Sage Journals: Discover world-class research

Abstract

Objective: The objective of this study was to valuate the effectiveness, tolerability and acceptability of various classes of antidepressants compared with placebo in the treatment of bulimia nervosa.

Method: A meta-analysis including 16 randomised controlled trials and 1300 bulimic patients was performed. Dichotomous outcomes were analysed by calculating relative risks, and continuous outcomes by calculating effect sizes. Methodological quality, heterogeneity in the results and selective publication were assessed.

Results: Short-term remission in bulimic symptoms was statistically more likely on antidepressants than placebo (Relative Risk = 0.88, 95%% Cl = 0.83–0.94, p < 0.0001). Drop-out rates were high but no statistical difference was found between treatment groups (34.6%% and 31.4%% for drug and placebo; RR = 1.03, 95%% CI = 0.80–1.32, p = 0.8). No difference in efficacy could be demonstrated among different classes of antidepressants.

Conclusions: The use of a single antidepressant agent was clinically effective for the treatment of bulimia nervosa when compared with placebo, with an overall greater remission rate and a higher rate of drop-outs. No differential effect regarding efficacy and tolerability among the various classes of antidepressants could be demonstrated.

Keywords

antidepressants bulimia nervosa meta-analysis systematic review

Bulimia Nervosa (BN) represents an important public health problem and is related to serious morbidity and even mortality. All types of anti-depressants seem to be beneficial to some degree in relieving bulimic symptoms in a large number of patients [1]. However, there is no clear evidence of a differential effect among the various antidepressants used. Moreover, it is not known if antidepressants have a specific anti-bulimic effect or if their success is a result of their antidepressant effects. This review attempts to systematically evaluate the use of anti-depressant medications compared with placebo for the treatment of BN, providing a statistical synthesis of the comparative data available to date.

Objectives

The aims were directed to the following questions: (i) are antidepressants clinically effective for the treatment of BN when compared with placebo? (ii) is there any differential effect for the various classes of antidepressants? and (iii) is the anti-bulimic effect of antidepressants independent of the effect on depressive symptoms?

Methods

Relevant trials for this review were identified by searching the following seven electronic databases, by means of the Cochrane Depression, Anxiety and Neurosis Group strategy for randomised controlled trials (RCTs): MEDLINE; EMBASE; LILACS; PsycLIT; the Cochrane Depression, Anxiety and Neurosis Group Database of Trials; the Cochrane Controlled Trials Register and the Science Citation Index (SCISEARCH). Reference lists of selected articles were inspected for further relevant studies. Pharmaceutical companies and first authors of all selected studies were contacted in order to obtain data on possible unpublished studies.

The review considered randomised, double-blind, placebo-controlled studies investigating any class of antidepressant medications in the treatment of patients with purging or non-purging type BN. Patients could have been diagnosed according to Russell, DSM-III, DSM-III-R, DSM-IV or ICD-10 criteria [2–6]. Parallel or crossover studies from which data for the first treatment period could be obtained and treated as a parallel trial were eligible. Data for at least one primary outcome of interest should have been reported or acquired.

As concealment of randomisation has been shown to affect trial outcomes [7],[8], two reviewers (JB and PH) independently graded the quality of allocation concealment according to the categories described in the Cochrane Collaboration Handbook (CCH) [7]. Trials with low or moderate risk of bias (categories A or B, respectively) were included. Trials with a high risk of bias due to inadequate allocation concealment were excluded. When no information on the randomisation process could be obtained from the authors, double-blind trials were included in category B. An inter-rater reliability study between reviewers was performed by means of the weighted kappa coefficient. The result indicated a very good agreement beyond chance (observed agreement of 91%%, K = 0.80, p < 0.0001). Studies comparing two active drugs or drugs with psychological approaches were not eligible for this review. Studies were also excluded if participants were patients with binge-eating/purging type anorexia nervosa or binge-eating disorder, as defined in DSM-IV [5], and if no information regarding the first treatment period of crossover studies could be obtained. In case of discrepancies, authors were contacted to have doubts resolved. Most excluded studies were reviews, subsets from other studies or did not provide any data for at least one primary outcome of interest.

The efficacy outcomes selected were remission, clinical improvement and changes in bulimic symptoms. Remission was defined as 100%% reduction in binge-eating episodes from baseline to endpoint, and clinical improvement as more than 50%% reduction. As we could not assume beforehand that all patients included in the studies would be purging-type bulimic individuals, binge-eating episodes were considered the main bulimic symptoms. A sensitivity analysis was performed to verify if purging and bingeing could be conflated.

Comorbid depression was evaluated through the mean difference in severity of depressive symptoms at endpoint.

Tolerability and acceptability of treatment were based on the number of patients dropping out during the study due to adverse events and for any cause.

Meta-analytic calculations were made by means of the statistical package ARCUS Quickstat for windows [9]. Remission, clinical improvement and drop-outs were analysed by calculating DerSimonian-Laird [10] estimate of relative risks (DL-RR) and approximate 95%% confidence interval. By convention, DL-RRs less than 1 indicated that an event was more likely to occur in the active treatment group. When overall results were significant, the number needed to harm (NNH) to produce or the number needed to treat (NNT) to prevent one outcome was calculated on the inverse of the absolute risk reduction [11].

Continuous outcomes (bulimic and depressive symptoms) were analysed by estimating effect size across all of the studies. The difference between the means of the antidepressant and placebo groups was divided by their pooled standard deviation (Hedges ‘g’ [12]). These g values were corrected for bias due to sample size, with the effect expressed as the unbiased estimator ‘d’. The pooled mean effect size was calculated using direct weights defined as the inverse of the variance of d for each study. According to Cohen [13], a large effect size is 0.8, a moderate effect size is 0.5 and a small effect size is 0.2.

Heterogeneity in the results of the trials was assessed by a test of heterogeneity using a Q (com-binability) statistic. A value of p < 0.05 indicated statistical heterogeneity.

The presence of selective publication was detected graphically in a funnel plot [14]. The number of studies (with null results) residing in the file drawers and needed to overturn the results of the combined significant test was assessed by means of the fail-safe N [12],[15].

Results

Sixteen studies fulfilled inclusion criteria and had data that could be used for at least one of the main comparisons (1300 patients in total). Six studies were considered relevant but were excluded from the meta-analysis. Characteristics of these 22 trials and reasons for exclusion from analysis are displayed in Table 1.

Table 1.

Characteristics of all eligible studies

Two studies were multi-centre, multinational [16],[17], both comparing fluoxetine to placebo. Ten trials were conducted in the USA [18–27], two in Canada [28],[29], one in the UK [30] and one in Norway [31]. Trials recruited outpatients seeking treatment ‘spontaneously’ from eating disorders clinics or programs, or included also volunteers for research recruited through advertisement.

The study population included in the trials presented reasonably comparable demographic and behavioural features. Only one study [23] included non-purging type bulimic subjects. Patients were mostly adult and young adult females, few adolescents being included. Underweight and overweight patients were constantly excluded.

All comparisons concerned one single antidepressant agent versus placebo. The only dose-response study conducted was the first multi-centre fluoxetine study [16]. Fluoxetine (60 mg/day) was clearly superior to both placebo and the usual 20 mg per day antidepressant dosages. The doses of other anti-depresssants have generally been similar to those employed in the treatment of depression.

Only two studies [17],[30] were graded as A according to the CCH criteria [7], thanks to information provided by the authors. All other 14 trials were graded as B. Blinding procedures were inadequately described in most trials. Some did not specify number or reasons for drop-outs, excluded patients after randomisation from analysis and did not report an intention-to-treat analysis. Some trials also omitted the standard deviations of continuous variables.

Efficacy

Remission of bulimic episodes (binge-eating or purge) was reported in eight studies (Table 1). Five reported remission of binge episodes [20],[22],[23],[26],[27], one of purge episodes [21] and two of both [17],[29]. In general, short-term remission of bulimic episodes was more likely on antidepressants than placebo (19.2%% vs 8.0%%). DerSimonian–Laird pooled estimates for each class of drugs are presented in Table 2. Only Monoamine oxidase inhibitors (MAOIs) (two small trials), and bupropion (one trial) showed significant differences over placebo. The risks of food and drug interactions with MAOIs and the increased risk of seizures in bulimic patients treated with bupropion contraindicate the use of these drugs in clinical practice.

Table 2.

Composite RR for efficacy, homogeneity tests and NNT

Pooled DL-RR for the eight studies reporting remission of bulimic episodes favoured drugs. The subgroup analysis confirmed that results were significant when bingeing and purging were analysed separately (DL-RR = 0.88, 95%% CI = 0.83–0.94 for bingeing and DL-RR = 0.90, 95%% CI = 0.81–0.99 for purging). According to the NNT, it would be necessary to treat 9 bulimic patients with an antidepressant to obtain one complete remission in a short-term, when compared with placebo, considering the 92%% non-remission rate of bulimic episodes in the placebo controls as a measure of baseline risk.

Clinical improvement was reported in eight trials (Table 1). Effects were stronger than those observed for remission (NNT = 4).

No published negative study was found; however, the search detected at least one unpublished multi-centre study performed in France and the UK [32], showing a lack of greater efficacy of fluvoxamine compared with placebo. Authors were contacted but did not provide their unpublished results. The number of trials necessary to overturn the results of the combined significance test (fail-safe N) was 18 for remission and 122 for clinical improvement.

The analysis of the mean difference in bulimic and depressive symptoms was based on the number of patients reporting these symptoms at the end of the trials. Five trials reported mean and standard deviations for bulimic symptoms at endpoint. DerSimonian-Laird pooled estimate of standardised effect size for difference in bulimic symptoms was −0.59 in favour of the active antidepressant drugs (95%% CI = − 0.87, − 0.31, p < 0.0001). According to Cohen's nomenclature this effect size is moderate [13].

Comorbidity (depressive symptoms)

Seven trials were included in the analysis. Standardised effect size was −0.23. No statistically significant difference was observed among classes of drugs. As individual patient data were not available, it was not possible to determine if the effects of anti-depressants on bulimic symptoms were independent of their effect on depressive symptoms.

Tolerability and acceptability

The pooled analysis of 12 trials reporting drop-outs for adverse events showed an overall rate of 10.5%% for antidepressants compared with 5.1%% for placebo (p = 0.01). No significant differences among and within classes of drugs were found (Table 3). Compared with placebo, MAOIs presented the lowest tolerability. The analysis of drop-outs due to any cause showed no statistically significant difference in acceptability of treatment between antidepressants and placebo, but overall short-term drop-out rates were high (34.6%% for drug and 31.4%% for placebo).

Table 3.

Composite RR for tolerability and acceptability of treatment, homogeneity tests and NNH

Tricyclic antidepressants (TCAs) and selective serotonin reuptake inhibitors (SSRIs) showed statistically significant results, but in opposite directions (Table 3). The pooled analysis of TCA studies showed that drop-outs were more likely in the active drug group (29%% vs 14.4%%), while in the case of SSRIs they were more likely for placebo (37.2%% for drug and 40.1%% for placebo).

Sensitivity analysis

Four pre-specified sensitivity analyses, subsidiary to the main review questions and based on non-randomised comparisons, were performed. They intended to confirm the hypotheses that response differed according to: (i) quality of the trials; (ii) class of drugs; (iii) length of treatment (< 8 weeks vs > 8 weeks); and (iv) diagnostic criteria (DSM-III vs DSM-III-R). Assessments were made by looking at separate subgroup of studies. None of the differences were statistically significant.

Discussion and conclusions

Bulimia nervosa is considered a complex and multi-factorial clinical condition, difficult to treat, with a chronic course [33]. Almost nothing is known about its natural course in the absence of treatment [34]. Investigations of the use of antidepressants for its treatment have been limited by several conceptual and methodological problems [35].

Indeed, most of the papers reviewed did not provide relevant information, such as details about adequacy of randomisation, blinding procedures and reasons of drop-outs. Crossover studies did not provide information for the first period, so carry-over or withdrawn effects were not taken into account. These methodological inconsistencies may have altered the perceived differences in efficacy between antidepressants and placebo and should be taken into account in the appraisal of results of the present review.

The most frequently reported outcome for efficacy was the mean per cent reduction in bulimic episodes, but measures of variance were missing in some trials. This outcome was not considered helpful to guide clinical decision-making as it does not translate benefits of treatment. A 70%% reduction in bulimic episodes in the experimental group, compared with a 50%% reduction in the placebo control group does not inform which is the probability of being free of bulimic symptoms or presenting a relevant clinical improvement when treated with a specific drug. Remission and clinical improvement were considered more appropriate outcomes for efficacy. Nevertheless, this information was not available for a considerable number of trials.

Concerning generalisability of the main findings, BN patients included in the trials seem similar to those seen in clinical settings in terms of age, duration of illness and severity of symptoms. However, most trials used stringent definition for bulimia or BN and usually did not include patients with comorbid severe depression, serious personality disorders or other ‘atypical’ conditions more difficult to treat (e.g. multi-impulsive patients or substance abusers). Therefore, results of this review are directed to bulimic patients without severe comorbidity.

This meta-analysis provides statistical evidence that a single antidepressant agent is clinically effective for the treatment of BN when compared with placebo, but the effect is modest. Although there were differences in terms of definition of illness according to different diagnostic criteria, length of treatment, quality of trials, and classes of drugs, a homogeneous therapeutic effect was found. No statistically significant differential effect regarding efficacy and tolerability among TCAs, SSRIs, MAOIs or other classes of antidepressants could be demonstrated. Remission rates were low and a considerable fraction of patients did not show a reduction of at least 50%% in bulimic symptoms in the short-term. Clinical improvement was consistent for all classes of antidepressants, though most of the patients would still fulfil criteria for BN at the end of the trials. It is proposed that single pharmacotherapy is less accepted for patients with BN than alternative single psychological therapies [36–38]. Medication studies without concurrent psychological approaches included in this review confirmed the high drop-out rates observed in most trials evaluating single pharmacological treatments, in part because of side-effects and in part because of some patients' negative attitudes toward medication use. The better acceptability of SSRIs may be related to their short-term effects on appetite and weight.

Few studies followed patients after the short-term double-blind treatment periods. Two trials reported a significant relapse rate (30–45%%) on improved patients followed for 4–6 months [20],[39]. We could find only one randomised, double-blind, placebo-controlled trial designed to specifically evaluate prevention of relapse with fluvoxamine [40]. Relapse rates during a 15-week maintenance period after successful psychotherapy treatment were significantly higher for placebo, but drop-out rates were very high (19 out of 33 patients randomised to fluvoxamine). The paucity of knowledge concerning the natural history of BN [34] makes it difficult to assess the long-term effects of treatments. Given the chronic, frequently relapsing course of treated patients, short-term results are of limited clinical value and long-term studies are warranted.

Antidepressants are effective for the treatment of minor depression [41] and also show a positive effect in the treatment of BN. Depressive symptoms are common in bulimic patients [42]. It is possible that antidepressants have both a direct effect over bulimic symptoms and an effect over depressive symptoms, reducing indirectly the disturbed eating behaviours. However, as individual patient data for concurrent depression was not systematically provided, it was not possible to conclude how depression affected short-term efficacy of antidepressants on bulimic symptoms.

It is important to further investigate and report whether TCAs and SSRIs present the same efficacy and frequency of drop-outs due to adverse events, describing these events and the respective remission rates. Therefore, it may be appropriate to directly compare these two classes of antidepressants regarding their tolerability and cost-effectiveness.

Future research should systematically include bulimic patients with comorbid major depression, personality disorders, substance abuse and other relevant clinical conditions. The impact of depression and other comorbid disorders on prognosis should be evaluated, improving generalisability of results. The option of trying a second antidepressant agent to reduce side-effects or increase efficacy should also be evaluated.

Footnotes

Acknowledgements

The authors wish to acknowledge Irismar R. Oliveira, Sergio L. Bartczak and Mauricio S. de Lima for advice and Roberta P. Trefiglio for assistance.

References

1. Wolfe

. Dimensions of response to antidepressant agents in bulimia nervosa: a review. Archives of Psychiatric Nursing 1995; 9: 111–121.

2. Russell

GFM

. Bulimia nervosa: an ominous variant of anorexia nervosa. Psychological Medicine 1979; 9: 429–448.

3. American Psychiatric Association . Diagnostic and statistical manual of mental disorders3rd edn. American Psychiatric Press, Washington 1980.

4. American Psychiatric Association . Diagnostic and statistical manual of mental disorders3rd edn. American Psychiatric Press, Washington 1987, Revised.

5. American Psychiatric Association . Diagnostic and statistical manual of mental disorders4th edn. American Psychiatric Press, Washington 1994.

6. World Health Organization . The ICD–10 classification of mental and behavioral disorders. Clinical descriptions and diagnostic guidelines. World Health Organization, Geneva 1992.

7. Cochrane collaboration handbook, Mulrow

Oxman

. BMJ Publishing, London 1996, Available in the Cochrane Library [[Database on Disk and CDROM]]. The Cochrane Collaboration. Issue 3. Oxford: Updated Software, 1996.

8. Schultz

Chalmers

Hayes

Altman

. Empirical evidence of bias: dimensions of methodological quality associated with estimates of treatment effects in controlled trials. Journal of the American Medical Association 1995; 273: 408–412.

9. Buchan

. Arcus Quickstat Biomedical for Windows. Research Solutions, Cambridge 1997.

10.

10. DerSimonian

Laird

. Meta–analysis in clinical trials. Controlled Clinical Trials 1986; 7: 177–188.

11.

11. Cook

Sachett

. The number needed to treat: a clinically useful measure of treatment effect. British Medical Journal 1995; 310: 452–454.

12.

12. Hedges

Olkin

. Statistical methods for meta–analysis. Academic Press, San Diego 1985.

13.

13. Cohen

. Statistical power for the behavioral sciences2nd edn. Lawrence Erlbaum, Hillsdale, NJ 1988.

14.

14. Light

Pillemer

. Summing up: the science of reviewing research. Harvard University Press, Cambridge, MA 1984.

15.

15. Rosenthal

. The –file drawer problem’ and the tolerance for null results. Psychological Bulletin 1979; 86: 638–641.

16.

16. Fluoxetine Bulimia Nervosa Collaborative Study Group ((FBNCSG)) . Fluoxetine in the treatment of bulimia nervosa. A multicenter, placebo–controlled, double–blind trial. Archives of General Psychiatry 1992; 49: 139–147.

17.

17. Goldstein

Wilson

Thomson

Potvin

Rampey

Jr . Long–term fluoxetine treatment of bulimia nervosa. British Journal of Psychiatry 1995; 166: 660–666.

18.

18. Mitchell

Pyle

Eckert

Hatukami

Pomeroy

Zimmerman

. A comparison study of antidepressants and structured intensive group psychotherapy in the treatment of bulimia nervosa. Archives of General Psychiatry 1990; 47: 149–157.

19.

19. Rothschild

Quitkin

Quiktin

. A double–blind placebo–controlled comparison of phenelzine and imipramine in the treatment of bulimia in atypical depressives. International Journal of Eating Disorders 1994; 15: 1–9.

20.

20. Walsh

Hadigan

Devlin

Gladis

Roose

. Long–term outcome of antidepressant treatment for bulimia nervosa. American Journal of Psychiatry 1991; 148: 1206–1212.

21.

21. Agras

Dorian

Kirkley

Arnow

Bachman

. Imipramine in the treatment of bulimia: a double–blind controlled study. International Journal of Eating Disorders 1987; 6: 29–38.

22.

Horne

Ferguson

Pope

. Treatment of bulimia with bupropion: a multicenter controlled trial. Journal of Clinical Psychiatry 1988; 49: 262–266.

23.

23. Mccann

Agras

. Successful treatment of nonpurging bulimia nervosa with desipramine: a double–blind, placebo–controlled study. American Journal of Psychiatry 1990; 147: 1509–1513.

24.

24. Mitchell

Groat

. A placebo–controlled double–blind trial of amitriptyline in bulimia. Journal of Clinical Psychopharmacology 1984; 4: 186–193.

25.

25. Pope

Jr Hudson

Jonas

Yurgelum–Todd

. Bulimia treated with imipramine: a placebo–controlled, double–blind study. American Journal of Psychiatry 1983; 140: 554–558.

26.

26. Pope

Keck

Mcelroy

Hudson

. A placebo–controlled study of trazodone in bulimia nervosa. Journal of Clinical Psychopharmacology 1989; 9: 254–259.

27.

27. Walsh

Gladis

Roose

Stewart

Glassman

. Phenelzine vs placebo in 50 patients with bulimia. Archives of General Psychiatry 1988; 45: 471–475.

28.

28. Kennedy

Piran

Warsh

. A trial of isocarboxazid in the treatment of bulimia nervosa. Journal of Clinical Psychopharmacology 1988; 8: 391–396.

29.

29. Kennedy

Goldbloom

Ralevski

Davis

D'souza

Lofchy

. Is there a role for selective monoamine oxidase inhibitor therapy in bulimia nervosa? A placebo–controlled trial of braforamine. Journal of Clinical Psychopharmacology 1993; 13: 415–422.

30.

30. Sabine

Yonace

Farrington

Barrant

Wakeling

. Bulimia nervosa: a placebo–controlled double–blind therapeutic trial of mianserin. British Journal of Clinical Pharmacology 1983; 15: 195–202.

31.

31. Kanerva

Rissanen

Sarna

. Fluoxetine in the treatment of anxiety, depressive symptoms, and eating–related symptoms in bulimia nervosa. Nordic Journal of Psychiatry 1994; 49: 237–242.

32.

32. Corcos

Flament

Atger

Jeammet

. Traitement pharmacologique de la bulimie. L'Encephale 1996; 22: 133–142.

33.

33. Foreyt

Poston

WSC

Goodrick

. Future directions in obesity and eating disorders. Addictive Behaviours 1996; 21: 767–778.

34.

34. Hsu

LKG

. Outcome of bulimia nervosa. Eating disorders and obesity. A comprehensive handbook, Browel

Fairburn

. Guilford, New York 1995.

35.

35. Shaw

Garfinkel

. Problems of research on eating disorders. International Journal of Eating Disorders 1990; 5: 545–555.

36.

36. Wilson

. Empirically validated treatments as a basis for clinical practice. Problems and prospects. Scientific standards of psychological practice: issues and recommendations, Hayes

Follette

Risley

Dawes

Grady

. Context, Reno 1995; 163–196.

37.

37. Bacaltchuk

Hay

Trefiglio

Mari

. Antidepressants versus psychotherapy for the treatment of bulimia nervosa. Paper presented at the Sixth International Cochrane Colloquium, Baltimore, Maryland, 1998.

38.

38. Bacaltchuk

Trefiglio

Oliveira

Lima

Mari

. Antidepressants versus psychotherapy for bulimia nervosa: a systematic review. Journal of Clinical Pharmacy and Therapeutics 1999; 24: 23–31.

39.

39. Pyle

Mitchell

Eckert

Hatsukami

Pomeroy

Zimmerman

. Maintenance treatment and 6–month outcome for bulimic patients who respond to initial treatment. American Journal of Psychiatry 1990; 147: 871–875.

40.

40. Fichter

Krüger

Rief

Holland

Döhne

. Fluvoxamine in prevention of relapse in bulimia nervosa: effects on eating specific psychopathology. Journal of Clinical Psychopharmacology 1996; 16: 9–18.

41.

41. Lima

Montcrieff

. A comparison of drugs versus placebo for the treatment of dysthymia: a systematic review ((Cochrane review)). The Cochrane Library. Update Software, Oxford 1998, Issue 2.

42.

42. Mitchell

Specker

de Zwaan

. Comorbidity and medical complications of bulimia nervosa. Journal of Clinical Psychiatry 1991; 52: 13–20.

43.

43. Hughes

Wells

Cunningham

Ilstrup

. Treating bulimia with desipramine. A double–blind, placebo–controlled study. Archives of General Psychiatry 1986; 43: 82–86.

44.

44. Barlow

Bloin

Blouin

Perez

. Treatment of bulimia with desipramine: a double–blind crossover study. Canadian Journal of Psychiatry 1988; 33: 129–133.

45.

45. Blouin

Blouin

Perez

Bushnik

Zuro

Mulder

. Treatment of bulimia with fenfluramine and desipramine. Journal of Clinical Psychopharmacology 1988; 8: 261–269.

46.

46. Alger

Shwalberg

Bigaouette

Michalek

Howard

. Effect of a tricyclic antidepressant and opiate antagonist on binge–eating behavior in normoweight bulimic and obese, binge–eating subjects. American Journal of Clinical Nutrition 1991; 53: 865–871.

47.

47. Marcus

Wing

Ewing

Kern

Mcdermott

Gooding

. A double–blind placebo–controlled trial of fluoxetine plus behavioral modification in the treatment of obese binge–eaters. American Journal of Psychiatry 1990; 147: 876–881.

48.

48. Scrimali

Grimali

Corriere

Grasso

Frisone

. La fluoxetina nel tratamento farmacologico della bulimia nervosa. Rivista Di Psichiatria 1994; 29: 171–177.