First step in managing bulimia nervosa: controlled trial of therapeutic manual

BMJ 1994; 308 doi: (Published 12 March 1994) Cite this as: BMJ 1994;308:686
  1. J Treasure,
  2. U Schmidt,
  3. N Troop,
  4. J Tiller,
  5. G Todd,
  6. M Keilen,
  7. E Dodge
  1. Eating Disorder Research Group, Institute of Psychiatry, London SE5 8AF Psychiatry Department, St Mary's Hospital, London W2 1NY University of Cologne, Cologne, Germany
  1. Correspondence to: Dr Treasure.
  • Accepted 12 November 1993


Objective: To test the short term efficacy of a self directed treatment manual for bulimia nervosa.

Design: Randomised controlled trial of the manual against cognitive behavioural therapy and a waiting list.

Setting: Tertiary referral centre.

Subjects: 81 consecutive referrals presenting with bulimia nervosa or atypical bulimia nervosa.

Main outcome measures: Frequency of binge eating, vomiting, and other behaviours to control weight as well as abstinence from these behaviours.

Results: Cognitive behavioural treatment produced a significant reduction in the frequency of binge eating, vomiting, and other behaviours to control weight. The manual significantly reduced frequency of binge eating and weight control behaviours other than vomiting, and there was no change in the group on the waiting list. Full remission was achieved in five (24%) of the group assigned to cognitive behavioural treatment, nine (22%) of the group who used the manual, and two (11%) of the group on the waiting list.

Conclusions: A self directed treatment manual may be a useful first intervention in the treatment of bulimia nervosa.

Clinical implications

  • Clinical implications

  • Bulimia nervosa affects between 4% and 8% of women at some point in their lives

  • Psychological therapy is the treatment of choice but is labour intensive

  • This study found that use of a self directed treatment manual produced a smaller symptomatic improvement than cognitive behavioural therapy but was considerably better than remaining on a waiting list

  • Complete remission from binge eating, vomiting, and other weight control behaviours was achieved in 22% of women who used the manual


Bulimia nervosa is increasing in prevalence among young women,1 and the lifetime prevalence ranges from 4% to 8% in women. Physical and psychological health is impaired,2 as well as social functioning.3 Whereas psychological treatment is generally the most successful for bulimia nervosa,4 it is often costly in terms of time and training of the therapist. Treatments with drugs can be administered more easily and cheaply, but their effects are less profound,6 are short lasting, and are associated with more side effects.7

A brief course of psychoeducation within a group with an emphasis on management of symptoms produced an abstinence from binge eating and vomiting in 20% of patients in the short term.8 It was less effective than full cognitive behavioural therapy at follow up but only in a more severely affected subgroup.9 Such group approaches, however, can pose organisational problems, and a lack of skilled staff may make it difficult to use in non-specialist centres.10

Books, computers, videotapes, and audiotapes, which require even less contact with a therapist than group treatment, have been shown to be effective in the treatment of depression, anxiety, phobias, and problem drinking and can considerably improve standard treatment by general practitioners of panic disorder and generalised anxiety disorder,*RF 11- 16* with therapeutic gains being maintained at two years' follow up.17 These approaches are cost effective and in line with the recommendations of the King's Fund Report that “the users should be empowered to master their own illness and problems.”18

We devised a problem oriented self help manual19 for the treatment of bulimia nervosa with the aim of improving the cost effectiveness of treatment, reducing the amount of contact with a therapist, and arming non-specialist centres with a usable and effective tool. The manual incorporates information and education with skills training including self monitoring, goal setting, assertiveness, cognitive restructuring, problem solving, and strategies to prevent relapse and was written to be easily comprehensible, according to the criteria of Albert and Chadwick.20

In an uncontrolled study we found significant reductions in bulimic symptoms and an increase in knowledge about the origins of the condition and the factors that maintain it.21

We present the results of a comparison of the manual with the first stage (sessions 1-8) of a course of 16 sessions of cognitive behavioural therapy and a control group from the hospital waiting list, by using a randomised controlled trial. This period was considered suitable for comparison as the first eight sessions of cognitive behavioural therapy are designed to produce symptomatic control.5 Our hypotheses were, firstly, that cognitive behavioural therapy would be considerably more effective than the manual and, secondly, that the manual would be considerably more effective than being on the waiting list at reducing bulimic symptoms.

Subjects and methods Subjects

One hundred and twenty five patients attending the Maudsley Hospital and fulfilling criteria for bulimia nervosa or atypical bulimia nervosa (International Classification of Diseases, tenth revision) were eligible for inclusion in the study. Seven were excluded because of severe comorbidity (diabetes mellitus, high risk of suicide, or dependence on alcohol), eight were unable to commit themselves to treatment because of travel difficulties or moving away, and 29 dropped out of treatment after starting: 14 were assigned to use the manual, seven were assigned to cognitive behavioural treatment, and eight were from the waiting list (the same ratio as were entered to each of these conditions). The remaining 81 patients (70 with bulimia nervosa and 11 with atypical bulimia nervosa) completed first and second assessments.

Seven of these patients were not binge eating at the first assessment but were included as bulimic symptoms wax and wane, and they were not considered to have recovered.


A clinician based assessment of bulimic symptoms was used to measure frequency of binging, vomiting, or other behaviours to control weight (such as the use of laxatives, diuretics, or diet pills; fasting; or excessive exercise), abnormal dietary pattern, degree of dietary restraint, excess concern with weight and shape, distress, and overall severity (see table III for the key for scoring severity). This interview was devised for a recent multicentre study of fluvoxamine in the treatment of bulimia (British Bulimia Group and Duphar Laboratories, personal communication) and was performed at first and second assessment, eight weeks apart. The structured clinical interview for DSM-III-R22 was used to measure additional psychiatric morbidity.

Measures of self assessment included the eating disorders inventory,23 the bulimic investigatory test, Edinburgh,24 the Beck depression inventory,25 and the self concept questionnaire,26 a measure of self esteem. As we aimed to show short term reduction in symptoms rather than psychological improvement, and in the interests of space, questionnaire scores are given only for comparison with other groups and change over eight weeks is not reported. Social class was assessed by father's occupation by using the Hollingshead two factor index.27


Ethics committee approval was obtained from the Bethlem and Maudsley Hospital. Eligible patients were assigned (by random numbers within envelopes included in the assessment pack) to one of three conditions: manual, cognitive behavioural therapy, and waiting list in the ratio of 2:1:1. Forty one were assigned to use the manual, 21 to cognitive behavioural therapy, and 19 to the waiting list. Figure 1 shows the full design of the study.


Design of study of self help manual v cognitive behavioural therapy v remaining on waiting list in treatment of patients with bulimia nervosa

Patients assigned to the manual condition were given the self treatment manual. They were told that it contained all the essential components of treatment that had been found to be effective in bulimia nervosa. They were encouraged to work through the book carefully, to put what they had learnt into practice, and to complete the exercises within the book. They were told that their progress would be reviewed at eight weeks, and they gave their informed consent to take part in the study.

Patients assigned to cognitive behavioural therapy were allocated to a therapist and immediately embarked on 16 sessions. Those on the waiting list were told that they would have to wait for a therapist to become available before starting therapy. All patients were reassessed eight weeks after their initial assessment by the same investigator, independently of the therapist.

Statistical analysis

Data were analysed with SPSS/PC+.28 Comparisons between groups were conducted by using a parametric analysis of variance for demographic and questionnaire data, a non-parametric Kruskal-Wallis analysis of variance, and confidence interval analysis for bulimic symptom scores (severity of symptoms was measured on non-linear scales).29 We used X2 tests for categorical data.


Sample characteristics

There were no significant differences between treatment groups on any demographic variables or questionnaire measures (table I and II) or on severity of bulimic symptoms (see table III, for scores) at P<=0.05. Patients who dropped out of treatment before the second assessment also did not differ significantly from patients who completed the first and second assessments.


Mean (SD) characteristics of sample of bulimic patients at first assessment according to assigned treatment

View this table:

Characteristics of sample of bulimic patients before treatment according to assigned treatment. Figures are numbers (percentages)

View this table:

Median and quartile symptom scores before and after treatment, median improvement, and approximate 95% confidence intervals in bulimic atients according to allocated treatment. Figures in parentheses are 25th and 75th percentile scores unless stated otherwise

View this table:

Effects of treatment

Within group differences - Patients using the manual showed significant improvement in symptoms (table III), and similar improvement was found in the group allocated to cognitive behavioural therapy. Those on the waiting list were unchanged.

Between group differences - With a 95% confidence interval analysis the subjects assigned to use the manual and those undergoing cognitive behavioural therapy were found to be significantly more improved than those on the waiting list. The dietary pattern, however, improved more in those undergoing cognitive behavioural therapy than those assigned to the manual and waiting list groups. Figure 2 shows the total change in symptoms, calculated by summing the frequency of binge eating, vomiting, and use of other weight control behaviours at the first and second assessments. Subjects assigned to the manual and cognitive behavioural therapy did significantly better than those on the waiting list but did not differ from each other. Surprisingly, no differences in the amount of improvement were found between the patients new to treatment and those who had had earlier treatment.


Abstinence rates

Table IV shows rates of abstinence from binge eating, vomiting, and other weight control behaviours as well as a global abstinence - that is, full remission - after eight weeks of treatment. Although differences in rates fail to reach significance, the manual seemed to be intermediate between cognitive behavioural therapy and no treatment (waiting list) in encouraging abstinence of binge eating and global abstinence.


Rates of abstinence from binge eating, vomiting, and other weight control behaviours at second assessment

View this table:


In accordance with our hypotheses, a treatment handbook can reduce bulimic symptoms and, in some, produce abstinence, although individual cognitive behavioural therapy seems to be more effective. Although both the manual and cognitive behavioural therapy reduced the frequency of binge eating, the use of weight control behaviours other than vomiting, concern with weight and shape, distress, clinician assessed overall severity, and a measure of global symptoms, therapy also reduced the frequency of vomiting, abnormal dietary pattern, and degree of dietary restraint.

The rates of abstinence produced by the manual are comparable with those produced in the educational group treatment of Olmstead et al9 and those of fluoxetine in a multicentre trial.6 The abstinence rate for binging in the cognitive behavioural treatment was slightly lower than those reported by others,4,30 while abstinence from vomiting was similar to some studies31 but less than others.4,30 It should be noted, however, that patients in our study are only halfway through treatment (eight sessions) to allow a direct comparison of the three treatment conditions. We thought that this would be an appropriate point of comparison as the main focus of the first eight weeks of cognitive behavioural therapy is reduction of symptoms. Fairburn and Cooper advise, “if by the end of eight weeks the patient's eating habits have not significantly improved, this treatment approach should be abandoned.”5 This criterion seems to be fulfilled by the manual.

Issues raised

Our study raises several issues. Firstly, data presented here are limited to short term effects. It will be important to consider rates of relapse and the need for subsequent treatment, such as consultation with a general practitioner or specialist treatment, or both. Further follow up is under way.

Secondly, a stepped care approach to the management of bulimia nervosa has been recommended32,33 and endorsed by the Royal College of Psychiatrists' report on eating disorders.34 Our study was conducted in a tertiary referral centre with many of the patients having had previous treatment. The manual might be more effective as a first step in the treatment of less severely ill bulimic patients presenting to primary care settings.

Thirdly, given the notorious ambivalence that patients with eating disorders have about treatment, eight weeks with a manual may have been too long a time to maintain motivation. Some patients reported feeling overwhelmed by being given so much information and advice at once and felt they did not know where to start. Others started to do some of the exercises but then gave up. Efficacy might be improved if motivation could be sustained, perhaps by dividing the manual into its individual chapters and sending them weekly or else by interspersing the manual with a few individual sessions. Treatment would then be given in small units, just as in cognitive behavioural therapy.


In conclusion, this study provides preliminary evidence that a self help manual may be a useful addition to the range of possible interventions in the treatment of bulimia nervosa, perhaps as the first phase of treatment or alongside existing treatments to reduce the amount of contact needed with a therapist, decreasing the dependence of the patient on the therapist, and freeing the therapist to work on more difficult, individual issues. It may also offer a means of help for sufferers who feel unable to overcome the barriers to seeking treatment that are often present.35

We thank the Mental Health Foundation for sponsoring this trial. NT is funded by the Medical Research Council. Royalties from the sale of the manual, Getting Better Bit(e) by Bit(e),19 are donated to the Mental Health Foundation and Eating Disorders Association.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
View Abstract