs of marginal means for each time were made, again using the Tukey method. For the studentized range statistic, the same degrees of freedom were used as in the Greenhouse-Geisser correction. (Again, this procedure was followed for all analyses when the Greenhouse-Geisser correction had been used. If the number of degrees of freedom used in the correction was not available in the output, p(n-l) was used (see Winer, 1971) to reduce the degrees of freedom; p = number of groups and n = number of subjects in a group). Figure 1 shows a small decline of self-reported pleasure over time; however, only the mean of time 1 (5.52) was significantly higher ( p_ < 0.01) than the reported mean at time 8 (4.83). All other pairs of means failed to reach significance ( p_ > 0.017). In addition, the group means at time 1 were compared with a one-way AN0VA; F (3,36) = 5.0457, p_'< 0.01. Tukey comparisons of group means showed that the mean of the control group (6.75) was significantly higher than the mean reported pleasure of the restrictlng-anorexic group (4.75) (p_ < 0.01) and of the bulimic group (5.15) (p. < 0.05). The difference with the bulimic-anorexic group (5.45) failed to reach significance 70 low PLEASURE high (0.10 > p_ > 0.05). In summary, the control group reported a higher level of pleasure than the eating disorder groups at the start of the experiment and this continued to be the case throughout the study. As the study continued this level declined somewhat. Second, univariate analyses for self-reported arousal were performed in a similar manner as described above for reported pleasure scores. Results showed (see Figure 2, the actual data are also presented in Appendix 8) that the groups reported significantly different levels of arousal, F (3,36) = 8.25, p_ < 0.0003). Tukey comparisons of group means showed that the control group reported the lowest level of arousal (3.79), whereas the restricting and the bulimic anorexic groups reported significantly higher levels (controls vs bulimic-anorexics (5.13) and vs restricting-anorexics (5.78): p_ < 0.017). This finding had been expected. The comparison with the bulimics (4.94) failed to reach significance: 0.05 > p_ > 0.017. The effect for time was also significant, F (4.65,167.38) = 9.68, p_ < 0001). During the second part of the study, subjects reported higher levels of arousal than during the f irst part. The highest mean by all groups was reported after task 2 (eating). This mean (time 7) (5.60) was significantly higher (p_ < 0.01) than all means reported during part 1 and at time 8 (at the end of study). Further, the mean at time 5 (5.45) was higher than the means in task 1 except for the mean immediately after task 1 (time 3, 4.86). Therefore, the highest means within each task were reported following that task. If one compares the groups at time 1, no significant differences emerge (whereas, as mentioned earlier, self-reported pleasure did show differences). Finally, univariate analyses for the third self-report measure -72 Figure 2 Self-Report: Arousal Legend • Control  O Iwflmlct • fllttflctin(j-AflOf«xJc« Self-Reports anxiety - were performed (see Figure 3 and Appendix 8) . The control group reported the lowest level of anxiety (2.72), while the eating disorder groups indicated a significantly higher level of anxiety throughout the experiment, F (3,36) = 15.18, p. < 0.0001 (restricting-anorexics: 6.36; bulimic-anorexics: 5.23; and bulimics: 5.80). Again, this finding had been expected. Comparisons of marginal means between the control group and each eating disorder group showed significant differences (p. < 0.005). The means of the eating disorder groups did not differ from each other. If one looks at the actual scores, the control subjects can be described as non-anxious (scores are less than 5.0), whereas the eating disorder groups report themselves to be anxious. A one-way ANOVA showed that the groups differed already from each other at time 1 ( F (3,36) = 4.0344, p. < 0.05). At the init ial report of anxiety, the control group reported a significantly lower level (3.67) than the bulimic (5.83) and the restricting-anorexic (6.07) groups (p. < 0.05), but not the bulimic-anorexic group (4.93) (p_ > 0.05). The effect for time was also significant, F (4.87,175.46) = 4.87, p_ < 0.001). The lowest level of anxiety was reported at time 3 (4.39), i .e. , following task 1 (rating the slides). This anxiety rating was significantly lower than ratings at time 5 (5.52) and time 7 (5.27); ratings at time 6 (5.13) and time 8 (5.21) also showed a trend of being higher: 0.05 > p_ > .017. The highest rating was given at time 5, at the start of task 2 when subjects had just selected and were given the food. This increase in reported level of anxiety from time 4 (4.61), the end of task 1, to time 5, the beginning of task 2, was significant (p_ < 0.01). Although the interaction of time by group was non-significant in the multivariate analysis, the univariate analysis did find a significant Interaction effect for this dependent variable. Figure 3 shows that as the study progressed the level of anxiety declined somewhat for the 74 Figure 3 Self-Report: Anxiety Legend • Conifoli  O luftmict -i i 1 1—— r r 1 2 3 4 5 6 Self-Reports control subjects but increased for the eating disorder subjects. In summary, control subjects reported a high level of pleasure, a moderate level of arousal, and a low level of anxiety throughout the study. By contrast, the eating disorder subjects reported a lower level of pleasure, and higher levels of arousal and anxiety. For each measure, the restricting-anorexics differed most from the control subjects. Finally, comparing subjects' responses to each task, all task 1 and corresponding task 2 self-report measures were collapsed and a 4 x 2 (groups by tasks) MANOVA was performed. The MANOVA task effect (ISO, = 25.57, F (3,34) = 8.05, p. < 0.01) was significant. The effect was also significant (p_ < 0.01) for each of the three self-report measures analyzed individually. In all cases ratings of anxiety (4.79) and arousal (4.57) were lower during task 1 than during task 2 (anxiety: 5.28, arousal: 5.24) and ratings of pleasure were higher during task 1 (5.39) than during task 2 (4.98). This indicates that overall task 1 was more pleasant, less anxiety provoking and less arousing than task 2. Further, the MANOVA interaction of task by group was significant (Wilks's lambda = 0.53, £ (9,82.9) = 2.74, p. < 0.01). The individual measures were significant in the case of anxiety and pleasure (p. < 0.0167) but not arousal. The interaction in the case of anxiety indicates that anxiety increased slightly on task 2 for all eating disorder groups (task 1 vs task 2: RA - 6.0 vs 6.62; BA - 4.58 vs 5.96; B - 5.54 vs 6.07) and decreased slightly on task 2 for the control subjects (3.04 vs 2.4). In the case of pleasure there was a slight increase during task 2 for control (6.49 vs 6.6) and bulimic-anorexic (4.05 vs 4.48) subjects and a slight decrease for the restricting-anorexics (4.42 vs 3.99) and bulimics (5.16 vs 4.84). These analyses show that as intended the f irst task was more pleasant, less anxiety arousing, and less arousing than the eating task. The 76 interactions furnish some evidence that the eating task had a differential effect on the eating disorder groups. Self-Report Inventory 2 This questionnaire was designed specifically for the study. It allowed subjects to comment on thoughts about eating. Two versions of this inventory were designed. One (version A) asked questions concerning thoughts while anticipating eating, whereas the other (version B) contained questions about positive or negative thoughts following eating. Comments by eating disordered patients, diagnostic criteria, and research on cognitive distortions had guided the selection of questions. However, no psychometric properties of this inventory were established. It is , therefore, inappropriate to interpret the findings in a manner similar to established and better understood questionnaires. S t i l l , statistical analyses were conducted to formally summarize the results. The detailed analyses of each question are presented in Appendix 9. The following paragraph summarizes the findings. As expected, several of the questions were able to distinguish between the different groups. Prior to eating control subjects reported greater ease in thinking about other things than eating than anorexics did. Compared with the eating disorder groups, controls also reported lower anxiety regarding eating, did not believe that the meal would affect their weight, and showed no fear of fatness. Similarly, after eating controls subscribed less to the belief that the meal would increase their weight and showed less fear of fatness. Regarding urges and probability of overeating, both controls and restricting anorexics responded in a similar fashion, reporting few urges to overeat and a low probability of this happening. In summary, it appears that, as had been expected, these questions were useful in distinguishing among groups. 77 Most Prominent Thought at Self-Report 1 At the end of each Self-Report Inventory 1, subjects had been asked to record their most prominent thought. Because task 2 followed task 1 in the same session and subjects knew what to expect, it was possible that especially eating disorder subjects would be anticipating task 2 while performing task 1. It was expected that during task 2 subjects would refer to the task at hand in positive or negative ways (I am hungry, I am anxious about eating etc.) and, in fact, subjects did so frequently. Similar thoughts during task 1 would obviously influence the interpretation of subjects' responses. Therefore, responses were scored as positive if they referred clearly to task 2 (e.g., feeling fat, tense stomach, anxious about this study, references to eating) and as negative if the responses were unrelated to the study (plans for after the study), or referred to task 1 or the study in a general way (e.g., the results of this study). Table 6 summarizes the results of responses during task 1. These results indicate that most subjects were able to concentrate on task 1 and, although they may have been anxious about task 2, did not report such anxiety. Psychophysiological Measures Two major psychophysiological measures were obtained: heart rate and skin conductance. The results of these measures will be discussed in detail in the following sections. A third, respiration, helped interpret the other measures, particularly skin conductance. An attempt was also made to record electrogastric activity; however, the recordings could not be scored reliably because necessary f i l ters had not been available at the time of data collection. Heart Rate Scoring Criteria As described earlier, heart rate was recorded continuously 78 Table 6 Prominent Thoughts during Task 1 (Number of Positive Responses, max. = 10) Group SR 1-1* SR 1-2 SR 1-3 SR 1-4 Controls 2 3 1 0 Restricting Anorexics 2 1 1 3 Bulimic Anorexics 1 0 0 3 Bulimics 4 3 2 1 *SR 1-1 and SR 1-2 were given prior to task 1, the others after task 1. 79 throughout the study. Scoring occurred at selected times. Every time a self-report measure was given, heart rate was scored five seconds prior to the instructions and the last ten seconds before completion of the questionnaire. For part 2 of the study, when subjects received Self-Report Inventory 2, heart rate was scored prior to completion of Self-Report Inventory 2. Therefore, there were eight sets of scores (each set containing a pre- and a during-score) available for analysis. Subjects had also four rest periods, one each before and after each task. A score for average heart rate over three or four minutes during each rest period was obtained, i.e. a total of four scores. In order to assess the responses of subjects immediately following presentation of the slides in terms of heart rate decrease or increase, heart rate was scored for ten seconds after each slide. The f irst score was obtained 1 - 2 seconds after the slide was turned on. A total of ten scores was recorded for each slide. The five second period prior to presentation of a slide provided a baseline measure, so that the scores after the slide could be expressed as a decrease or increase relative to the baseline. The scores for the ten slides were then averaged, so that there was a set of ten scores available for each subject. If a heart rate response occurred during the slide presentation that was clearly attributable to respiration (e.g., the subject took a deep breath as the slide was turned on and heart rate decreased), the responses to that particular slide were omitted from the analysis. (This procedure was also used when scoring heart rate prior to or during self-reports; i f respiration seemed to cause a particular heart rate response, scores were obtained just prior to onset of the respiration response). A final set of scores was obtained during task 2 (eating). The average heart rate of the f i rst , middle, and last minute of eating was obtained, three scores per subject. 80 Interrater Reliability Scoring of the heart rate responses was done by hand and, obviously, there was some room for error. An effort was made to reduce such errors. First, strict criteria for scoring were established and applied. Secondly, records were scored blindly, i .e . , the person who scored the record was unaware of the diagnosis of the subject. And thirdly, because the experimenter scored most (about 80%) of the records, a second scorer scored the results for the remaining records. Interrater reliabil ity was calculated by having the record of one subject scored independently by the two scorers. The Pearson product-moment correlation coefficient was .99 indicating high interrater agreement. Heart Rate Variances When the heart rate scores were entered on scoring sheets, it was observed that the range of scores in the experimental groups, particularly the restricting anorexic group, was larger than in the control group. For example, in the f irst rest period heart rate ranged from 60 to 82 bpm (beats per minute) in the control group, but from 42 to 118 in the restricting-anorexic group, from 54 to 93 in the bulimic-anorexic group, and from 55 to 119 in the bulimic group. The scores for each rest period and during task 2 were tested for homogeneity of variances using Bartlett's test. For both the f irst and second rest period variances were found to be heterogeneous ( Bartlett's test = 11.7, F (3.23) = 3.74, p. < 0.05 for rest period 1 and Bartlett's test = 9.72, F (3.23) = 3.1, p_ < 0.05 for rest period 2). For the other rest periods the F ratios failed to reach significance. Similarly, Bartlett's test was also non-significant for each set of scores obtained during task 2. Visual inspection of the data showed that as the study progressed scores 1n the lower end of the range increased, but scores 1n the upper range remained at the same level. Despite the violation of assumption of homogeneity of variances in the two rest periods, the scores were 81 analyzed without transformation, as the number of subjects in each group was equal. Tonic Heart Rate As described earlier, average heart rate had been scored during each of the four rest periods and during the f i rst , middle, and last minutes of eating. In addition, an average score during the slide task was obtained by calculating the mean heart rate slide baselines. These mean heart rate scores (and the mean skin conductance levels corresponding to the time periods) were analyzed in a 4 x 8 with two dependent variables multivariate analyses of variance (MANOVA). Again, the four subject groups served as independent variables and the two physiological measures (heart rate and skin conductance) as dependent variables assessed on eight occasions. Results did not find any group differences; i .e. , all groups responded in similar fashion. The effect for time was significant: TSQ = 632.897; £ (14,23) = 28.88, p_ < 0.01. The group by time interaction was also non-significant. Since the multivariate analyses indicated significant effects for time, univariate analyses were performed for each dependent variable. The type 1 error rate was adjusted to 0.05/2 or 0.025 for the univariate significance tests and any subsequent Tukey comparisons of marginal means. The univariate analyses of tonic heart rate (the analyses for skin conductance will be presented later) did not show any group differences, but a significant effect for time: F (4,68.61) = 12.12, p_ < 0.01. The group by time interaction was not significant. The changes over time are illustrated in Figure 4 (the actual means and standard deviations are presented in Appendix 10). Comparison of marginal means showed that heart rate during eating was significantly higher (p. < 0.01) than during the 82 90-, 64 84 £ «J SO cc © X 76 74 72 W Figure 4 Heart Rate Pattern Legend I Control! O BudmJci 70 -L-f 6 , M , W " ""2 r.i!3 7^1 Time Periods rest periods or the slide task. Heart rate increased during eating; i .e. , heart rate during the middle (84.5 bpm) and the end of eating (85.2 bpm) was significantly higher (p. < 0.01) than during the f irst minute of eating (82.0 bpm). In fact, the highest heart rate was observed at the end of eating. Heart rate during the last rest period (after eating) (76.6 bpm) was significantly higher (p. < 0.025) than during the rest period prior to eating (74.1). Heart rate during rest period 3 was slightly lower than the heart rates during task 1 and, thus, did not show a hypothesized increase prior to eating. Heart rate during the slides (76.7 bpm) was somewhat higher than rest periods 1 (74.8 bpm), 2 (74.6 bpm), and 3; however, only the difference with rest period 3 reached significance (p. < 0.025). In summary, expected differences between groups were not found. All groups showed a similar heart rate pattern throughout the study. Heart rate was clearly highest during eating and was lowest during the rest periods. Heart Rate during Task 1 As described earlier, heart rate was scored for a ten second period after onset of the slides and expressed as deviation from a five second baseline measured prior to onset of each slide. The scores were then averaged for the ten slides so that a total of ten scores per subject were available for analysis. These scores were subjected to a 4 x 10 (group by times) repeated measures analysis. It had been expected that the scores would show a decrease of heart rate compared with the baseline and that all groups would show this decrease. The analyses confirmed this expectation. Neither group nor time differences were found; all groups decreased their heart rate following presentation of a slide. The control group showed an average decrease of 1.611 bpm (beats per minute), the restricting-anorexics 1.238 bpm, the bulimic-anorexics 0.288 bpm, and the bulimic group 0.347 bpm. 84 Heart Rate during Self-Report Measures Heart rate was also measured prior to and during the final 10 seconds of each self-report measure. It had been expected that heart rate would increase when subjects completed a self-report measure. One difficulty in scoring arose at time 7, the self-report after eating. Because the subjects completed the self-report measures immediately after the task, no convenient pre-score ( i .e . , baseline) was available. It was decided to use the baseline score prior to self-report 1 at time 6 (prior to task 2). A pre-score could be more readily obtained after task 1 as there was an interval between completion of the task (rating of the slides) and the start of self-report 3. The obtained set of scores (a pre-score and a during score for each self-report, thus a set of eight scores per subject) were subjected to a 4 x 8 repeated measures analysis of covariance. The four groups served as independent variables and the heart rate scores during the self-report measures as the dependent variables assessed on eight occasions. The pre-score of each self-report measure was the covarlate for each during score. Although the analysis was univariate, the type 1 error rate was adjusted to 0.05/2 or 0.025 since corresponding analyses were also performed for skin conductance levels (these will be reported later). Results did not find any group differences. The effect for time also failed to reach significance (0.05 > p. > 0.025). The group by time interaction also was non-significant. In addition, the difference scores (during score - pre-score) were also calculated and the means and standard deviations are presented in Table 7. Overall, the change scores were positive, i .e. , as expected heart rate during self-report measures was higher than prior to i t . Table 7 shows that the largest increase occurred at time 7 (after eating). However, this score was not a true change score since the pre-score was obtained prior to task 2. Previous analyses showed that heart 85 Table 7 Heart Rate Change Scores (bpm) for the Self-Report Measures Means for each Group (Standard Deviations in Parentheses) Group c* RA* BA* B* Margin; Total Self-Report 1 2.3 (5.4) 3.8 (3.7) 4.4 (7.1) 3.3 (6.6) 3.45 Self-Report 2 2.9 (5.3) 2.4 (6.7) 1.6 (6.4) 0.2 (3.9) 1.76 Self-Report 3 -1.0 (7.1) 2.3 (5.3) -0.8 (4.2) -1.2 (6.0) -0.18 Self-Report 4 0.2 (4.2) 4.9 (4.5) 2.6 (4.7) 3.4 (5.6) 2.78 Self-Report (Version A) 5 1.9 (4.7) 1.6 (5.7) -0.7 (4.7) 0.2 (5.6) 0.75 Self-Report (Version A) 6 -0.1 (6.0) 3.6 (5.7) 2.8 (5.0) 2.6 (5.8) 2.23 Self-Report (Version B) 7 4.4 (4.8) 6.2 (3.0) 4.3 (6.9) 4.5 (4.0) 4.85 Self-Report (Version B) 8 1.1 (5.0) 0.2 (7.0) 1.8 (5.5) -1.3 (6.5) 0.45 Marginal Total. 1.5 3.1 2.0 1.5 * C - Controls RA - Restricting-Anorexics BA - Bulimic-Anorexics B - Bulimics rate increased during eating and consequently, the derived score at time 7 was inflated. Overall i t appears that the scores for self-reports during the two tasks were comparable. Skin Conductance Scoring Criteria The second major psychophysiological measure obtained was skin conductance. Since the recording of skin conductance was continuous, readings could be obtained at any point during the study. Various aspects of skin conductance were measured. First, tonic levels were recorded at different periods throughout the study similar to heart rate. During rest periods skin conductance levels were measured at eight points from 20 seconds after onset of the rest period to the beginning of the next self-report measure. These eight measurements were averaged to obtain a single score for each rest period. During task 2 three scores of tonic levels were obtained, one in each interval where heart rate had been measured. During task 1, tonic level was measured in a five second interval prior to onset of each slide and the average tonic level of all slides was calculated. As with heart rate, tonic level was measured five seconds prior to onset of completing a questionnaire and during the last ten seconds. In addition to tonic levels, amplitude of the response was scored when subjects were instructed to start the following activities: to complete a self-report measure, to get ready for task 1, and to begin task 2. In addition, amplitude of responses to the slides was also scored. The amplitude of a response was scored if it was Ajjimhos or larger. Smaller responses were not scored. One difficulty, however, arose. Large responses could not always be scored accurately 1f the response was larger than the space available for recording. In such cases, the measured response was smaller than the actual response. 87 Further, the number of non-specific fluctuations were measured for the following periods: during self-report measures and task 2 (thirty seconds after onset of the task to end of task) and during rest periods (starting twenty seconds after onset of the rest period). Fluctuations were scored if they were larger than .l^jimhos. The obtained number of fluctuations in each interval was then standardized on a sixty second basis so that the score for the analyses represents the average number of fluctuations per minute in the interval. Finally, recovery half time was measured when non-specific fluctuations occurred. These measures were averaged for each interval so that only one number per interval entered the analyses. As with heart rate, responses that were due to respiration were not included in the scoring. Interrater Reliabilitv Skin conductance was scored by hand and, more so than with heart rate, measurement error was possible. For example, with large responses the base tonic level changes. Therefore, interrater reliabil ity was assessed by having tonic skin conductance levels from one subject scored independently by two raters and correlating the obtained scores. A Pearson product moment correlation coefficient of r = .99 was obtained indicating high interrater agreement. Variances of Tonic Skin Conductance As with heart rate, the set of scores obtained in each rest period and during task 2 were tested for homogeneity of variances. Bartlett's tests were all non-significant indicating that the variances in each group were homogeneous. Tonic Skin Conductance The mean tonic skin conductance levels during the rest periods, the eating task, and slide task had served as a dependent variable in a 4 x 8 repeated measures multivariate analysis of variance (tonic heart rate was 88 the other dependent variable). As reported earlier, results did not find any group differences, but a significant effect for time. Therefore, a 4 x 8 (groups by times) univariate analysis of variance was performed. As with the heart rate analyses, type 1 error rate was set at 0.05/2 or 0.025. As expected, results did not find significant group differences, but a significant effect for time; F (1.91,68.61) = 12.12, p, < 0.01. Figure 5 shows the pattern of tonic skin conductance levels for all groups (the actual means and standard deviations are presented in Appendix 11). Comparison of marginal means showed that skin conductance was significantly higher (p. < 0.01) at the beginning of eating (3.97 increased when the rest periods are compared. Skin conductance during preceding and subsequent rest periods, this increase was, however, non-significant. The group by time interaction was not significant. In summary, skin conductance levels increased during the study in all groups. The highest level was at minute 1 of task 2. The changes were particularly evident in the restricting-anorexic group. In a manner similar to the self-report measures, a 4 x 2 (groups by tasks) MANOVA was performed contrasting heart rate and skin conductance responses during task 1 with those during task 2. The task effect was significant; JiQ = 239.574, F (2,35) = 116.46 and p_ < 0.01. The effect was also significant for both dependent variables. Both heart rate and skin conductance levels were lower during task 1 than task 2, Indicating than at all other time periods including during the middle (3.44 and the end of eating (3.46 /^mhos). Skin conductance levels also rest period 4 (3.45 Jimhos) was significantly higher than skin conductance during rest period 1 (2.88 yU.nfihos, p_ < 0.01) and during rest period 2 (2.99 ^mhos, p. < 0.025). The comparison with rest period 3 (3.29 y^mhos) failed to reach significance. Skin conductance level during the slide task (3.19 //mhos) was slightly higher than during the 89 Figure 5 Skin Conductance Pattern o 3. 5.1 -i 4.7-4.3 -3.« -3.5-3.1-2.7-2.3 -1.5-1.1--O 8jjflnuct A 8ufiroic-Anor«»ic« • floulcting-Anon»lc» reit 1 |ii<5«i felt 2 ceil 3 e«t 1 «it 2 eat 3 r« i t 4 Time Periods 90 that physiological arousal during task 1 was lower than during task 2. Skin Conductance during Task 1 Tonic skin conductance levels of the pre-slide time periods were averaged for each subject and analyzed with a 4 -way analysis of variance. No specific expectations had been formulated; it was simply of interest to see if the tonic levels would differ. The F -ratio was significant, F (3,36) = 3.10, p. < 0.05. The lowest mean tonic level was obtained in the restricting-anorexic group (mean = 1.92 ^xmhos, standard deviation = 1.58). This mean was significantly lower (p_ < 0.05) than the mean tonic level in the bulimic group (mean = 4.171 ^mhos, standard deviation = 2.1). The means in the other two groups were also higher than the mean in the restricting-anorexic group but these differences were non-significant (for the control group: mean = 3.65 jjumkos, standard deviation = 1.85; and for the bul1m1c-anorex1c group: mean = 2.99 ^Lmhos, standard deviation =1.3). Skin Conductance during Self-Reports Average skin conductance levels were also scored prior to and during the self-report measures. As with heart rate, the scores during the self-report were analyzed in a 4 x 8 repeated measures analysis of covariance with the pre-scores of each self-report serving as the covariate of the during score. Type 1 error rate was again reduced to 0.05/2 or 0.025. Results of the analysis showed no significant group differences, but a significant effect for time: F (7,251) = 5.71, p. < 0.01. The group by time interaction was not significant. Comparison of marginal means showed that skin conductance level was lowest (2.9 /^.mhos) during self-report 5 after subjects had received the food and Instructions for task 2. This level was significantly lower (p_ < 0.025) than during all other self-report measures (which ranged from 3.52 y^ Lmhos to 3.60 y/,mhos) except during self-reports 1 and 3. Those were also preceded by a period of 91 activity ( initial description of the task and the slide task). It appears that the size of skin conductance level during a self-report reflects partly the level of preceding activity. As with heart rate analyses, actual change scores during the self-report measures (during score - pre-score) were calculated. The means and standard deviations are presented in Table 8. Overall, change scores were positive except for time 5, the beginning of task 2. Self-report 5 followed a period of activity. It is likely that the pre-score of self-report 5 was inflated. Unlike heart rate scores, change scores of self-report 7 (after eating) were positive; however, previous analysis showed that skin conductance levels declined during eating after an initial increase. Heart rate, on the other hand, increased as task 2 proceeded. Overall, i t appears that the size of the change scores is influenced by the level of activity preceding the self-report measure. The analyses of tonic skin conductance levels throughout the task were of major interest. Skin conductance responses can also be evaluated by scoring amplitudes of responses to stimuli and the number of spontaneous fluctuations during time periods. These components have, however, measurement problems. Amplitude could not be scored accurately in all cases: a small response was more likely to be scored accurately than a larger response. Scoring errors tended to be conservative (a larger response was likely recorded smaller than its actual size). It also was diff icult to analyze spontaneous fluctuations (and consequently recovery rates) since a different proportion of subjects in each group were hyporesponslve. Since research in this area is sparse, it was of interest to further examine skin conductance responses. Therefore, univariate analyses for skin conductance amplitudes and spontaneous fluctuations were performed with a traditional type 1 error rate of 0.05. Findings from these analyses must be Interpreted cautiously since overall error rate is likely inflated. 92 Table 8 Skin Conductance Change Scores (urnhos) for the Self-Report Measures Means for each Group (Standard Deviations in Parentheses) Group C* RA* BA* B* Marginal Totals. Self-Report 1 0.004 (0.332) -0.052 (0.292) 0.063 (0.350) -0.129 (0.337) -0. ,029 Self-Report 2 0.714 (0.792) 0.389 (0.738) 0.501 (0.570) 0.347 (0.896) 0. .488 Self-Report 3 -0.120 (0.511) 0.140 (0.461) 0.068 (0.357) -0.047 (0.686) 0. ,010 Self-Report 4 0.813 (1.238) 0.055 (0.232) 0.345 (0.546) 0.406 (0.453) 0. ,405 Self-Report (Version A) 5 -0.507 (0.298) 0.021 (0.595) -0.484 (0.594) -0.249 (0.822) -0. .305 Self-Report (Version A) 6 0.640 (1.174) 0.743 (1.290) 0.306 (0.556) 0.220 (0.756) 0. ,477 Self-Report (version B) 7 0.408 (1.066) 0.924 (1.896) 0.373 (0.758) 0.270 (0.941) 0. .494 Self-Report (version B) 8 0.475 (0.617) 0.077 (0.276) 0.252 (0.508) 0.625 (0.617) 0. 357 Marginal Totals. . 0.303 0.287 0.178 0.180 * C - Controls RA - Restricting Anorexics BA - Bulimic Anorexics B - Bulimics Skin Conductance Amplitude Amplitude during Both Tasks The amplitude of skin conductance during task 2 ( i .e . , the instructions to eat) was compared with amplitude of responses during task 1. Table 9 summarizes the results of these analyses. First, the amplitude of the response to the f irst slide was used for comparison. A 4 x 2 repeated measures analysis showed that the mean amplitude for task 2 was significantly larger than the response to the f irst slide, F (1,36) = 75.07, p. < 0.001). This was observed in all groups. Next, the largest amplitude of a response to a slide was used for comparison. Again, amplitude for task 2 was significantly larger, F (1,36) = 58.23, p_ < 0.001). No differences among groups emerged. Finally, the amplitude to the instructions for task 1 ("get ready for the slides") was compared with the amplitude to task 2 ("start to eat now"). The results yielded F ratios that were not significant. The mean amplitude was similar for both tasks, £ (1,36) = 2.22, p. > 0.05, and for all groups, £ (3,36) = 1.12, p. > 0.05. It was also observed that the responses of subjects to the slides habituated very quickly, i .e. , for most slides no measurable skin conductance responses were present. Consequently, no further analyses regarding responses to slides were done. Amplitude during Self-Report Measures Mean amplitudes to the instructions to complete a self-report measure (largest response in the f irst thirty seconds of f i l l ing out a self-report measure) were analyzed in a 4 x 8 (group by time) repeated measures ANOVA design. Both the effects for group and time were significant; the group by time interaction was non-significant. For between groups, the F ratio was significant, £ (3,36) = 4.83, p. < 0.01. The lowest mean amplitude was observed in the restricting-anorexic group 94 Table 9 Skin Conductance Amplitudes (nmhos) for Task 1 and Task 2 Means (Standard Deviations in Parentheses) Group Task 1 - slide Task 2 f irst largest instruction instruction C* .27 .42 1.02 .99 (.52) (.51) (.54) (.44) RA* .04 .16 .54 1.01 (.10) (.25) (.52) (.71) BA* .07 .25 .87 1.04 (•09) (•28) (.56) (.61) B* .14 .26 1.19 1.14 (.24) (.39) (.58) (.63) C - Controls RA - Restricting-Anorexics BA - Bulimic-Anorexics B - Bulimics (0.35yiA,mhos). This mean amplitude was significantly lower than the mean amplitude in the bulimic group (0.86 ^cmhos) (p_ < 0.05) and in the control group (0.97yttmhos) (p_ < 0.01). The mean amplitude of the bulimic-anorexic group (0.53 y/.mhos) fell in-between the restricting-anorexic and bulimic group and was not significantly different from either one. The time main effect was significant, F (3,36) = 8.88, p. < 0.01. The mean amplitudes for the self-report measures given during task 1 were very similar to each other and to the mean amplitudes to the self-report measures prior to task 2 and the final one in task 2. The mean amplitude of the f irst self-report measure of task 2 (0.40 ^imhos) (at time 5) was significantly lower than the mean amplitudes of the self-report measures during task 1 (p_ < 0.05). The lowest mean amplitude was obtained for the self-report measure after task 2 (0.26 ^tmhos) (at time 7), which was significantly lower (p_ < 0.01) than the mean amplitudes of all self-report measures except the one at time 5. Figure 6 summarizes these results. The two low mean amplitudes were obtained following a period of activity. Before onset of task 2 (time 5), the experimenter had entered the room, distributed materials and talked to the subject. Visual inspection of the recordings during this time period showed considerable skin conductance activity with large responses. The other low mean amplitude was obtained at time 7 after task 2 where the number of fluctuations (see below) also indicated activity. Task 1, by comparison, was an activity characterized by l i t t le physical movement. These results can perhaps be seen best in the context of physical activity. Non-Specific Fluctuations The number of non-specific skin conductance fluctuations -another measure of arousal - was counted during various time periods. Skin conductance responses were judged to be non-specific fluctuations if they occurred spontaneously, i .e . , no stimulus was observed that could have 96 Figure 6 Skin Conductance Amplitude Legend • i Control* ZZ) Bufimka Buflmk-Anor«xic* CD R«itrict('ng-Anofexic« elicited the response. For example, skin conductance responses (greater than .1 jjimhos) following a change in breathing or some activity (such as turning a page) were not counted. During self-report measures, the number of fluctuations were counted in the interval thirty seconds after starting (in order to allow for habituation of the response to the instructions to complete the self-report) until the end of the self-report. During rest periods, the number of fluctuations were assessed during the same intervals measures for the tonic level were obtained. Finally, during task 2 all skin conductance fluctuations were counted if they occurred 30 seconds after onset of the task and were not due to a change in the breathing pattern. The above criteria were modified for this time period since i t was difficult to distinguish between responses that followed some activity (associated with eating) and those that were non-specific. In all cases, the number of fluctuations was scored on a one minute basis and this score entered the analyses. Rest Periods The scores (Table 10) for skin conductance fluctuations during the four rest periods (four scores per subject) were analyzed in a 4 x 4 (group by time) repeated measures ANOVA. The results showed significant group differences, F (3,36) = 3.25, p. < 0.05). Tukey comparisons of marginal means showed that the fluctuations in the restricting-anorexic group were significantly lower (p. < 0.05) than the fluctuations in the control group. No other group comparisons were significant. The analyses for time and the group by time Interaction were not significant. When the scores that entered the analysis were inspected, 1t was noted that skin conductance fluctuations were not present 1n every possible interval. In such cases a score of zero was entered into the analysis. Although scores of zero occurred in each group, they were particularly frequent in the restricting-anorexic group. For example, in 98 Table 10 Mean Number of Skin Conductance Fluctuations for Rest Periods in each Group (Standard Deviations in Parentheses) Rest Period Group 1 2 3 4 Marginal Total. c* 0.985 (0.897) 1.204 (0.929) 1.438 (1.639) 1.437 (0.971) 1.266 RA* 0.024 (0.076) 0.246 (0.423) 0.248 (0.393) 1.029 (2.464) 0.387 BA* 0.403 (0.455) 0.515 (0.730) 1.011 (1.066) 0.720 (0.752) 0.662 B* 0.926 (0.778) 1.083 (0.753) 1.145 (1.171) 1.276 (1.401) 1.108 Marginal Totals.0.584 0.762 0.961 1.116 RA** 0.24 n=l 0.82 (0.32) 0.62 (0.40) 2.57 (3.59) C - Controls RA - Restricting Anorexics BA - Bulimic Anorexics B - Bulimics ** RA - Restricting Anorexics with zero responses omitted the f irst rest period the numbers of subjects that showed no skin conductance fluctuations in each group were as follows: two in each of the control group and the bulimic group, four in the bulimic-anorexic group, and nine in the restricting-anorexic group. The number of intervals with an absence of fluctuations varied for the different time periods, but the largest number of subjects that showed fluctuations during a rest period was four in the restricting-anorexic group. In order to further understand the skin conductance fluctuations, the means and standard deviations of fluctuations in the restricting-anorexic group were calculated omitting all zero responses. As Table 10 shows these new means are more similar to the other groups. Self-Reports The number of skin conductance fluctuations that occurred during the self-report measures (eight scores per subject) were subjected to a 4 x 8 repeated measures ANOVA. The F-ratio for the group effect was significant, F (3,36) = 4.10, p_ < 0.05. Neither the main effect for time nor the group by time Interaction was significant. Comparisons of marginal group means showed that the restricting-anorexic group had significantly fewer skin conductance fluctuations than the bulimic group (p_ < 0.05). No other group comparisons were significant. Similarly to the number of skin conductance fluctuations during rest periods, subjects in the restricting-anorexic group frequently gave zero responses during self-reports. Again, the average number of fluctuations was calculated for those subjects in this group that did give a response. These scores are included in Table 11 that presents the means of skin conductance fluctuations during the self-report measures for all groups. Although no statistical analyses were performed, it appears that the mean number of fluctuations given by subjects in the restricting-anorexic group is similar to those given by subjects in the other groups 1f restricting-anorexic subjects with a response of zero are excluded. 100 Table 11 Mean Number of Skin Conductance Fluctuations During Self-Report Measures (Standard Deviations in Parentheses) Groups Self-Report 1 c* 2.206 (1.313) RA* 0.121 (0.383) BA* 1.682 (1.610) B* 1.766 (1.740) Marginal Totals. 1.443 RA** 1.21 n=l 2 2.224 (1.836) 0.282 (0.470) 0.939 (1.053) 2.619 (2.310) 1. .516 1.03 n=2 3 1.964 (1.310) 1.003 (2.127) 0.559 (0.529) 2.223 (2.289) 1. .437 2.01 (2.8) 4 2.342 (2.241) 0.337 (0.494) 0.735 (1.039) 2.887 (2.564) 1. .575 0.84 (0.4) 5 2.557 (2.654) 0.582 (0.809) 1.114 (1.146) 1.879 (2.011) 1. .533 1.46 (0.5) 6 1.962 (1.102) 0.518 (0.984) 1.497 (1.111) 2.421 (2.426) 1. .599 1.73 (1.1) 7 1.598 (1.416) 0.773 (1.369) 0.681 (0.977) 1.840 (1.860) 1. ,223 1.42 (1.8) 8 1.544 (1.488) 0.925 (2.167) 1.122 (1.596) 2.544 (2.079) 1. ,533 2.31 (3.1) Marginal Totals. 2.049 0.567 1.041 2.272 C* - Control Group RA* - Restricting-Anorexic Group BA* - Bulimic-Anorexic Group B* - Bulimic Group RA**- Restricting-Anorexic Group excluding zero responses S k i n Conductance F 1 u c t u a t i o n s d u r i n g Task 2 F i n a l l y , t he number o f s k i n conductance f l u c t u a t i o n s t h a t o c c u r r e d d u r i n g e a t i n g were a n a l y z e d i n a one-way between-groups ANOVA. As d e s c r i b e d e a r l i e r , a l l f l u c t u a t i o n s were i n c l u d e d i n the s c o r i n g s i n c e i t was too d i f f i c u l t t o a s s e s s whether a f l u c t u a t i o n was n o n - s p e c i f i c . The a n a l y s i s y i e l d e d a n o n - s i g n i f i c a n t F - r a t i o , i n d i c a t i n g t he absence o f s i g n i f i c a n t group d i f f e r e n c e s . The mean number o f f l u c t u a t i o n s p er minute was 1.16 f o r both a n o r e x i c groups, 1.64 f o r the c o n t r o l group, and 1.86 f o r t he b u l i m i c group. U n l i k e t he r e s t p e r i o d s and s e l f - r e p o r t measures, more s u b j e c t s (6 out o f 10) i n the r e s t r i c t i n g - a n o r e x i c group showed s k i n conductance f l u c t u a t i o n s . I f the z e r o r e s p o n s e s a r e e x c l u d e d , the mean r i s e s t o 1.93. A l l s u b j e c t s i n the c o n t r o l group and 9 i n each o f the b u l i m i c - a n o r e x i c and b u l i m i c group had s k i n conductance f l u c t u a t i o n s d u r i n g t a s k 2. Summary In summary, the number o f s k i n conductance f l u c t u a t i o n s g i v e n by a l l groups were s i m i l a r t h r o u g h o u t the study; I.e., t h e r e were no s i g n i f i c a n t time e f f e c t s . Both d u r i n g r e s t p e r i o d s and d u r i n g s e l f - r e p o r t measures, s u b j e c t s i n the r e s t r i c t i n g - a n o r e x i c group gave fewer s k i n conductance f l u c t u a t i o n s . T h i s r e s u l t appears t o be due t o a r e l a t i v e l y l a r g e number o f z e r o r e s p o n s e s i n t h i s group. I f the non-responders are e x c l u d e d from the a n a l y s e s , t he means become more s i m i l a r . Furthermore, d u r i n g t a s k 2, when more s u b j e c t s i n the r e s t r i c t i n g - a n o r e x i c group gave r e s p o n s e s no s i g n i f i c a n t group d i f f e r e n c e s were found. Recovery H a l f Time H a l f time r e c o v e r y r a t e , I.e., the time i t t a k e s f o r the c u r v e I n d i c a t i n g a s k i n conductance response t o d e c r e a s e t o h a l f o f the s i z e o f the r e s ponse, was measured f o r the s k i n conductance f l u c t u a t i o n s . I f more than one measurement c o u l d be o b t a i n e d , t he s c o r e s were averaged. On some 102 o c c a s i o n s i t was not p o s s i b l e t o measure h a l f time r e c o v e r y r a t e because e i t h e r the f l u c t u a t i o n o c c u r r e d a t the end o f an i n t e r v a l and the c u r v e then d e c l i n e d o u t s i d e o f t h e s c o r i n g i n t e r v a l o r because r e s p o n s e s were superimposed on each o t h e r , i . e . , a new response o c c u r r e d b e f o r e the p r e v i o u s one had d e c r e a s e d s u f f i c i e n t l y t o make s c o r i n g p o s s i b l e . For th o s e i n t e r v a l s where s k i n conductance f l u c t u a t i o n s were absent, r e c o v e r y r a t e s were a l s o absent. As d i s c u s s e d i n the p r e v i o u s s e c t i o n , the r e s t r i c t i n g - a n o r e x i c group was c h a r a c t e r i z e d by many i n t e r v a l s w i t h no s k i n conductance f l u c t u a t i o n s . T h e r e f o r e , s c o r e s f o r some i n t e r v a l s were based on a few ( 1 , 2, o r 3) s u b j e c t s . In any c a s e , group s i z e s f o r the d i f f e r e n t s c o r e s were unequal and s m a l l . I t was, t h e r e f o r e , not p o s s i b l e t o conduct formal s t a t i s t i c a l a n a l y s e s ; r a t h e r t he means and s t a n d a r d d e v i a t i o n s f o r d i f f e r e n t i n t e r v a l s w i l l be p r e s e n t e d . Rest P e r i o d s T a b l e 12 c o n t a i n s the mean h a l f time r e c o v e r y r a t e s f o r the s k i n conductance f l u c t u a t i o n s t h a t o c c u r r e d d u r i n g r e s t p e r i o d s . A l t h o u g h the average s c o r e s ranged from 2.58 t o 7 seconds ( t h e l a t t e r b e i n g based on one s u b j e c t ) , most s c o r e s were between 3 and 5 seconds. No d i f f e r e n c e s r e g a r d i n g group o r time seem a p p a r e n t . S e l f - R e p o r t s In T a b l e 13 the mean h a l f time r e c o v e r y r a t e s f o r the f l u c t u a t i o n s t h a t o c c u r r e d d u r i n g t he s e l f - r e p o r t measures are p r e s e n t e d . S u b j e c t s i n the b u l i m i c - a n o r e x i c group had l o n g e r r e c o v e r y r a t e s than s u b j e c t s i n the o t h e r groups, but the d i f f e r e n c e s were not marked. E a t i n g Task F i n a l l y , t h e r e c o v e r y h a l f times f o r s k i n conductance r e s p o n s e s between groups d u r i n g t a s k 2 were compared. The l o n g e s t r e c o v e r y r a t e o c c u r r e d i n the r e s t r i c t i n g - a n o r e x i c group (6.25 seconds, sd. = 6.37) f o l l o w e d by 4.58 seconds ( s d . = 2.12) 1n the b u l i m i c - a n o r e x i c group. The mean s c o r e s i n the r e m a i n i n g groups were s i m i l a r : 3.33 seconds ( s d . = 103 Table 12 Mean Half Time Recovery Rates for Skin Conductance Fluctuations during Rest Periods (Standard Deviations in Parentheses) Group Period C* RA* BA* B* 1 4.49 7.0 5.28 2.75 (2.08) n=l (1.37) (1.57) 2 3.09 3.3 3.6 3.1 (2.65) (1.84) (1.56) (1.62) 3 3.29 5.4 3.12 3.7 (1.00) (0.85) (1.13) (2.54) 4 3.00 4.04 4.96 2.58 (1.19) (3.63) (2.88) (0.67) * C - Control Group RA - Restricting-Anorexic Group BA - Bulimic-Anorexic Group B - Bulimic Group Table 13 Mean Recovery Half Time for Skin Conductance Fluctuations during Self-Reports (Standard Deviations in Parentheses) Group C* RA* BA* B* Self-Report 1 3.06 2.4 3.32 2.13 (1.52) n=l (1.83) (0.61) 2 3.16 0 4.6 2.61 (1.48) (2.02) (1.09) 3 2.86 2.45 6.4 2.33 (1.24) (1.62) (2.35) (0.87) 4 3.13 3.6 4.98 2.29 (1.09) n=l (3.50) (0.73) 5 2.27 5.4 2.85 2.26 (0.65) (2.86) (0.69) (0.92) 6 3.08 2.7 5.89 2.16 (0.86) n=l (5.96) (0.69) 7 2.26 3.74 4.2 2.2 (0.75) (2.82) (2.12) (0.63) 8 2.53 4.45 4.54 2.1 (1.09) (4.17) (1.91) (0.93) * C - Control Group RA - Restricting-Anorexic Group BA - Bulimic-Anorexic Group B - Bulimic Group 3.37) i n the c o n t r o l group, and 3.08 seconds ( s d . = 0.87) i n the b u l i m i c group. The b e h a v i o u r a l measure o f major i n t e r e s t was the q u a n t i t y e a t e n . P r e f e r e n c e r a t i n g s o f the s l i d e s and the f o o d were a l s o a n a l y z e d i n o r d e r t o a s s e s s any p o s s i b l e d i f f e r e n c e s . S i n c e the t h r e e measures d i f f e r e d i n importance w i t h r e s p e c t t o the purpose o f the study, s e p a r a t e u n i v a r i a t e a n a l y s e s were performed. As t h e subsequent p a r a g r a p h s d e s c r i b i n g the a n a l y s e s show a l l o b s e r v e d d i f f e r e n c e s were s i g n i f i c a n t a t 0.05/3 o r 0.017. Q u a n t i t y Eaten The q u a n t i t y o f the f o o d eaten was measured i n terms o f the amount o f c a l o r i e s consumed. S u b j e c t s c o u l d e a t 275 c a l o r i e s a t the most. I t was h y p o t h e s i z e d t h a t a n o r e x i c s u b j e c t s would e a t s i g n i f i c a n t l y l e s s than c o n t r o l s u b j e c t s . R e s u l t s o f a one-way ANOVA i n d i c a t e d s i g n i f i c a n t group d i f f e r e n c e s , F (3,36) = 5.13, p. < 0.01). The f o l l o w i n g t a b l e shows the means and s t a n d a r d d e v i a t i o n s o f c a l o r i c consumption i n each group. B e h a v i o u r a l Measures T a b l e 14 Mean Number o f C a l o r i e s Eaten ( S t a n d a r d D e v i a t i o n s i n P a r e n t h e s e s ) C a l o r i e s Consumed C o n t r o l s 254.5 (55.68) R e s t r i c t i n g - A n o r e x i c s 115.8 (109.4) B u i i m i c - A n o r e x i c s 181.0 (108.8) B u i i m i c s 247.4 (68.59) S u b j e c t s i n the r e s t r i c t i n g - a n o r e x i c group consumed s i g n i f i c a n t l y 106 fewer c a l o r i e s than s u b j e c t s i n the b u l i m i c group (p. < 0.017) and i n the c o n t r o l group (p. < 0.01). A l t h o u g h b u l i m i c - a n o r e x i c s a l s o consumed l e s s , the d i f f e r e n c e s were not s i g n i f i c a n t . E i g h t y p e r c e n t o f the s u b j e c t s i n the c o n t r o l group and 70% i n the b u l i m i c group a t e a l l the f o o d , whereas o n l y h a l f the s u b j e c t s i n the b u l i m i c - a n o r e x i c group and 30% i n the r e s t r i c t i n g - a n o r e x i c group d i d so. F u r t h e r d i f f e r e n c e s a r e r e v e a l e d i f one examines how many s u b j e c t s a t e l i t t l e . S i x s u b j e c t s i n the r e s t r i c t i n g - a n o r e x i c group consumed l e s s than 100 c a l o r i e s ( t h e l e a s t amount eaten was 5 c a l o r i e s ) , t h r e e s u b j e c t s d i d so i n the b u l i m i c -a n o r e x i c group ( t h e l o w e s t s c o r e was 20), and o n l y one i n each o f the b u l i m i c group (55 c a l o r i e s ) and the c o n t r o l group (98 c a l o r i e s ) . P r e f e r e n c e R a t i n g o f Task 2 S u b j e c t s had a l s o r a t e d t h e i r p r e f e r e n c e f o r the f o o d eaten on a 9 p o i n t s c a l e (where a s c o r e o f 1 i n d i c a t e d d i s l i k e and 9 p r e f e r e n c e ) . A one-way ANOVA f o r f o u r groups y i e l d e d a s i g n i f i c a n t F - r a t i o , F (3,36) = 9.28, p. < 0.01. Comparisons o f means showed t h a t the mean p r e f e r e n c e r a t i n g (3-3, sd. 2.0) o f the f o o d eaten i n the r e s t r i c t i n g - a n o r e x i c group was s i g n i f i c a n t l y lower (p_ < 0.01) than the mean r a t i n g s i n the b u l i m i c (6.7, sd. 1.7), the b u l i m i c - a n o r e x i c (6.9, sd. 2.5), and the c o n t r o l (7.5, sd. 1.6) groups. None o f the o t h e r groups d i f f e r e d from each o t h e r . S u b j e c t s i n the r e s t r i c t i n g - a n o r e x i c group a t e v e r y l i t t l e and d i s l i k e d the f o o d t h e y a t e . A l l o t h e r s u b j e c t s e x p r e s s e d a p o s i t i v e p r e f e r e n c e f o r the food. P r e f e r e n c e R a t i n g o f Task 1 S u b j e c t s had a l s o r a t e d t h e i r p r e f e r e n c e f o r the s l i d e s they saw. No s p e c i f i c e x p e c t a t i o n s were f o r m u l a t e d r e g a r d i n g such r a t i n g s as p r e f e r e n c e f o r a r t i s i n d i v i d u a l . These r a t i n g s were a l s o a n a l y z e d i n a one-way ANOVA w i t h f o u r groups. The a n a l y s e s y i e l d e d a s i g n i f i c a n t F r a t i o , F (3,36) = 5.6, p. < 0.01. S u b j e c t s i n the b u l i m i c - a n o r e x i c group gave s i g n i f i c a n t l y (p_ < 0.01) lower mean r a t i n g s (4.0, sd. 1.1) than 107 s u b j e c t s i n t h e c o n t r o l group (5.91, sd. 1.1). The r a t i n g s by the r e s t r i c t i n g - a n o r e x i c group (4.87, sd. 0.9) and the b u l i m i c group (4.81, sd. 1.1) f e l l in-between the o t h e r groups and d i d not d i f f e r s i g n i f i c a n t l y from e i t h e r . I f one examines the s i z e o f the means, i t appears t h a t s u b j e c t s i n the e a t i n g d i s o r d e r groups somewhat d i s l i k e d the p i c t u r e s ( a s c o r e o f 5 i n d i c a t e s a n e u t r a l r a t i n g ) , whereas c o n t r o l s u b j e c t s seemed t o l i k e t he s l i d e s , p o s s i b l y because t h e y were i n no way t h r e a t e n e d by the n a t u r e o f t h i s experiment. 1 0 8 Chapter 4 DISCUSSION The main purpose of the study was to measure levels of anxiety before, during, and after eating in three groups of eating disorder subjects and normal-weight controls. It was expected that controls would show l i t t le or no anxiety during eating, whereas eating disorder subjects would show high levels of anxiety (as assessed by self-report, psychophysiological, and behavioural measures). Confirming expectations, controls were indeed non-anxious, although psychophysiological arousal (particularly heart rate) during eating was high. By contrast, eating disorder subjects reported high levels of anxiety prior to and throughout eating. Self-report of anxiety was already high during a neutral task indicating that anorexic and bulimic subjects anticipated eating and felt threatened from the beginning of the study. During eating, anxiety increased even more. In terms of skin conductance measures, many anorexics were found to be hyporesponsive, although responsiveness increased during eating, again indicating increased arousal. The behavioural measures differentiated most among groups. Restricting-anorexics ate the least, whereas bulimics ate similar amounts as controls. Bulimic-anorexics were an intermediate group. If one considers the scores of normal-weight controls as a 'standard', then generally the scores of restricting-anorexics were most different from it. Weight status appeared to be an important variable. Overall, the results confirm that in anorexic and bulimic females anxiety during eating is prominent and that it is useful to measure this anxiety with the three-response system of the tri-partite model of anxiety. The following sections discuss the results of the study 1n more detail. The discussion will begin with the findings from the various dependent measures (self-report, psychophysiological, and behavioural) as 109 they relate to the expectations of the study. The covariation of these measures in the different groups is of interest. Differences between groups based on their responses will be elaborated along with diagnostic implications. Finally, conclusions drawn from the study will be presented, and an anxiety model of anorexia nervosa and bulimia will be described. Self-Report Measures Responses to the self-report inventory that measured affective states of pleasure, arousal, and anxiety generally confirmed expectations. Controls reported a higher level of pleasure than the eating disorder groups. Their level of arousal was moderate and their anxiety was low. This pattern was seen throughout the study. By contrast, the eating disorder groups reported a lower level of pleasure and higher levels of arousal and anxiety. In these groups anxiety increased further during the eating task. Levels of pleasure declined somewhat throughout the study, perhaps due to fatigue, boredom, or the repeated measurements. Levels of arousal were highest after each task which had required physical activity. Overall, arousal was higher during the experimental task which also required more activity. J . Russell (1980) described a circumplex model of emotion where mood states could be expressed on two independent bipolar dimensions. If one plots the mean levels of pleasure and arousal for each group on these two dimensions, the control group is positioned quite differently from the eating disorder groups (if the actual scores are plotted on the axes). As Figure 7 shows, the control group falls in the lower right quadrant indicating a pleasant and relaxed affective state. By contrast, the two anorexic groups appear in the upper left quadrant, while the bulimic group scores are close to the intersection of the axes. Although the anorexic groups are quite close to the axes, the restricting-anorexic group average is located closer to 110 F i g u r e 7 P o s i t i o n , o f A n o r e x i c B u l l a e , and C o n t r o l Group on Axes o f Arousal and Pl e a s u r e P Arousal high low 1 Pleasure 9 h i g h 11 low Legend B C - C o n t r o l s • RA - R e s t r i c t i n g - A n o r e x i c A BA - B u l i m i c - A n o r e x i c s O 8 - B u l i m i c s 111 adjectives describing a distressed affective state. The bulimic-anorexic and bulimic groups are more diff icult to describe because they are located almost at the intersection of the axes. The scale assessing anxiety discriminated best among the groups. Already at the time of the f irst assessment, the controls and two of the eating disorder groups differed on levels of pleasure and anxiety. This is consistent with the observation that eating disorder patients generally report a higher level of anxiety than control subjects (as, for example, MMPI data show, see Mizes, 1985). In the present case this was accompanied by a lower level of reported pleasure. The experimental task appeared to increase levels of anxiety in the eating disorder groups. The highest level of anxiety was reported after subjects had been given the food and were about to start eating it . The univariate analysis (although not analyzed fully since the multivariate analysis yielded non-significant results for the interaction of groups by anxiety) and the MANOVA comparing mean ratings of anxiety during the two tasks did indicate an increase in anxiety during the eating task for the eating disorder groups, whereas level of anxiety declined in the control group (see Figure 3, p. 75). Comments by subjects indicate that they were not anticipating the experimental task during the neutral (slide) task; furthermore, anxiety levels were lowest after the neutral task, suggesting that subjects were able to relax. Therefore it appears that eating disorder subjects are generally characterized by higher levels of anxiety; in addition, levels of anxiety increase prior to eating and remain high during eating. If the anxiety is caused by the anticipation of eating, subjects may not be aware of 1t. Results from condition 2 (see Appendix 3) suggest that the init ial anxiety was indeed caused by the anticipation of eating. Anorexics and bulimics reported higher levels of pleasure and lower levels of anxiety when they knew they would not have to eat the food. 112 Robinson et al . (1983) reported that bulimic patients had reported significantly more anxiety than restricting-anorexics and controls prior to and after a meal. After the meal, restricting-anorexics were also significantly more anxious than controls. The present study confirms part of those results. The two bulimic groups reported more anxiety than the controls did. However, restricting-anorexics were also significantly more anxious than controls and were not different from the bulimic groups. In fact, it was the restricting-anorexics who differed most from the controls in the present study. In contrast, the Robinson et a l . (1983) restricting-anorexics reported slightly (but not significantly) more anxiety and the bulimics reported less anxiety after eating. This pattern was reversed in the present study: restricting-anorexics reported slightly less anxiety and the bulimic groups somewhat more anxiety after eating (see Figure 3, p. 75). It is possible that restricting-anorexics were relieved when they were able to stop eating whereas the bulimic groups, who also ate more, were more ambivalent towards eating. These observations are quite speculative as the actual changes were small, but of interest. Leitenberg et al . (1984) had found that anxiety increased for bulimics while eating. Their results are however diff icult to compare since the present study only obtained a pre- and post-anxiety rating, but no ratings during eating itself. Furthermore, in the Leitenberg et al. study, bulimics were encouraged to eat until the urge to vomit was strong, whereas in the present study a relatively small quantity of food was consumed and no specific demands regarding the quantity to be eaten were made. Responses on the self-report inventory 2 which assessed attitudes to food and eating lend support to views that emphasize cognitive distortions in eating disorders. Anorexics were more preoccupied with food; all three eating disorder groups showed a greater fear of eating 113 (which is consistent with the high anxiety ratings on the other self-report measure); anorexics also believed that eating a small meal would affect their weight and make them fat. Prior to eating, the questions about urges to overeat and probabilities of doing so did not yield any differences, although they did so after eating when the bulimic groups reported a stronger urge to overeat and reported a higher probability of doing so. As expected, restricting-anorexics did not report any urges to overeat. Group differences emerged after eating: food consumption appears to increase urges to overeat and probability of binging. This finding is consistent with the starvation studies (Keys et a l . , 1950), where during refeeding volunteers reported increased hunger. Some bulimics also report that eating any size of meal will result in binge-eating. Perhaps for bulimics, who may be under their ideal weight set-point, eating leads to increased biological pressure to eat more and recover the lost weight. Alternatively or in addition, i f a small meal functions like a preload in restraint studies, then bulimics can be expected to counterregulate, i .e . , report an increased urge to eat. Of course, a meal of 275 calories would be considered small only in the context of a normal ( i .e . , 2100 calories/day) diet. For dieters on a much more restricted diet, say 1000 calories, 275 calories represent a more substantial portion of their daily intake. In any case, it appears that in bulimics eating increases the likelihood of binge-eating; the termination of eating becomes then an important focus. One further methodological point is in order. Failure to find significant differences on some of the questions might also have been in part due to the particular wording of the questions. The use of the word 'overeat' is ambiguous; it Implies eating more than some unspecified standard and it is also commonly used. The word 'binge' has a negative connotation Implying abnormal eating. Therefore, 1f the word 'binge' had been used instead, the controls might have responded differently to the questions. Since psychometric properties of the 114 instrument had not been established, caution must be used when interpreting the results. S t i l l , the findings do confirm the importance of cognitive factors in eating disorders and the value of further investigation of such factors. Psychophysiological Measures Two psychophysiological measures - heart rate and skin conductance -were obtained and were useful in distinguishing between groups and tasks. They will be discussed in turn. Heart Rate Before the heart rate findings are discussed, heart rate variability in the different groups will be commented on. Initial data analyses showed that heart rate variances were heterogeneous in the init ial rest periods. The largest variability was found in the restricting-anorexic group where some subjects were characterized by bradycardia, a frequent accompaniment of semi-starvation, but others by tachycardia. The cause of such high heart rates is not clear. It 1s interesting to note that Keys et al . (1950), when measuring pulse rate, had observed 'relative tachycardia1 in some subjects during refeeding with unlimited food access. It would be interesting to investigate further how heart rate is affected by increased caloric intake. In the present study heart rate variability diminished as the study progressed because lower scores tended to increase. This increase appeared to be smaller for the higher scores suggesting a ceiling effect. Such a limitation might, of course, reduce possible group effects if groups consist of differing numbers of subjects who show limited heart rate Increases. This appears to be the case in this study. In addition, heart rate responsltivity for different subjects might vary due to other, unknown factors such as nutritional status. The results must therefore be Interpreted cautiously. Overall, heart rate levels during the study yielded two findings: a 115 lack of group differences and a large increase during the eating task. Throughout both tasks groups responded in a similar manner. It had been expected that all groups would show l i t t le change in heart rate or that they would show heart rate deceleration after presentation of neutral slides. The results did confirm this expectation and are consistent with a heart rate orienting response. The visual stimuli had indeed been perceived as neutral. During eating, heart rate increased significantly in all groups. With respect to the three time samples taken (the in i t ia l , middle, and last minutes), heart rate increase was significant from the f irst minute when subjects anticipated, prepared for, and eventually started eating to the middle minute of the task when subjects were in fact eating. During the second half of eating, average heart rate did not increase in the control and restricting-anorexic groups and increased only slightly in the bulimic groups. Although the heart rate increase from time 1 to time 2 during eating was significant, the actual increase was comparatively small, about 2 - 3 beats per minute. This increase could be due to increased motor activity (getting the food, chewing, swallowing, etc.). Subjects, however, differed 1n the amount of activity performed; for example, some may have moved more than others when performing a comparable activity. This was not directly measured. Subjects also differed on the amount eaten and this presumably also affects motor activity. It 1s interesting to note that the control and the bulimic groups who ate the most did not show any further heart rate increases, whereas the anorexic groups who ate less did so (see Appendix 10, p. 159). Heart rate was also significantly higher during eating than during slide presentation. This difference was already significant at the f irst minute of eating. As Appendix 10 (p. 159) shows, both the control and the bulimic groups had very similar heart rates during the slide presentation and at time 1 of eating, whereas the anorexic groups had a comparatively 116 higher heart rate at time 1 of eating. This apparently higher heart rate was, however, not significant. The two tasks were not equivalent in terms of activity required. Viewing and rating slides involves less movement than eating food. Nonetheless, it was st i l l somewhat surprising that at the beginning of task 2 heart rate was so much higher than during task 1. If the average heart rate increase from rest period preceding the task to the task itself is calculated, then it appears that heart rate during task 1 is about 2 bpm higher than during rest period 1 (controls 2.6, restricting-anorexics 1.85, bulimic-anorexics 1.9, and bulimics 1.28) reflecting increased motor activity. Heart rate increase from rest period three to the beginning of task 2 is however considerably larger, almost 8 bpm (controls 6.9, restricting-anorexics 8.9, bulimic-anorexics 9.2, and bulimics 6.5). Between the respective rest period and the task, self-reports had been completed. Change scores during self-report measures were positive, but did not differ for self-reports 2 and 6. As Table 7 (see p. 86) shows, the actual change scores on self-reports cannot account for this large heart rate increase. As already mentioned earlier, the initial minute of eating was characterized by increased motor activity (to initiate eating) and also anticipation of this task. Subjects had been aware of the purpose of the experiment, i .e . , to study what "happens when people do different things including eating" and were possibly responding to some unspecified experimental demand. Heart rate was not higher during the rest period prior to eating. Any anticipation of eating subjects might have had did not result in higher physiological arousal before the task. The higher heart rate during the last period (after eating) reflects the increased heart rate during eating. Heart rate had apparently not yet returned to the level of the rest period prior to eating. As Figure 4 (p. 83) shows, heart rate was higher during the tasks 117 associated with increased motor activity. Such cardiac-somatic coupling has been described by Obrist (1981) and appears to occur when heart rate is under vagal control (as, for example, during exercise). Both heart rate and motor activity are integrated by the central nervous system. The large increase during task 2 (particularly the f irst minute) cannot, however, easily be accounted for by increased physical activity. Obrist (1981) has also suggested that under some conditions (as, for example, avoidance of aversive stimuli) cardiac-somatic uncoupling can occur and heart rate increases are then due to sympathetic excitation. Aversive stimuli have often elicited marked heart rate increases (see Obrist, 1981, Siddle & Turpin, 1980), and eating disorder patients often report conflicting emotions while eating. The results are, however, more diff icult to interpret for the control subjects. Measures that might have helped to interpret the results such as vasomotor activity could not be obtained because of recording diff iculties. Since heart rate changes are mediated by both the parasympathetic and sympathetic nervous systems, it is possible that the actual heart rate increases observed in the different groups might be due to differential arousal of the two nervous systems. It therefore would be interesting to estimate sympathetic activity during eating, particularly during the f irst minute. Various methods of estimating sympathetic activity have been proposed, mainly the analysis of T-wave amplitude (see Furedy & Heslegrave, 1983) or spectral analysis of respiratory sinus arrythmias and heart period (Porges, McCabe & Yongue, 1982). In order to simplify scoring the EKG was transformed to a tachographic output. Consequently, T-wave amplitudes were not available for analyses. Similary, spectral analysis requires recording of respiration data in a standard manner. Again, the data from the present study could not be analyzed in this manner. One can only speculate, i f indeed heart rate increase during eating was due perhaps to sympathetic activity in eating disorder subjects and reduced parasympathetic activity 118 in control subjects. It is therefore suggested that a future study measure these variables in the different groups. Skin Conductance Unlike heart rate, variances of tonic levels of skin conductance were homogeneous in the various groups. Generally, there were no differences among groups. Skin conductance levels increased throughout the study reflecting increased arousal. Overall, levels were higher during the second part of the study, but did not show any significant increases before and after each task. During eating, the highest level was obtained at the f irst minute of eating. At this time, a significant heart rate increase was also observed. Whereas heart rate continued to increase during eating, skin conductance levels declined. During presentation of the slides, tonic skin conductance levels of restricting-anorexics were significantly lower than those of bulimics with the scores of the other groups falling in-between. This finding differs from the Calloway et al . (1983) study where no differences of log skin conductance level were found when anorexic, bulimic, and control subjects were compared. Skin conductance levels during eating were higher than during the slide task reflecting increased arousal during eating. As the pattern of tonic level suggests (see Figure 5, p. 89), skin conductance was highest at the f irst minute of eating. The increase in skin conductance levels during the eating task was particularly pronounced in the anorexic groups, whereas bulimics responded more like controls. The size of the actual change scores during self-report 6 (see Table 8, p. 93) suggests that some increase in skin conductance level already occurred prior to the beginning of eating (although a similar size average change score is also observed at self-report 2, prior to the slide task. The measures are however not easily compared since the scores for the slides are averages of the ten slides). 119 Such observations lead to further speculations. If heart rate pattern is compared with the skin conductance pattern (see Figure 4, p. 83 and Figure 5, p. 89), it appears that heart rate increases at the beginning of eating in the control and bulimic groups are accompanied by relatively small skin conductance level increases, whereas in the anorexic groups somewhat higher heart rate increases are associated with larger increases in skin conductance levels. Hare (1978) suggested that during aversive stimulation small skin conductance changes occuring during heart rate increases indicate that l i t t le fear is present. If a high heart rate indicates 'environmental rejection' (a defensive response), then skin conductance increase can indicate how 'successful' this rejection is (in reducing fear arousal). Assuming that eating is aversive to eating disorder subjects bulimics indicate the presence of less fear than the anorexic subjects. Bulimics in this study ate more than the anorexics. Presumably, for controls eating is not aversive. Although it is interesting to compare heart rate and skin conductance increases in the groups and the figures show some apparent differences, it is emphasized that such observations are highly speculative as the statistical analyses did not find any differences among the groups. However, future research could study more specifically the relationship between heart rate and skin conductance responses in these groups. Before the results of the other components of the skin conductance response are discussed, the caution concerning a probable high type 1 error rate is repeated. S t i l l , 1t is useful to compare the findings of this study with previous research. All conclusions are, necessarily, tentative. Amplitude of the skin conductance responses was also measured. Analyses failed to yield meaningful task differences. If one takes increased physical activity into account, amplitudes to the self-report measures showed few differences when compared across the tasks. Responses 120 to instructions to start a task were also similar and perhaps signaled readiness and anticipation. The only differences between tasks were found when responses to the slides were compared with the instruction to start eating; admittedly, the stimuli used to provide these data were not very comparable. Generally, mean amplitudes 1n the restricting-anorexic group were significantly lower than amplitudes in the bulimic and control groups. Mean amplitudes of the bulimic-anorexics were somewhat higher than those of the restricting-anorexics. This finding could suggest that restricting-anorexics are hyporesponsive; however, in a comparison of task instructions, no group differences emerged (see Table 9, p. 95). When asked to get ready for these tasks (viewing and rating the slides and eating) skin conductance amplitudes of restricting-anorexics increased and were, thus, no longer different from those of the other groups. It appears, therefore, that under some conditions restricting-anorexics show responsiveness that is comparable to the other groups. The number of spontaneous fluctuations, I.e., skin conductance fluctuations that were not due to clearly observable stimuli, was also analyzed. Two main findings emerged. Except for the eating task, fewer spontaneous fluctuations were observed in the restricting-anorexic group. Subjects who showed no spontaneous fluctuations were found in each group and time period; however, the largest number of zero responses was found in the restricting-anorexic group, resulting in a lower overall mean. Again, restricting-anorexics were more likely to be hyporesponsive or not to respond at a l l . Those subjects who did show spontaneous fluctuations were similar to the subjects in the other groups. These results are contrary to the findings in the Calloway et al . (1983) study where restricting-anorexics had the highest number of spontaneous fluctuations and the number of fluctuations was significantly lower 1n the bulimic and the bulimic-anorexic groups. The authors had concluded that the different 121 skin conductance pattern - showing hyporesponsiveness - in the bulimic groups was associated with binging and vomiting, and not with weight, as the weight of both anorexic groups was similar. In their study restricting-anorexics and controls also showed similar patterns. In the present study the number of spontaneous fluctuations did appear to be associated with weight; the group with the lowest weight had the highest number of subjects who showed no responses. Bulimic-anorexics, who had the second lowest weight, had more non-responders than the bulimics or controls did (but fewer than the restricting-anorexics). The bulimic-anorexics did not differ from any other group significantly, but appeared to be intermediate among the eating disorder groups. Interestingly, and as with skin conductance amplitudes, more restricting-anorexics responded during eating and group differences disappeared. This increase 1n the number of fluctuations probably reflects the increased arousal during this task and was not found during other time periods such as self-report measures and rest periods. The Calloway et al. study found no group differences with respect to recovery half-time (the time it takes for a skin conductance response to decrease to half of the size of the response). In the present study group comparisons could not be made as some groups, particularly the restricting-anorexic group, became very small (e.g., n < 5) when non-responders were excluded. Inspection of the means suggests similar recovery times in each group. However, during the eating task, the anorexic groups (particularly the restricting-anorexics) showed a much longer recovery half-time, suggesting Increased arousal. In summary, psychophysiological measures failed to distinguish between groups when group data were analyzed. All groups appeared to respond in a similar manner showing higher arousal during eating. As previous studies have found (see Calloway et a l . , 1983; Leitenberg et a l . , 1984) some eating disorder subjects were hyporesponsive. In the 122 present study restricting-anorexics in particular and also to some extent bulimic-anorexics were more likely to be non-responders. Physiological responsiveness appeared to be associated with weight status. Eating disorder subjects who were physiological responders gave responses of similar magnitude (except for skin conductance amplitudes by restricting-anorexics) as the control subjects. Despite the absence of group differences, restricting-anorexics and also bulimic-anorexics appeared to show increased arousal during eating: otherwise earlier non-responders responded during eating with an increased number of fluctuations and several interactions of anxiety measure by time approached significance. The increased heart rate during eating (found in all groups) is an interesting finding. Perhaps the increased physiological arousal during eating which is associated with verbal reports of anxiety increases perceptions of anxiety by anorexic subjects. Controls, by contrast, might interpret their increased arousal as due to increased activity or, even, pleasurable feelings (e.g., I enjoy this meal). This is of course speculation, since this study did not test directly the subjects' perception of arousal. It would be of Interest to further investigate heart rate during eating and subjects' perception of it . Skin conductance measures are more diff icult to obtain and, in this study, hyporesponsitivity was a problem. Behavioural Measures Because most bulimics (and controls) ate all the food, the scores for quantity eaten were not normally distributed and consequently could not be used for correlational analyses. The results are therefore discussed in a descriptive manner. As expected, anorexic subjects ate l i t t le during the study although they knew the caloric value of the food prior to the start of the study and were thus presumably able to adjust their earlier and anticipated 123 daily intake taking this meal into account. For some anorexics, the relatively small snack might have represented a substantial portion of their food intake for that particular day. Some anorexics reported that they had planned to eat all of the food but found themselves unable to do so. It is noteworthy that some restricting-anorexics ate all the food; anorexics appear to increase consumption under certain conditions. Bulimics ate a similar amount as controls; perhaps the quantity was sufficiently small not to affect their daily intake significantly and thus increase anxiety. In addition, vomiting was not controlled for; it was possible for subjects to vomit after the study, but no provision had been made to ascertain if indeed any subjects did. It seems, however, unlikely that subjects planned on vomiting as they did not finish the study until some time after eating (and thus had no timely opportunity to do so). Furthermore, self-reported anxiety after eating did not increase in the bulimic groups. These results differ from those obtained by Rosen et al . (1985) where bulimics, when prevented from vomiting, became increasingly anxious while eating and also ate less than a control group. The data suggest that the bulimics in the Rosen et al . study ate less than the bulimics in this study. In summary, food consumption appears to be correlated with weight status, i .e. , subjects who weigh less also eat less. Preference ratings of the food had also been obtained. Restricting-anorexics appeared to dislike the food they ate whereas all other groups seemed to like it . Surprising is the relatively high food rating by the bulimic-anorexics who, in fact, had not eaten much. Discordance of Measures The three measures - self-report, psychophysiological, and behavioural - showed different covariances in each group. For the controls, the self-report and behavioural measures suggested low anxiety although psychophysiological arousal during eating was high. During the 124 neutral slide task, psychophysiological arousal was much lower and, thus, there was concordance among the three measures of anxiety, suggesting a non-anxious state. The design of the present study cannot specify the cause of the high physiological arousal during eating, but this arousal was clearly not interpreted as being a state of anxiety. Restricting-anorexics showed concordance of the three measures during eating. Physiological arousal was high, self-reports indicated a high level of anxiety, and low food consumption suggested behavioural avoidance. During the presentation of the neutral slides, discordance was observed: self-reports were already indicating a high level of anxiety. It is , however, probable that overall levels of anxiety are high in restricting-anorexics. Bulimic-anorexics also reported high levels of anxiety and showed high levels of physiological arousal, but were less avoidant since consumption of food was higher than in the restricting-anorexic group. This group, therefore, showed less concordance among anxiety measures. Finally, the bulimics reported similar levels of anxiety and physiological arousal as the anorexic groups, but showed no behavioural avoidance. Measures of anxiety in this group can be described as discordant. Degrees of concordance have been found to be influenced by several variables such as emotional arousal and levels of demand (Rachman, & Hodgson, 1974). The present study did not assess effects of different levels of demand (e.g., subjects' perception of demand regarding food consumption). Emotional arousal was high in the eating disorder groups, but only the anorexic groups showed good concordance among anxiety indices. Differences Between Eating Disorder Groups Self-reports describing levels of pleasure, arousal, and anxiety did not differentiate among the three eating disorder groups. It was therefore not possible to correlate anxiety scale scores with other 125 measures. There were some group differences with respect to the physiological measures: for example, restricting-anorexics showed lower skin conductance amplitudes and fewer spontaneous fluctuations. Overall, all groups showed high physiological arousal during eating. Differences were most apparent during the behavioural avoidance task, i .e. , the quantity of food eaten. Here bulimics were very similar to controls. Generally, restricting-anorexics were most different from controls (as the various relevant figures illustrate), whereas bulimics were most similar to controls in some aspects such as food consumption. Responses by bulimic-anorexics were often intermediate, at times being similar to the responses of restricting-anorexics (e.g., a relatively higher level of tonic skin conductance during eating). These findings provide only partial support for the results of a study comparing bulimics of different weight status and reporting very few differences among them (Garner et a l . , 1985). In the present study if the eating disorder groups differed on a variable, then the finding was consistent with the weight status of the group (as, for example, with food consumption). These results emphasize the importance of weight restoration in the treatment of anorexia nervosa. In addition, bulimic-anorexics appear to be sufficiently different from the restricting-anorexics and the bulimics to assign them to a separate subgroup. Conclusion As was pointed out in the introduction, many variables (such as body image disturbances and sociocultural factors) are considered relevant to the etiology and maintenance of anorexia nervosa and bulimia. The results of the present study suggest that anxiety also plays an important role in eating disorders. Both anorexics and bulimics generally report a high level of anxiety. Unlike phobics who can avoid a specific situation (such as a dog or high bridge), anorexics do not experience marked reductions 126 of anxiety (as assessed by self-report) when not eating. Because the body strives for food at all times, anorexics must struggle continuously not to give in to their urges to eat. Bulimics who may be below weight set-point probably also experience biological pressures to eat. Since this study did not assess probable set-point of the subjects, it is not known if all eating disorder subjects were indeed below set-point. It is recommended that such a measure be Included in future research. In addition, bulimics might be anxious prior to eating since they anticipate difficulty in stopping. In any case, anxiety is all pervasive and must be addressed in therapy. During eating, when conflicts about body shape and weight are prominent, anxiety is also high. Not only did subjects report themselves to be more anxious, but physiological arousal increased as well. Anorexics, in addition, showed avoidance behaviours, and thus can be described as showing a fear of eating in a manner similar to phobics. Bulimics did not show the expected avoidance behaviour. They generally appear to eat more than anorexics do, as is also evidenced by their higher weight. They therefore experience more exposure to the feared situation. Their main difficulty lies with terminating rather than starting to eat. Because the food available during the study was limited, they did not have to worry about losing control and overeating. Such external controls appeared to have been effective in preventing a report of increased anxiety after eating. S t i l l , as in the anorexic groups, self-reports of anxiety were high as was physiological arousal. Although this study did not fully support the anxiety reduction model of bulimia (perhaps because of experimental limitations), 1t does appear that anxiety is an important variable in bulimia. The study confirmed the value of the tri-partite model of anxiety. All response systems contributed to the assessment of anxiety; different concordances for individual groups pointed to important group 127 differences. Although physiological arousal was also high for controls, they did not report themselves to be anxious. The further investigation of physiological arousal during eating (and decrease after eating) will increase our understanding of its parameters. Eating disorder subjects probably label this arousal as anxiety and, in addition, anorexics also show avoidance behaviours. Bulimics might do so (as previous research suggests) under certain conditions (e.g., the availability of unlimited food). As discussed in the Introduction, anorexia nervosa cannot be unambiguously conceptualized as phobic behaviour. Anorexics, for instance, not only avoid the feared stimulus (normal weight which they view as being ' fat ' ) , but also seem to pursue its opposite (extreme thinness). On the other hand, i t has been repeatedly noted (Foa & Kozak, 1985; Lang, 1985; Rachman, 1985) that phobias are by no means homogeneous in their underlying structure. For example, whereas simple phobics fear a specific stimulus, agoraphobics fear the anxiety which some situations presumably evoke. Perhaps anorexics and bulimics, in phobic terms, can be described as fearing the perceived 'harm' that they associate with eating: they are not afraid of the food but they are afraid of its effects. Eating also leads to immediate discomfort: many anorexics report feeling bloated and uncomfortable after eating, and bulimics report increased urges to binge after a meal. The anxiety increases when significant weight gain is expected from eating. If such formulations are correct, then the high level of anxiety found in anorexics and bulimics is not surprising (and somewhat similar to general levels of anxiety found in socially anxious subjects). One can also expect that the fear structures are less coherent and more discordant. The causes of initial weight loss probably differ for different anorexics. However, as clinicians observe and anorexics report, 128 anticipation of eating and eating itself results in anxiety. Lasegue (1873) already observed this anxiety during eating and referred to it as an "uneasiness after food" which leads to refusal of food (see introduction, p. 5). The present study confirms the high level of anxiety anorexics and bulimics report prior to and during eating. This anxiety appears to be init ial ly secondary to other causes of anorexia (e.g., a distorted body image or a fear of fatness), but may gain primary importance once weight loss has occurred. Therapy must be two-fold: restoration of weight to the natural set point and the establishment of appropriate eating behaviours, and, then additionally, treatment of other psychological correlates in order to maintain improvement and to prevent relapse. In practice, overlap between both objectives exists; for example, cognitive therapy will aid the anorexic in gaining weight and in maintaining it . Weight restoration is an important f irst step in the treatment of anorexia since psychotherapy with a severely underweight anorexic is unlikely to be successful. The establishment of appropriate eating behaviours is also a significant f irst goal in the treatment of bulimia. As long as issues of eating (what, when, and how much to eat and when to stop) are prominent, meaningful psychotherapy is unlikely. Hence, it follows that the anxiety surrounding eating must be addressed early in treatment. Since similarities between phobias and eating disorders exist, treatment approaches that have been found effective can be applied to anorexia nervosa and bulimia as well. The use of exposure with response prevention is one example of such a strategy: therapy found to be effective with obsessive-compulsive disorders was applied to treat bulimics who also vomited, which then led to the formulation of the anxiety reduction model of bulimia. Since anxiety about eating can be assessed with three response systems, therapy should address each. Progressive relaxation prior to and during eating (focussing on body 129 parts not directly involved in the act of eating) may reduce physiological arousal. Subsequent studies should investigate further the parameters of this arousal, and patients may benefit from information about it . If, for example, more studies confirm that physiological arousal and particularly heart rate increase significantly at the start of eating, then patients could be reassured about the normality of this increase. This increased arousal will then no longer be attributed to anxiety. Cognitive therapy can address dysfunctional thoughts regarding eating, the effects of food on weight, and body image. Garner and associates (1985) have adapted this approach from the treatment of depression. Education on weight regulation, effects of dieting and exercise, and body ideals in the cultural context have also been suggested (Garner et a l . , 1985). Anorexics who are very underweight may in particular benefit from an understanding of the effects of starvation and refeeding on the body. As already suggested above, the patient may no longer attribute physiological arousal during eating to anxiety, i f she knows of other likely causes. Lastly, particularly for anorexics, exposure to the feared stimulus, I.e., eating should increase food intake. As with phobias, a hierarchy of foods can be developed through which the anorexic proceeds. The hierarchy could start with small portions of perceived 'safe' foods and gradually include more 'dif f icult ' foods and larger, more normal portions. The gradual approach is derived from the starvation studies, where the subjects with unlimited access to food during eating appeared to have more problems with overeating (Keys et a l . , 1950). Of course, the minimum portions must take the physical condition of the patient Into account. The Inclusion of a variety of foods is an important part of normalization of eating. For example, so-called fattening foods such as chocolates can be a target of treatment. This approach was suggested by Rosen and Leitenberg (1985) 1n the 130 treatment of bulimia. Exposure to different foods is graduated on a hierarchy, where foods that provoke a lot of anxiety are placed on top of the l is t . Although bulimics have more difficulty with the termination of eating, exposure to different foods could be an important component to treatment. Bulimics also need to learn to eat 'normal' portions and include all foods; in fact, to become unrestrained eaters. This exposure must be accompanied with response prevention of purging. This treatment approach has been found to be effective (Leitenberg et a l . , 1984; Rossiter & Wilson, 1985). The anxiety model addresses only bingers and vomiters, partly because vomiting usually occurs temporally close to binging and thus reinforces it . Other methods of counteracting the binge such as the use of diuretics or strict dieting are more delayed. S t i l l , response prevention can follow the exposure to the food. In practice, it is more diff icult to implement this treatment. Antecedents of, e.g., the use of diuretics have to be identified and treated accordingly. If, for example, thoughts about impending weight gain lead to a decision to take diuretics, cognitive restructuring might be employed. Self monitoring, behavioural contracting, and other behavioural methods such as stimulus control might be used to decrease excessive dieting. To develop further such speculations, the progression of treatment can be considered. Initially, therapy can combine all treatments addressing the three response systems. For example, an emaciated anorexic might use relaxation skil ls and some cognitive restructuring when eating increasingly more 'dif f icult ' meals. After weight restoration and considerable normalization of eating, cognitive therapy might be emphasized and address dysfunctional beliefs that led to the original weight loss. Thus, different components of the treatment need to be emphasized at different times. Weight restoration by Itself is not sufficient to successfully treat anorexia nervosa; 1f the other concerns are not addressed relapse may be likely. In practice, anorexics will gain 131 weight in order to be released from hospital, but will subsequently lose it . Weight gain achieved by gradual exposure might give patients a sense of control over eating, counteracting their sense of ineffectiveness (see Bruch, 1973). Exposure to food and weight restoration must come early in the treatment sequence since psychotherapy is unlikely to be effective if a patient struggles continually with issues of eating. Similarly, for bulimics, response prevention must be present early in treatment in order to reduce binging and purging and give a sense of control to the patient. Within this framework, the treatment of anorexics and bulimics will be quite similar. The major difference is the addition of response prevention after food exposure. The other treatment components remain the same. Anorexia nervosa and bulimia appear to be multi-dimensional disorders. The variety of treatment approaches 1s therefore not surprising. An anxiety model of anorexia nervosa and bulimia can Integrate the different aspects of the disorders. Such a model will be described and with it the hypothetical development of anorexia nervosa and bulimia. Because of a variety of factors such as sociocultural demands, body image disturbances, and the demands of puberty, an adolescent or young adult woman (who may be vulnerable because of genetic and developmental factors) begins to diet and lose weight. After significant weight loss, the anorexic fears to regain the weight, starvation symptoms appear, and the struggle to remain thin centers around eating. Loss of control is feared. The anorexic who also binges experiences such loss of control. Weight 1s probably higher than with restricting-anorexics; however, physical sequelae of binging and often purging are also present. Anxiety during eating remains high and is relieved by purging. Purging thus reinforces binging and the frequency increases. Bulimics have not lost much weight although they might be 132 below set point; starvation symptoms will not be readily apparent. Biological pressures to eat can increase the likelihood of binging. Normal eating is largely absent. Dysfunctional beliefs regarding self-worth, body image perception and ideals, and fear of obesity strengthen diet behaviour and in turn binging. Again, eating becomes associated with anxiety. In this manner, cognitive, physiological, and behavioural variables interact to produce chaotic eating behaviour. This model not only recognizes the different variables implicated in the development of these disorders, but also allows more individual treatment approaches by emphasizing certain components depending on the individual patient. Of course, a model that includes so many factors has l i t t le value if i t cannot easily distinguish itself from alternative conceptualizations. In practice, treatment approaches are not intrinsically different. This perspective has, however, certain implications. First, the presence of anxiety during eating is crit ical in eating disorders. This anxiety is best measured with the three-response system, and consequently, treatment must address all three systems. Exposure to the feared stimulus should be emphasized. The model also poses some questions. Replication and generalization of the findings are essential; in addition, the parameters of this anxiety need to be studied further. Several issues need to be examined: the conditions that increase or decrease anxiety (e.g., anxiety probably increases if bulimics are allowed unlimited access to food); the choice and timing of treatment; the covariation of the anxiety response systems and the effect of treatment (e.g., does treatment produce desynchrony?); patterns of possible desynchrony of the anxiety response and its relationship to outcome; the description of new subgroups based on differences of the anxiety response systems among eating disorder groups. In summary, anxiety while eating appears to be critical 1n anorexia nervosa and bulimia. The investigation of this anxiety (with three 133 response systems) appears to be fruitful in increasing our theoretical understanding of anorexia nervosa and bulimia and for suggesting additional treatment strategies that have been found helpful in the treatment of anxiety disorders. 134 REFERENCES American Psychiatric Association. (1980). Diagnostic and  statistical manual of mental disorders (DSM-III) (3rd ed.). Washington, DC: Author. American Psychiatric Association. (1987). Diagnostic and  statistical manual of mental disorders (DSM-III-R) (rev. 3rd ed.). Washington, DC: Author. Beck, A. T. (1976). Cognitive therapy and the emotional  disorders. New York: International Universities Press. Bemis, K. M. (1983). A comparison of functional relationships in anorexia nervosa and phobia. In P. L. Darby, P. E. Garfinkel, D. M. Garner & D. V. Coscina (Eds), Anorexia  nervosa: recent developments and research (pp. 403-415). New York: Alan R. Liss. Bennett, W., & Gurin, J . (1982). The dieter's dilemma. New York: Basic Books. Beumont, P. J . , Booth, A. L., Abraham, S. F., Griffiths, D. A., & Turner T. R. (1983). A temporal sequence of symptoms in patients with anorexia nervosa: A preliminary report. In P. L. Darby, P. E. Garfinkel, D. M. Garner & D. V. Coscina (Eds), Anorexia nervosa: Recent developments and  research (pp. 129-136). New York: Alan R. Liss. Beumont, P. J . , George, G. C , & Smart. D. E. (1976). 'Dieters' and 'vomiters and purgers in anorexia nervosa. Psychological Medicine. 6. 617-622. Boskind-White, M., & White, W, C. (1983). Bulimarexia: The  binge/purge cvcle. New York: Norton. Bray, G. A. (1986). Effects of obesity on health and happiness. In K. D. Brownell & J . P. Foreyt (Eds.), Handbook of eating disorders (pp. 3-44). New York: Basic Books. Bruch, H. (1962). Perceptual and conceptual disturbances in anorexia nervosa. Psychosomatic Medicine, 24T 187-194. Bruch, H. (1973). Eating disorders. New York: Basic Books. Bruch, H. (1974). Perils of behaviour modification in treatment of anorexia nervosa. Journal of American Medical  Association. 280. 1419-1422. Button, E. J . , & Whitehouse, A. (1981). Subclinical anorexia nervosa. Psychological Medicine, 11, 509-516. Calloway, P., Fonagy, P., & Wakeling, A. (1983). Autonomic arousal in eating disorders: Further evidence for the clinical subdivision of anorexia nervosa. British Journal  of Psychiatry. 142. 38-42. Casper, R. C , Eckert, E. P., Halmi, K. A. , Goldberg, S. C , & Davis, J . M. (1980). Bulimia: Its incidence and clinical importance in patients with anorexia nervosa. Archives of  General Psvchiatrv. 37. 1030-1035. Crisp, A. H. (1983). Some aspects of the psychopathology of anorexia nervosa. In P. L. Darby, P. E. Garfinkel, D. M. Garner & D. V. Coscina (Eds), Anorexia nervosa: Recent  developments and research (pp. 15-28). New York: Alan R. Li ss. Crisp, A. H., Palmer, R. L., & Kalucy, R. S. (1976). How common is anorexia nervosa? A prevalence study. British  Journal of Psvchiatrv. 128. 549-554. Dixon, W. J . (Chief Ed), BMDP statistical software (1983 rev. printing). Berkeley: University of California Press. E l l i s , A. (1962). Reason and emotion in psychotherapy. New York: Lyle Stuart. Feighner, J . P., Robins, E. , Guze, S. B., Woodruff, R. A., Winokur, G., & Munoz, R. (1972). Diagnostic crieteria for use in psychiatric research. Archives of General  Psvchiatrv. 26. 57-63. Foa, E. B., & Kozak, M. J . (1985). Treatment of anxiety disorders: Implications for psychopathology. In A. H. Tuma & J . Maser (Eds.), Anxiety and the anxiety disorders (pp. 421-452). Hillsdale, N. J . : Lawrence Erlbaum Associates. Fairburn, C. G. (1985). Cognitive-behavioral treatment for bulimia. In D. M. Garner & P. E. Garfinkel (Eds), Handbook  of psychotherapy for anorexia nervosa and bulimia (pp. 160-192). New York: The Guildford Press. Fairburn, C. G., & Cooper, P. J . (1984). The clinical features of bulimia nervosa. British Journal of  Psvchiatrv. 144. 238-246. Fairburn, G. C , Cooper, Z. , & Cooper, P. J . (1986). The clinical features and maintenance of bulimia nervosa. In K. D. Brownell & J . P. Foreyt (Eds.), Handbook of eating  disorders (pp. 389-404). New York: Basic Books. Forth, A. E., & Hare, R. D. (1986). Effects of anticipation  and escape on autonomic responses to phobic stimuli. Paper presented at the conference of the Canadian Psychological Association, Toronto, Canada. Freeman, R. J . , Beach, B., Davis, R., & Solyom, L. (1985). The prediction of relapse in bulimia nervosa. Journal of  Psychiatric Research. 19. 349-353. Furedy, J . J . , & Heslegrave, R. J . (1983). A consideration of recent criticisms of the T-wave amplitude index of myocardial sympathetic activity. Psvchophvsioloqv. 20. 204-211. Garfinkel, P. E., & Garner, D. M. (1982). Anorexia nervosa: a  multidimensional perspective. New York: Brunner/Mazel. Garfinkel, P. E., Moldofsky, H., & Garner, D. M. (1980). The heterogeneity of anorexia nervosa. Archives of General  Psvchiatrv. 37. 1036-1040. Garner, D. M., & Bemis, K. M. (1985). Cognitive therapy for anorexia nervosa. In D. M. Garner & P. E. Garfinkel (Eds), Handbook of psychotherapy for anorexia nervosa and  bulimia (pp. 107-146). New York: The Guildford Press. Garner, D. M., Garfinkel, P. E., & Olmsted, M. P. (1983). An overview of sociocultural factors in the development of anorexia nervosa. In P. L. Darby, P. E. Garfinkel, D. M. Garner & D. V. Coscina (Eds), Anorexia nervosa: Recent  developments and research (pp. 65-82). New York: Alan R. Li ss. Garner, D. M., Olmsted, M. P., Polivy, J . , & Garfinkel, P. E. (1984). Comparison between weight-preoccupied women and anorexia nervosa. Psychosomatic Medicine, 46, 255-266. Garner, D. M., Rockert, W., Olmsted, M. P., Johnson, C , & Coscina, D. V. (1985). Psychoeductlonal principles in the treatment of bulimia and anorexia nervosa. In D. M. Garner & P. E. Garfinkel (Eds), Handbook of psychotherapy  for anorexia nervosa and bulimia (pp. 513-572). New York: The Guildford Press. Goodsitt, A. (1985). Self psychology and the treatment of anorexia nervosa. In D. M. Garner & P. E. Garfinkel (Eds), Handbook of psychotherapy for anorexia nervosa and  bulimia (pp. 55-82). New York: The Guildford Press. Gormally, J . (1984). The obese binge eater: diagnosis, etiology, and clinical Issues. In R. C. Hawkins, W. J . Fremouw &P. F. Clement (Eds), The binge-purge syndrome (pp. 47-73). New York: Springer. Grey, S. J . , Rachman, S., & Sartory, G. (1981). Return of fear: The role of Inhibition. Behaviour Research and  Therapy. 19. 135-143. Gull, W.W. (1874). Anorexia nervosa. Trans. Clin. Soc. (London), 7:22. Reprinted in R. M. Kaufman & M. Heiman (Eds.), Evolution of psychosomatic concepts.  Anorexia nervosa: A paradigm (pp. 132-138). New York: International Universities Press, 1964. Halmi, K. A. (1985). Review: Classfication of the eating disorders. Journal of Psychiatric Research, 19T 113-120. Halmi, K. A., Falk, J . R., & Schwartz, E. (1981). Binge-eating and vomiting: a survey of a college population. Psychological Medicine. 11. 697-706. Hatsukami, D., Eckert, E., Mitchell, J . E., & Pyle, R. (1984). Affective disorder and substance abuse in women with bulimia. Psychological Medicine. 14. 701-704. Hare, R. D. (1973). Orienting and defensive responses to visual stimuli. Psvchophvsiologv. 10. 453-464. Hare, R. D. (1978). Electrodermal and cardiovascular correlates of psychopathy. In R. D. Hare & D. Schalling (Eds.), Psychopathic behaviour: Approaches to research (pp. 107-143). Chichester, England: Wiley. Hare, R. D., & Blevings, G. (1975). Defensive responses to phobic stimuli. Biological Psychology. 3. 1-13. Hart, K. J . , & Ollendick, T. H. (1985). Prevalence of bulimia in working and university women. American Journal of  Psychiatry. 142. 951-854. Healy, K., Conroy, R. M., & Walsh, N. (1985). The prevalence of binge-eating and bulimia in 1063 college students. Journal of Psychiatric Research. 19. 161-166. Herman, C. P., Pol ivy, J . , Lank, C. N., & Heatherton, T. F. (1987). Anxiety, hunger, and eating behavior. Journal of  Abnormal Psychology. 96. 264-269. Hinz, L. D., & Williamson, D. A. (1987). Bulimia and depression: A review of the affective variant hypothesis. Psychological Bulletin. 102. 150-158. Hodgson, R., & Rachman, S. (1974). I. Synchrony and desynchrony in fear and avoidance. Behaviour Research and  Therapy. 12. 311-318. Holzl, R., & Lautenbacher, S. (1984). Psychophysiological indices of the feeding response in anorexia nervosa patients. In K. M. Pirke & D. Ploog (Eds.), The  psvchobioloqv of anorexia nervosa (pp. 93-113). Berlin, W. Germany: Springer. Hsu, L. K. (1980). Outcome of anorexia nervosa. Archives of  General Psychiatry. 37. 1041-1046. Hudson, J . I., Pope, H. G., & Jonas, J . M. (1985). Antidepressant treatment of bulimia. Advances in Behaviour  Research and Therapy, 7. 173-179. Johnson, C. L., Stuckey, M. K., Lewis, L. D., & Schwartz, D. M. (1982). Bulimia: A descriptive survey of 316 cases. International Journal of Eating Disorders. 2. 3-16. Johnson, W. G., Schlundt, D. G., Kelley, M. L., & Ruggiero, L. (1984). Exposure with response prevention and energy regulation in the treatment of bulimia. The International  Journal of Eating Disorders, 3. 37-46. Katzman, M. A., & Wolchik, S. A. (1984). Bulimia and binge eating in college women: A comparison of personality and behavioral characteristics. Journal of Consulting and  Clinical Psychology. 52. 423-428. Keesey, R. E. (1986). A set-point theory of obesity. In K. D. Brownell & J . P. Foreyt (Eds.), Handbook of eating  disorders (pp. 63-87). New York: Basic Books. Keys, A., Brozek, J . , Henschel, A. , Mickelsen, 0., & Taylor, H. L. (1950). The biology of human starvation. Minneapolis: University of Minnesota Press. Kraus, B. (1983). The Barbara Kraus 1983 calorie guide to  brand names and basic foods (rev. ed). Scarborough, Ontario: New American Library of Canada. Lacey, J . H., & Gibson, E. (1985). Controlling weight by purgation and vomiting: A comparative study of bulimics. Journal of Psychiatric Research. 19. 337-341. Lacey, J . I. (1967). Somatic response patterning and stress: Some revisions of activation theory. In M. H. Appley & R. Trumbull (Eds.), psychological stress: Issues in research (pp. 14-42). New York: Appleton-Century-Crofts. Lang, P. J . (1985). The cognitive psychophysiology in emotion: fear and anxiety. In A. H. Tuma & J . Maser (Eds.), Anxiety and the anxiety disorders (pp. 131-170). Hillsdale, N. J . : Lawrence Erlbaum Associates. Lang, P. J . , Levin, D. N., Miller, G. A., & Kozak, M. J . (1983). Fear bevahior, fear imagery, and the psychophysiology of emotion: The problem of affective response integration. Journal of Abnormal Psychology, 92. 276-306. Lasegue, C. (1873). De I'anorexie hysterlque. Arch. Gen, de.  Med. 385. Reprinted in R. M. Kaufman & M. Heiman (Eds), Evolution of psychosomatic concepts. Anorexia nervosa: A  paradigm (pp. 143-155). New York: International Universities Press, 1964. Leitenberg, H., Gross, J . , Peterson, J . & Rosen, J . C. (1984). Analysis of an anxiety model and the process of change during exposure plus response prevention treatment of bulimia nervosa. Behavior Therapy, 15. 3-20. Loro, A. D. (1984). Binge-eating: a cognitive-behavioral treatment approach. In R. C. Hawkins, W. J . Fremouw & P. F. Clement (Eds), The binge-purge syndrome (pp. 183-210). New York: Springer. Luck, P., & Wakeling, A. (1980). Altered thresholds for thermoregulatory sweating and vasodilation in anorexia nervosa. British Medical Journal. 281. 906-908. Mackay W., & Lidell , A. (1986). An investigation into the matching of specific agoraphobic response characteristics with specific types of treatment. Behaviour Research and  Therapy. 24. 361-364. McCann, B. S., Woolfolk, R. L., & Lehrer, P. M. (1987). Specificity in response to treatment: A study of interpersonal anxiety. Behaviour Research and Therapy. 25. 129-136. Metropolitan Life Tables (1983). Metropolitan Life Insurance Company. Mitchell, J . E. (1986). Bulimia: Medical and physiological aspects. In K. D. Brownell & J . P. Foreyt (Eds.), Handbook  of eating disorders (pp. 379-388). New York: Basic Books. Mizes, J . S. (1985). Bulimia: A review of its symptomatology and treatment. Advanced Behaviour Research and Therapy. 7. 91-142. Obrist, P. A. (1981). Cardiovascular psvchophysiologv. New York: Plenum Press. Owen, W. P., Halmi, K. A., Gibbs, J . , & Smith, G. P. (1985). Satiety responses in eating disorders. Journal of  Psychiatric Research. 19. 279-284. Piran, N., Kennedy, S., Garfinkel, P. E., & Owens, M. (1985). Affective disturbance in eating disorders. Journal of  Nervous and Mental Disease. 173. 395-400. Polivy, J . , & Herman, C. P. (1985). Dieting and binging: a causal analysis. American Psychologist. 40. 193-201. Polivy, J . , Herman, C. P., Olmsted, M. P., & Jazwinsky, C. (1984). Restraint and binge eating. In R. C. Hawkins, W. J . Fremouw & P. F. Clement (Eds), The binge-purge syndrome (pp. 104-122). New York: Springer. Porges, S. W., McCabe, P. M. & Yongue, B. G. (1982). Respiratory heart rate interations: Psychophysiological implications for pathophysiology and behavior. In J . T. Cacioppo & R. E. Petty Eds.), Perspectives in cardiovascular psychophysiology pp. 223-264). New York: Guildford Press. Pyle, R. L , Mitchell, J . E., & Eckert, E. D. (1981). Bulimia: A report of 34 cases. Journal of Clinical  Psychiatry. 42. 60-64. Rachman, S. J . (1985). The treatment of anxiety disorders: A critique of the implications for psychopathology. In A. H. Tuma & J . Maser (Eds.), Anxiety and the anxiety  disorders (pp. 453-462). Hillsdale, N. J . : Lawrence Erlbaum Associates. Rachman, S., & Hodgson, R. (1974). II. Desynchrony in measures of fear. Behaviour Research and Therapy. 12. 319-326. Rampling, D. (1985). Ascetic ideals and anorexia nervosa. Journal of Psychiatric Research. 19T 89-94. Rau, J . H., & Green, R. S. (1984). Neurological factors affecting binge eating: Body over mind. In R. C. Hawkins, W. J . Fremouw & P. F. Clement (Eds), The binge-purge  syndrome (pp. 123-143). New York: Springer. Robinson, R. G., Tortosa, M., Sullivan, J . , Buchanan, E., Andersen, A. E., & Folstein, M. F. (1983). Quantitative assessment of psychologic state of patients with anorexia nervosa or bulimia: Response to caloric stimulus. Psychosomatic Medicine. 45. 283-292. Rosen, J . C , & Leitenberg, H. (1985). Exposure plus response prevention treatment of bulimia. In D. M. Garner & P. E. Garfinkel (Eds), Handbook of psychotherapy for anorexia  nervosa and bulimia (pp. 193-209). New York: The Guildford Press. Rosen, J . C , Leitenberg, H., Gross, J . , & Willmuth, M. (1985). Standardized test meals in the assessment of bulimia nervosa. Advances in Behaviour Research and  Therapy. 7. 181-197. Rossiter, E. M., & Wilson, G. T. (1985). Cognitive restructuring and response prevention in the treatment of bulimia nervosa. Behaviour Research and Therapy. 23. 349-359. Ruderman, A. J . (1986). Bulimia and irrational beliefs. Behavior Research and Therapy. 24r 193-197. Russell, G. F. (1970). Anorexia nervosa: Its Identity as an illness and its treatment. In J . H. Price (Ed), Modern  trends in psychological medicine. 2. London: Butterworths. Russell, G. F. (1979). Bulimia nervosa: An ominous variant of anorexia nervosa. Psychological Medicine. 9. 429-448. Russell, J . A. (1980). A circumplex model of affect. Journal  of Personality and Social Psychology. 39. 1161-1178. Russell, J . A. , & Mehrabian, A. (1974). Distinguishing anger and anxiety in terms of emotional response factors. Journal of Consulting and Clinical Psychology. 42t 79-83. Salkind, M. R., Fincham, J . , & Silverstone, T. (1980). Is anorexia nervosa a phobic disorder? A psychophysiological enquiry. Biological Psychiatry. 15. 803-808. Sargent, J . , Liebman, R., & Silver, M. (1985). Family therapy for anorexia nervosa. In D. M. Garner & P. E. Garfinkel (Eds), Handbook of psychotherapy for anorexia nervosa and  bulimia (pp. 280-307). New York: The Guildford Press. Schlesier-Stropp, B. (1984). Bulimia: A review of the literature. Psychological Bulletin. 95. 247-257. Siddle, D. A., & Turpin, G. (Eds.), (1980). Measurement, quantification, and analysis of cardiac activity. In I. Martin & P. H. Venables (Eds.), Techniques in  psychophysiology. New York: John Wiley & Sons, Ltd. Silverman, J . A. (1985). Richard Morton, 1637-1698, limner of anorexia nervosa: His l i fe and times. A tercentenary essay. Journal of Psychiatric Research, 19T 83-88. Slade, P. (1985). A review of body-image studies in anorexia nervosa and bulimia nervosa. Journal of Psychiatric  Research. 19. 255-265. Solyom, L., Thomas, C. D., Freeman, R. J . , & Miles, J . E. (1983). Anorexia nervosa: Obsessive-compulsive disorder or phobia? A comparative study. In P. L. Darby, P. E. Garfinkel, D. M. Garner & D. V. Coscina (Eds), Anorexia  nervosa: recent developments and research (pp. 137-148). New York: Alan R. Liss. Striegel-Moore, R. H., Silberstein, L. R., & Rodin, J . (1986). Toward an understanding of risk factors for bulimia. American Psychologist, 41. 246-263. Strober, M. (1986). Anorexia nervosa: History and psychological concepts. In K. D. Brownell & J . P. Foreyt (Eds.), Handbook of eating disorders (pp. 231-246). New York: Basic Books. Stunkard, A. (1959). Eating patterns and obesity. Psychiatric  Quarterly. 33. 284-295. 142 Taylor, B., & Agras, S. (1981). Assessment of phobia. In D. H. Barlow (Ed.), Behavioral assessment of adult disorders (pp. 181-208). New York: Guilford Press. Toner, B. B., Garfinkel, P. E., & Garner, D. M. (1986). Long-term follow-up of anorexia nervosa. Psychosomatic  Medicine. 48. 520-529. Wakeling, A. (1985). Neurobiological aspects of feeding disorders. Journal of Psychiatric Research, 19. 191-201. Williamson, D. A., Kelley, M. L., Davis, C. J . , Ruggiero, L., & Veitia, M. C. (1985). The psychophysiology of bulimia. Advances in Behaviour Research and Therapy. 7. 163-172. Winer, B. J . (1971). Statistical principles in experimental design (2nd Ed.) (pp. 514-603). New York: McGraw-Hill. Ziolko, H. U. (1985). Bulimie. Fortschritte der Neuroloqie - Psvchiatrie. 53. 231-258. Appendix 1 Diagnostic Criteria for 307.10 Anorexia Nervosa DSM-III-R (1987) A. Refusal to maintain body weight over a minimal normal weight for age and height, e.g., weight loss leading to maintenance of body weight 15% below that expected; or failure to make expected weight gain during period of growth, leading to body weight 15% below that expected. B. Intense fear of gaining weight or becoming fat, even though underweight. C. Disturbance in the way in which one's body weight, size, or shape is experienced, e.g., the person claims to "feel fat" even when emaciated, believes that one area of the body is "too fat" even when obviously underweight. D. In females, absence of at least three consecutive menstrual cycles when otherwise expected to occur (primary or secondary amenorrhea). (A woman is considered to have amenorrhea i f her periods occur only following hormone, e.g., estrogen, administration.) 144 Appendix 2 Diagnostic Criteria for 307.51 Bulimia Nervosa DSM-III-R (1987) A. Recurrent episodes of binge eating (rapid consumption of a large amount of food in a discrete period of time). B. A feeling of lack of control over eating behavior during eating binges. C. The person regularly engages in either self-induced vomiting, use of laxatives or diuretics, strict dieting or fasting, or vigorous exercise in order to prevent weight gain. D. A minimum average of two binge eating episodes a week for at least three months. E. Persistent overconcern with body shape and weight. Appendix 3 Condition 2 Eight subjects (2 in each group) participated in this condition. The procedure was the same as in the main study except that subjects were exposed passively to the stimuli and were not reqired to respond. Thus, subjects viewed the slides but did not rate them, and were exposed to the food but did not eat i t . The procedure was modified in the following way: Part 1: Subjects received Self-Report 1 twice separated by a period of relaxation. They then viewed the slides but did not rate them. After presentation of the slides, subjects completed Self-Report 1 two more times, again separated by a period of relaxation. Part 2: Subjects completed twice Self-Report 1, but no additional questionnaires were given. Then, food was brought into the room and placed beside the subject. (It had been explained to the subjects that the purpose of the study was to measure the effect of various stimuli). The experimenter left the room and subjects were asked to complete Self-Report 1. Following a few minutes of relaxation, another Self-Report 1 was completed. The food was then removed from the room and subjects completed two more times Self-Report 1 separated by a period of relaxation. Because only two subjects per group participated in this condition, it was not possible to analyze the data in a formal way. The data were summarized and will be presented in the following paragraphs. Again, subjects indicated levels of pleasure, arousal, and anxiety on self-report measures given repeatedly. Eating disorder subjects reported higher levels of pleasure and lower levels of arousal and anxiety (when compared with the data from the main study). Control subjects responded in a similar fashion in both studies. Psychophysiological measures were also obtained. Following the presentation of slides, heart rate decreased or stayed the same in most subjects. The average change was -.16 bpm. During rest periods heart rate remained stable and did not increase (as was observed in the main study). During presentation of food, heart rate increased somewhat except for bulimic subjects; this increase was, however, modest and mean heart rates did not appear different from those obtained during rest periods or task 1. In fact, average heart rates during food presentation were higher when compared with mean heart rates during slides for the bulimic group; however, for the controls and the restricting-anorexics the opposite was found. Heart rate change scores (during self-reports) were mostly positive. Tonic levels of skin conductance showed l i t t le change during rest periods. Skin conductance increased init ial ly during task 2 and was higher than during task 1. The pattern of skin conductance following the initial minute of task 2 was variable. Change scores obtained during self-report measures were mostly positive except at those times following some activity, e.g. , the experimenter entering the room to bring or remove materials. This result is consistent with those obtained in the main study. Amplitudes to the self-report measures appeared to decrease in part 2 of the study. At times following much activity, amplitudes were quite small. Amplitudes to other stimuli could not be obtained due to the changed procedure. Spontaneous skin conductance fluctuations were scored in the various 146 intervals. Fewer fluctuations were observed in the rest periods than were observed in the main study ( if one compares absolute numbers). A large variation of scores was obtained during self-reports and exposure to food. During the latter interval some eating disorder subjects were non-responsive. In summary, eating disorder subjects reported a more positive affective state and less physiological arousal than when required to eat. Of course, they were also less active. Some of the responses (such as skin conductance amplitude) help to interpret the data obtained in the main study (effect of stimulation, e.g. , experimenter enters the room). The small number of subjects does not allow definitive conclusions. The findings, however, suggest that the results obtained in the main study were due to the demand of eating. 147 Appendix 4 Self-Report Inventory 1 which contains the scale assessing emotional state will not be published here for copyright reasons. The interested reader can find the scale in Mehrabian, A., & Russell, J.A. An approach to environmental psychology. Cambridge, Massachusetts: M. I. T. Press, 1974. 148 Appendix 5 Self-Report Inventory 2 - Version A (Prior to Eating) Please indicate how you feel at this moment by placing a checkmark along the line (Example : s/ : ). 1. I find it easy to think about other things than the food I am about to eat. very diff icult : : : : : : : : very easy 2. I am afraid to eat at this moment. not at all : : : : : : : :_ very much 3. Eating this meal will not affect my weight significantly, disagree : : : : : : : : agree 4. I am afraid that I will become fat i f I eat this meal, not at all : : : : : : : : very much 5. How strong is your urge to overeat right now? no urge at all : : : : : : : : very strong 6. Given the strength of your urge to overeat, how likely is it that you will give in and overeat, i f given the opportunity? very likely : : : : : : : : unlikely 149 Appendix 6 Self-Report Inventory 2 - Version B (After Eating) Please indicate how you feel at this moment by placing a checkmark along the line (Example :_^ Z_: ). 1. All my thoughts concentrate on the food I just ate. very much : : : : : : : : not at all 2. This meal will increase my weight. not at all : : : : : : : : very much 3. I am afraid that I will become fat. very much : : : : : : : : not at all 4. How strong is your urge to overeat right now? no urge : : : : : : : : very strong 5. Given the strength of your urge to overeat, how likely is it that you will give in and binge, i f given the opportunity? unlikely : : : : : : : : very likely 150 Appendix 7 Informed Consent Form Principal Investigator: D. Papageorgis, Ph.D. Project Manager: B. Buree, M.A. Title: Anxiety during Eating in Anorexia Nervosa and Bulimia CONSENT FORM I, , voluntarily give consent to participate in a study involving the assessment of factors that contribute to anxiety in anorexia nervosa and bulimia. The procedures to be followed have been explained to me and I understand them. They are as follows: 1. During the f irst session, I will complete two questionnaires and provide background information regarding my eating behaviours. 2. During the second session, I will complete two different tasks. First I will rate ten landscape paintings according to my preference. Later I will eat a small meal. Throughout the session the following psychophysiological recordings will be obtained: Heart rate will be measured with 2 electrodes attached on the sternum (middle of ribs); skin conductance (sweating) will be measured with 2 electrodes attached to two fingers; muscle tension will be measured with 2 electrodes attached on the forehead; stomach activity will be measured with 2 electrodes (1 attached on the stomach, the other on the leg); respiration will be measured with a belt around the ribcage; and pulse amplitude will be measured with an electrode on the thumb. I will also complete a series of self-report measures. Parts of the session will be videotaped. All psychophysiological recording procedures used are standard and safe, and involve minimal risks. There may be slight discomfort experienced when the electrodes are attached (the skin needs to be cleaned thoroughly). My total time commitment is about 2 1/2 - 3 hours. All data collected will be kept strictly confidential and results will be reported as group findings only. Refusal to participate or withdraw from the study will not jeopardize my psychiatric treatment in any way. My questions concerning the study have been answered to my satisfaction. Date Witness Subject 151 Appendix 8 Means of Self-Reported Pleasure (Standard Deviations in Parentheses) Group c* RA* BA* B* Margin; Total: Time 1 6.75 (1.47) 4.75 (0.95) 5.45 (1.49) 5.15 (0.81) 5.52 Time 2 6.22 (1.33) 4.63 (0.95) 5.42 (1.52) 5.02 (0.84) 5.32 Time 3 6.75 (0.95) 4.55 (1.33) 5.23 (0.92) 5.40 (0.83) 5.48 Time 4 6.23 (1.14) 4.15 (1.37) 5.55 (1.26) 5.07 (1.00) 5.25 Time 5 6.60 (1.08) 3.87 (1.45) 4.53 (1.22) 5.05 (1.23) 5.01 Time 6 6.57 (1.22) 3.77 (1.37) 4.82 (1.81) 4.71 (0.89) 4.98 Time 7 6.72 (1.04) 4.32 (1.47) 4.38 (1.47) 4.93 (1.34) 5.09 Time 8 6.53 (1.23) 4.00 (1.29) 4.18 (1.28) 4.62 (1.10) 4.83 Marginal Totals. 6.55 4.25 4.95 5.00 5.19 * C - Controls RA- Restricting Anorexics BA- Bulimic Anorexics B - Bulimics 152 Appendix 8, continued Means of Self-Reported Arousal (Standard Deviations in Parentheses) Group C RA BA B Margin; Total Time 1 3.90 (1.12) 5.20 (0.96) 4.28 (1.86) 4.85 (1.18) 4.56 Time 2 3.47 (1.29) 5.48 (1.01) 4.20 (1.60) 4.80 (1.27) 4.49 Time 3 4.00 (0.67) 5.93 (1.29) 4.57 (1.45) 4.93 (1.26) 4.86 Time 4 3.05 (0.83) 5.42 (1.17) 4.85 (1-51) 4.23 (1.40) 4.39 Time 5 3.98 (0.90) 6.18 (0.99) 6.17 (0.95) 5.45 (0.68) 5.45 Time 6 3.58 (1.22) 6.30 (1.46) 5.55 (1.86) 4.90 (0.89) 5.08 Time 7 4.63 (0.84) 6.10 (1.24) 6.00 (1-71) 5.67 (0.87) 5.60 Time 8 3.67 (0.58) 5.58 (1.69) 5.43 (1.58) 4.65 (0.88) 4.83 Marginal Totals 3.79 5.78 5.13 4.94 153 Appendix 8, continued Means of Self-Reported Anxiety (Standard Deviations in Parentheses) Group C RA BA B Marginal Totals Time 1 3.67 6.07 4.93 5.83 5.12 (1.87) (1.17) (2.02) (1.67) Time 2 3.47 5.77 5.20 5.67 5.03 (1.55) (1.35) (2.11) (2.11) Time 3 2.50 5.93 3.87 5.27 4.39 (0.96) (1.68) (1.62) (1.45) Time 4 2.50 6.23 4.30 5.40 4.61 (0.89) (1.98) (2.02) (1.78) Time 5 2.43 7.00 6.57 6.10 5.52 (1.02) (1.24) (1.03) (1.38) Time 6 2.27 6.67 5.63 5.93 5.13 (1.12) (1.45) (2.32) (1.80) Time 7 2.60 6.07 5.90 6.17 5.27 (1.53) (1.99) (2.18) (1.50) Time 8 2.30 6.73 5.73 6.07 5.21 d-12) (1.59) (1-62) (1.92) Marginal 2.72 6.36 5.27 5.80 Totals 154 Appendix 9 Results of Self-Report Inventory 2 As explained In the text, the questionnaires were designed to assess more formally distorted beliefs eating disorder subjects may have. No psychometric properties were established. The Self-Report Inventory 2 was given after Self-Report Inventory 1 during task 2 only. Therefore, each subject completed version A (prior to the task) twice and, similarly, version B (after task 2) twice. Subjects rated on nine point scales how statements about eating applied to them. These ratings (one for each statement) were subjected to 4 by 2 (groups by times) repeated measures analyses of variance. These statistical analyses (F-ratios and Tukey comparisons of marginal means where applicable) should, however, be seen as a more formal way of summarizing those statements rather than form the basis for interpreting and contrasting thought contents of the groups. The means and standard deviations of the ratings at each time are summarized in Table 15. Self-Report Inventory 2 - Version A. (prior to eating): This inventory contained six questions. The analyses of the ratings for each question are presented in this section. Statement 1: I find it easy to think about other things than the food I am about to eat. (very diff icult -9- -1- very easy). The F-ratio for between group comparison was significant, £ (3,36) = 5.52, p_ < .05. Control subjects gave a significantly lower rating than the restricting-anorexics (p_ < 0.05) and the bulimic-anorexics (p_ < 0.01), but not the bulimic subjects (p. > 0.05). The ratings given on the two occasions were similar. The group by time interaction was significant (p. < 0.05). The control and restricting-anorexic groups gave a higher rating the second time, whereas the other two groups gave a lower rating. Statement 2: I am afraid to eat at this moment, (not at all -1- -9-very much). Again, the F-rat1o for between group comparison was significant (F (3,36) = 13.98, p_ < 0.0001). The control group gave a significantly lower rating than each eating disorder group (p. < 0.01 for Controls vs Restricting-Anorexics and Bulimic-Anorexics and p_ < 0.05 for Controls vs Bulimics), indicating a lack of anxiety regarding eating. These ratings were similar at both times, and the group by time interaction was not significant. Statement 3: Eating this meal will not affect my weight significantly (disagree -9- -1-agree). The F-ratio for between group comparisons was significant (£ (3,36) = 4.77, p. < 0.01). The other F-ratios were not significant. The control subjects gave the lowest rating. This rating was significantly lower than the one given by the restricting-anorexics (p_ < 0.01) and the bul1mic-anorexics (p. < 0.05). The rating by bulimic subjects fell between the rating of the control subjects and the ratings by the anorexic subjects. None of the group mean comparisons with the bulimic subjects was significant. Statement 4: I am afraid that I will become fat 1f I eat this meal (not at all -1- -9- very much). The F-ratio for between group comparisons was significant ( £ (3,36) = 9.34, p_ < 0.0001). Control subjects expressed l i t t le fear of fatness whereas restricting and bulimic anorexics gave a higher rating (p. < 0.01). The rating by bulimic subjects was lower than the ones by the other two eating disorder groups, but s t i l l significantly higher than the 155 Table 15 Self-Report Inventory 2 - Version A Means of Ratings for Time 1 and Time 2; (Standard Deviations in Parentheses) C* RA* BA* B* Question 1: "I find i t easy to think about other things than the food I am about to eat." Time 1 3.4 Time 2 3.9 (2.1) 6.7 (2.6) 7.6 (1.3) 5.7 (2.5) (2.4) 7.1 (2.4) 6.7 (2.5) 4.6 (2.7) Question 2: "I am afraid to eat at this moment." Time 1 1.7 (1.6) 7.0 (2.2) 6.8 (2.4) 4.4 (2.5) Time 2 1.2 (0.4) 5.8 (2.8) 6.4 (2.5) 4.1 (2.8) Question 3: "Eating this meal will not affect my weight significantly." Time 1 2 TIME 2 2 7 (2.5) 6.5 (2.2) 5.8 (2.3) 4.0 (2.7) 3 (2.2) 5.9 (3.0) 5.4 (2.4) 4.1 (3.1) Question 4: "I am afraid that I will become fat i f I eat this meal." Time 1 1.6 Time 2 1.3 (0.8) 5.6 (2.2) 5.8 (2.7) 4.6 (2.6) (0.5) 5.7 (2.1) 5.9 (2.6) 4.2 (2.6) Question 5: "How strong is your urge to overeat right now?" Time 1 2.9 Time 2 2.8 (2.5) 1.8 (2.5) 3.3 (2.5) 5.0 (2.6) (2.9) 1.9 (2.2) 4.4 (3.4) 4.6 (2.7) Question 6: "Given the strength of your urge to overeat, how likely is it that you will give in and overeat, i f given the opportunity?" Time 1 4.3 Time 2 3.8 (2.5) 2.4 (2.5) 5.2 (2.7) 4.9 (3.2) (2.0) 1.6 (1.6) 4.7 (3.1) 4.0 (2.8) * C - Controls RA- Restricting Anorexics BA- Bulimic Anorexics B - Bulimics Table 15, continued Self-Report Inventory 2 - Version B Means of Ratings at Time 1 and Time 2; Standard Deviations in Parentheses C RA BA B Question 1: "All my thoughts concentrate on the food I just ate." Time 1 3 Time 2 .8 (1.5) 6.5 (2.8) 7.2 (2.3) 6.8 (2.3) 1.5 (0.7) 4.8 (2.5) 6.1 (2.1) 4.8 (2.6) Question 2: "This meal will increase my weight." Time 1 1.7 (0.9) 3.4 (2.3) 5.7 (2.4) 4.9 (2.6) Time 2 1.5 (0.7) 3.7 (2.2) 5.0 (2.6) 3.7 (2.6) Question 3: "I am afraid that I will become fat." Time 1 2.5 (2.0) 5.2 (2.8) 7.4 (2.3) 5.3 (2.9) Time 2 2.0 (1.6) 6.1 (2.0) 6.8 (2.6) 5.0 (3.0) Question 4: "How strong is your urge to overeat right now?" .1 (1.0) 1.9 (2.5) 5.1 (3.1) 4.7 (3.1) .5 (0.7) 1.3 (0.9) 5.0 (3.3) 4.3 (2.8) Time 1 2 Time 2 1 Question 5: "Given the strength of your urge to overeat, how likely is 1t that you will give in and binge, if given the opportunity?" Time 1 2 Time 2 2 .9 (2.0) 2.6 (3.4) 5.6 (2.9) 4.8 (2.8) .4 (2.0) 1.5 (1.6) 5.5 (3.2) 5.0 (2.7) one by control subjects (p_ < 0.05). The other F ratios were not significant. Statement 5: How strong is your urge to overeat right now? (no urge at all -1- -9- very strong). No differences emerged in this analysis. Statement 6: Given the strength of your urge to overeat, how likely is it that you will give in and overeat, i f given the opportunity? (very likely -9- -1-unlikely). No group differences emerged. The rating given on the f irst occasion (4.2) was significantly (p_ < 0.05) higher than the rating given prior to eating (3.52). Self-Report Inventory 2 - Version B (after eating): This inventory consisted of five questions. Again, the ratings for each question will be described. Statement 1: All my thoughts concentrate on the food I just ate (very much -9- -1- not at a l l ) . Both the F-ratios for groups (F (3,36) = 8.29, p. < 0.001) and over time (F (1,36) = 26.83, p_ < 0.0001) were significant. The control subjects gave a significantly lower rating than the other three groups (p < 0.01). The ratings also declined over time (p. < 0.01). Statement 2: This meal will increase my weight (not at all -1- -9- very much). The analysis of this question yielded three significant F-ratios (groups: £ (3,36) = 5.68, p. < 0.01; time: £ (1,36) = 7.04, p_ < 0.05); time by group: £ (3,36) = 3.62, p_ < 0.05). The control subjects gave the lowest rating to this question and their rating was significantly lower than the ratings given by the bulimic subjects (p < 0.05) and by the bulimic-anorexic subjects (p < 0.01). The rating by restricting-anorexics was the second lowest and not significantly different from the other groups (p > 0.05). The ratings given immediately after eating (3.92) were significantly higher ( p_ < 0.05) than the ratings given at the end of the study (3.475). Although the interaction was significant, the ordering of the groups remains. Statement 3: I am afraid I will become fat (very much -9--1- not at a l l ) . The F-ratio for between groups was significant ( £ (3,36) = 8.63, p. < 0.001). The control subjects gave significantly lower ratings than the anorexic (p. < 0.01) and bulimic (p. < 0.05) subjects. The highest rating was given by the bulimic-anorexics (7.1) compared with the rating by controls (2.25). The other F tests were not significant. Statement 4: How strong is your urge to overeat right now? (no urge -1--9- very strong) The F tests for groups (£ (3,36) = 5.99, p. < 0.01) and time (£ (1,36) = 4.33, p. < 0.05) were significant. The lowest rating was given by the restricting-anorexic group and it was significantly lower than the ratings by the bulimic-anorexic and bulimic groups (p. < 0.05). The control subjects gave the second lowest rating, and 1t was significantly lower than the rating by the bul 1mic-anorexic group (p_ < 0.05). It failed to reach significance when compared with the bulimic group. The ratings by the restricting-anorexic and control subjects were very similar. The ratings declined as time elapsed after eating. Although this decline was significant (p_ < 0.05), the actual ratings did not change markedly ( from 3.45 to 3.025). Statement 5: Given the strength of your urge to overeat, how likely is it that you will give in and binge, i f given the opportunity? (unlikely -1--9- very likely). Only the F test for groups yielded a significant ratio (£ (3,36) = 4.69, p_ < 0.01). The lowest rating was given by the restricting-anorexic group (2.05), followed by the control group (2.65), the bulimic group (4.9), 158 and the bulimic-anorexic group (5.55). When Tukey comparisons of marginal means were calculated, only the contrast between restricting and bulimic anorexics reached significance (p. < 0.05). In summary, it appears that several of the questions were able to distinguish between the different groups. The effect of time was relatively less important, i.e. subjects did not markedly change their thoughts or feelings over time. 159 Appendix 10 Means of Heart Rate (bpm) (Standard Deviations 1n Parentheses) Group C* RA* BA* B* Margina Totals Rest 1 73.00 (7.06) 77.30 (24.03) 74.80 (13.97) 74.20 (16.36) 74.82 SI ides 75.56 (9.04) 79.15 (25.11) 76.69 (14.09) 75.48 (16.66) 76.72 Rest 2 72.10 (7.09) 76.30 (21.85) 75.30 (14.52) 74.90 (15.52) 74.65 Rest 3 72.80 (8.69) 76.70 (21.68) 74.00 (14.86) 73.00 (14.36) 74.12 Eat 1 79.70 (9.43) 85.60 (20.75) 83.20 (14.18) 79.50 (15.73) 82.00 Eat 2 81.90 (9.59) 87.80 (19.38) 86.00 (15.20) 82.30 (16.04) 84.50 Eat 3 81.80 (9.65) 87.50 (20.83) 87.30 (15.52) 84.30 (14.93) 85.22 Rest 4 75.00 (8.31) 81.20 (22.07) 75.70 (16.43) 74.50 (14.42) 76.60 Marginal 76.48 81.44 79.12 77.27 Totals C - Controls RA - Restricting-Anorexics BA - Bulimic-Anorexics B - Bulimics 160 Appendix 11 Means of Tonic Skin Conductance Levels (/junhos) (Standard Deviations in Parentheses^ Group C* RA* BA* B* Marginal Totals Rest 1 3.24 1.44 2.78 4.07 2.88 (1.42) (1.09) (1.40) (2.02) Slides 3.66 1.94 2.99 4.19 3.20 (1.85) (1.59) (1.30) (2.11) Rest 2 3.32 1.91 2.84 3.92 3.00 (1.58) (1.36) (1.30) (2.09) Rest 3 3.37 2.21 3.54 4.06 3.30 (1.50) (1.37) (1.60) (2.36) Eat 1 3.99 3.17 4.12 4.61 3.97 (2.32) (2.30) (1.79) (2.72) Eat 2 3.61 2.73 3.32 4.09 3.44 (2.11) (2.12) (1.64) (2.52) Eat 3 3.37 2.74 3.30 4.10 3.38 (2.04) (1.86) (1.58) (2.50) Rest 4 3.45 2.98 3.36 4.04 3.46 (2.02) (2.45) (1.86) (2.65) Marginal 3.50 2.39 3.28 4.14 Totals C - Controls RA - Restricting-Anorexics BA - Bulimic-Anorexics B - Bulimics 161 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items