Skip to main content
NCBI home page
BMJ Clinical Evidence logoLink to BMJ Clinical Evidence
. 2009 Mar 10;2009:1011.

Anorexia nervosa

James Lock 1,#, Kathleen Kara Fitzpatrick 2,#
PMCID: PMC2907776  PMID: 19445758

Abstract

Introduction

Anorexia nervosa is characterised by a low body mass index (BMI), fear of gaining weight, denial of current low weight and its impact on health, and amenorrhoea. Estimated prevalence is highest in teenage girls, and up to 0.7% of this age group may be affected. While most people with anorexia nervosa recover completely or partially, about 5% die of the condition, and 20% develop a chronic eating disorder. Young women with anorexia nervosa are at increased risk of bone fractures later in life.

Methods and outcomes

We conducted a systematic review which aimed to answer the following clinical questions: What are the effects of treatments for anorexia nervosa? What are the effects of interventions to prevent or treat complications of anorexia nervosa? We searched: Medline, Embase, The Cochrane Library, and other important databases up to August 2007 (Clinical Evidence reviews are updated periodically, please check our website for the most up-to-date version of this review). We included harms alerts from relevant organisations such as the US Food and Drug Administration (FDA) and the UK Medicines and Healthcare products Regulatory Agency (MHRA).

Results

We found 40 systematic reviews, RCTs, or observational studies that met our inclusion criteria. We performed a GRADE evaluation of the quality of evidence for interventions.

Conclusions

In this systematic review we present information relating to the effectiveness and safety of the following interventions: anxiolytic drugs, cyproheptadine, inpatient/outpatient treatment setting, oestrogen treatment, psychotherapy, refeeding, selective serotonin reuptake inhibitors (SSRIs), and tricyclic antidepressants.

Key Points

Anorexia nervosa is characterised by a low BMI, fear of gaining weight, denial of current low weight and its impact on health, and amenorrhoea.

  • Estimated prevalence is highest in teenage girls, and may affect up to 0.7% of this group.

  • Anorexia nervosa is related to family, sociocultural, genetic, and other biological factors. Psychiatric and personality disorders such as depression, anxiety disorders, obsessive compulsive disorder, and perfectionism, are commonly found in people who have anorexia nervosa.

  • Most people with anorexia nervosa recover completely or partially, but about 5% die from the condition and 20% develop a chronic eating disorder.

  • Young women with anorexia nervosa are at increased risk of fractures later in life.

There is no strong research evidence that any treatments work well for anorexia nervosa. However, there is a gradual accumulation of evidence which suggests that early intervention is effective. Working with the family may also interrupt the development of a persistent form of the illness.

Evidence on the benefits of psychotherapy is unclear.

Refeeding is a necessary and effective component of treatment, but is not sufficient alone.

  • Very limited evidence from a quasi-experimental study suggests that a lenient approach to refeeding is as effective and more acceptable compared with a more strict approach.

  • Refeeding may be as effective in an outpatient setting as during hospital admission.

  • Nasogastric feeding is rarely required and can lead to problems due to hypophosphataemia.

  • Nutritional supplements, including zinc, have only limited evidence for their effectiveness, and additional evaluation of these measures are warranted.

Limited evidence from small RCTs has not shown significant weight gain from SSRIs or tricyclic antidepressants, some of which may cause serious adverse effects.

  • Tricyclic antidepressants may cause drowsiness, dry mouth, blurred vision, and a prolonged QT interval in people who have anorexia nervosa.

  • SSRIs have not been shown to be beneficial, but the evidence remains very limited; in the four RCTs we found, conclusions were limited due to small trial size and high withdrawal rates.

Anxiolytic drugs (mainly older generation antipsychotic drugs) may prolong the QT interval, increasing the risk of ventricular tachycardia, torsades de pointes, and sudden death.

  • Atypical antipsychotics have been evaluated for their potential role in reducing agitation and anxiety related to refeeding, as well as for potentially increasing appetite. Weak observational evidence has suggested that they may decrease obsessiveness and agitation. However, we found no RCTs of sufficient quality on the effects of atypical antipsychotics, and further evidence from large, well-conducted RCTs is necessary to draw reliable conclusions.

  • Some atypical antipsychotics do not appear to be associated with the same cardiac risks as older-generation antipsychotic drugs. However, further research needs to be done.

We found insufficient evidence assessing cyproheptadine for treating anorexia nervosa.

Oestrogen treatment has been hypothesized to reduce the negative effects on bone mineral density associated with anorexia nervosa. However, three small RCTs have failed to demonstrate significant changes in bone mineral density after treatment with oestrogen.

About this condition

Definition

Anorexia nervosa is characterised by a refusal to maintain weight at or above a minimally normal weight (less than 85% of expected weight for age and height, or BMI less than 17.5 kg/m2), or a failure to show the expected weight gain during growth. There is also often an intense fear of gaining weight, preoccupation with weight, denial of current low weight and its adverse impact on health, and amenorrhoea. Two subtypes of anorexia nervosa, binge-purge and restricting, have been defined.

Incidence/ Prevalence

One population-based study using consultation data from the General Practitioner Database in the UK found a mean incidence for anorexia nervosa of 4/100,000 in people aged 10-39 years.One systematic review (5 studies) assessing prevalence in European people aged over 19 years found a 12-month prevalence of 0.2-0.7%. Little is known of the incidence or prevalence in Asia, South America, or Africa. Population studies on the incidence of anorexia nervosa among adult ethnic minority populations in the USA have found a 12-month prevalence of 0.02% in Asian-Americans, 0.03% in Latinos, and 0.05% in African-American and Caribbean adults.

Aetiology/ Risk factors

Anorexia nervosa has been related to family, biological, social, and cultural factors. Studies have found that the condition is associated with a family history of anorexia nervosa (adjusted HR 11.4, 95% CI 1.1 to 89.0), bulimia nervosa (adjusted HR 3.5, 95% CI 1.1 to 14.0), depression, generalised anxiety disorder, obsessive compulsive disorder, or obsessive compulsive personality disorder (adjusted RR 3.6, 95% CI 1.6 to 8.0). A twin study suggested that anorexia nervosa may be related to genetic factors, but it was unable to estimate reliably how non-shared environmental factors contributed. Specific aspects of childhood temperament thought to be related include perfectionism, negative self-evaluation, and extreme compliance. Perinatal factors include prematurity, particularly if the baby was small for gestational age (prematurity: OR 3.2, 95% CI 1.6 to 6.2; prematurity and small for gestational age: OR 5.7, 95% CI 1.1 to 28.7). In a prospective cohort study (51 adolescents with the condition), people with anorexia nervosa were significantly more likely to have an affective disorder than were controls matched for sex, age, and school (lifetime risk of affective disorder 96% in people with anorexia nervosa v 23% in controls; ARI 73%, 95% CI 60% to 85%). It is unclear whether affective disorders precede anorexia nervosa or occur as a consequence of starvation. Similarly, obsessive compulsive disorder was significantly more likely to be present in people with anorexia nervosa compared with controls (30% v 10%; ARI 20%, 95% CI 10% to 41%). However, in two thirds of people with obsessive compulsive disorder and anorexia nervosa, obsessive compulsive disorder preceded the anorexia nervosa.

Prognosis

One prospective study followed up 51 people with teenage-onset anorexia nervosa, about half of whom received no or minimal treatment (fewer than 8 sessions). After 10 years, 14/51 people (27.5%) had a persistent eating disorder, three (5.9%) had ongoing anorexia nervosa, and six (11.8%) had experienced a period of bulimia nervosa. About half of all participants in the study continued to have poor psychosocial functioning after 10 years (assessed using the Morgan Russell scale and Global Assessment of Functioning Scale).An extended follow-up RCT of 38 participants, who completed either separate or conjoint family therapy, found that 75% of people had no eating disorder symptoms at 5-year follow-up. It found that those people who had good to intermediate outcomes at the end of treatment were more likely to have good outcomes at the end of the 5-year follow-up. A summary of treatment studies (119 studies published between 1953 and 1999, 5590 people, length of follow-up 1-29 years) found that 47% of people recover completely from anorexia nervosa (range 0-92%), 34% improve (range 0-75%), 21% develop a chronic eating disorder (range 0-79%), and 5% die from the condition (range 0-22%). Favourable prognostic factors include an early age at onset, and a short interval between onset of symptoms and the beginning of treatment.Family criticism, in particular maternal criticism, appears to influence the outcome of treatment. Unfavourable prognostic factors include vomiting, profound weight loss, chronicity, psychiatric comorbidity, psychosocial problems, and a history of premorbid developmental or clinical abnormalities.In particular, psychiatric comorbidities represent significant negative outcomes, with a recent review of outcomes indicating increased risk for personality disorders (in particular avoidant, dependent, obsessive-compulsive and passive-aggressive personality disorders), obsessive compulsive disorder, and depression.The all-cause standardised mortality ratio of eating disorders (anorexia nervosa and bulimia nervosa) has been estimated at 538, which is about three times higher than other psychiatric illnesses. In studies published between 1970 and 1996, the average annual mortality was 0.59% a year for females in 10 eating-disorder populations (1322 people), with a minimum follow-up of 6 years. Mortality was higher for people with lower weight and older age at presentation. Mortality by suicide represents a significant threat to people who have anorexia nervosa and is reported as the second most common cause of death in this population. A review of studies published on suicide and suicide attempts in people who have eating disorders found that suicide rates were markedly elevated compared with the general population: between 3-20% of the inpatient and outpatient groups assessed had attempted suicide at some point.The elevated rates of suicide attempts and death by suicide are present for those receiving both inpatient and outpatient treatment, and appear to exist independant of BMI at the time of death. Assessment of suicide risk remains a critical feature in the evaluation and treatment of people who have anorexia nervosa, particularly among those with comorbid psychiatric illnesses.Young women who have anorexia nervosa are at an increased risk of fractures later in life.

Aims of intervention

To restore physical health (weight within the normal range and no sequelae of starvation [e.g. regular menstruation, normal bone mass]), normal patterns of eating and attitudes towards weight and shape; to reduce any additional psychiatric comorbidities (e.g. depression, anxiety, obsessive compulsive disorder); to reduce the impact of extreme personality traits such as high sensitivity to threat, rigidity and detail rather than global information processing style; and to reduce the impact of the illness on social functioning and quality of life.

Outcomes

The most widely used measure of outcome in people who have anorexia nervosa is the Morgan Russell scale, which considers nutritional status, menstrual function, mental state, and sexual and social adjustment. Biological outcome criteria alone, such as weight (BMI or in relation to matched population weight) and menstrual function, are used infrequently as outcome measures. RCTs do not usually have long enough follow-up periods to examine mortality. Other validated outcome measures include eating-symptom measures. Bone mineral density is included as a proxy outcome for fracture risk.

Methods

Clinical Evidence search and appraisal August 2007. The following databases were used to identify studies for this systematic review: Medline 1966 to August 2007, Embase 1980 to August 2007, and The Cochrane Library (all databases) 2007, Issue 3. Additional searches were carried out using these websites: NHS Centre for Reviews and Dissemination (all databases), Turning Research into Practice (TRIP), and NICE. Abstracts of the studies retrieved from the initial search were assessed by an information specialist. Selected studies were then sent to the author for additional assessment, using pre-determined criteria to identify relevant studies. Study design criteria for assessment in this review were: published systematic reviews and RCTs in any language, with any level of blinding (including ‘open’ studies) and containing at least 30 individuals (with any loss to follow-up allowed). There was no minimum length of follow-up required to include studies. We also searched for observational studies of refeeding. In addition, we use a regular surveillance protocol to capture harms alerts from organisations such as the FDA and the Medicines and UK Healthcare products Regulatory Agency (MHRA), which are added to the reviews as required. We searched for RCTs comparing each listed intervention with placebo, no treatment, usual care, or any other listed intervention, and included all RCTs of sufficient quality. We have reported various outcome measures used in included studies (see Outcomes section). However, none of these outcome defintions have systematic support for their utility in predicting long-term remission or recovery. The lack of a clear consensus in the field on defining remission and recovery limits the interpretation of all RCTs concerned with anorexia nervosa. We have performed a GRADE evaluation of the quality of evidence for interventions included in this review, see table . To aid readability of the numerical data in our reviews, we round many percentages to the nearest whole number. Readers should be aware of this when relating percentages to summary statistics such as RRs and ORs.

Table.

GRADE evaluation of interventions for anorexia nervosa

Important outcomes Symptom improvement, bone mineral density
Number of studies (participants) Outcome Comparison Type of evidence Quality Consistency Directness Effect size GRADE Comment
What are the effects of treatments in anorexia nervosa?
1 (54) Symptom improvement Zinc v placebo 4 –3 0 0 0 Very low Quality points deducted for sparse data, incomplete reporting of results, and poor follow-up
2 (153) Symptom improvement Cyproheptadine v placebo 4 –3 0 0 0 Very low Quality points deducted for sparse data, incomplete reporting of results, unclear withdrawal rate, and lack of clarity about whether psychotherapy was also given
1 (90) Symptom improvement Inpatient v outpatient treatment setting 4 –3 0 –1 0 Very low Quality points deducted for sparse data, weak methods (lack of blinding, randomisation, treatment adherence), and incomplete reporting of results. Directness point deducted for no direct statistical comparison between groups
4 (182) Symptom improvement Psychotherapy v treatment as usual v dietary counselling 4 –2 –1 –1 0 Very low Quality points deducted for sparse data, and weak methods. Consistency point deducted for different results for different outcomes. Directness point deducted for no direct comparison between groups in one RCT
1 (93) Symptom improvement Psychotherapy v fluoxetine or psychotherapy plus fluoxetine 4 –3 0 0 0 Very low Quality points deducted for sparse data, incomplete reporting of results, and weak methods (poor adherence, withdrawls)
8 (371) Symptom improvement Different types of psychotherapy v each other 4 –3 0 0 0 Very low Quality points deducted for weak methods, incomplete reporting of results, and high rate of withdrawls
3 (124) Symptom improvement Fluoxetine or citalopramv placebo 4 –3 0 0 0 Very low Quality points deducted for sparse data, incomplete reporting of results, and high rate of withdrawls
2 (115) Symptom improvement Tricyclic antidepressants v placebo 4 –2 0 –1 0 Very low Quality points deducted for sparse data, and poor/unclear follow-up. Directness point deducted for no statistical comparison between groups in 1 RCT
What are the effects of interventions to prevent or treat complications of anorexia nervosa?
3 (210) Bone mineral density Oestrogen treatment v control 4 –2 0 –1 0 Very low Quality points deducted for high rate of withdrawls, and incomplete reporting of results. Directness point deducted for use of a surrogate outcome measure
Open in a new tab

Type of evidence: 4 = RCT. Consistency: similarity of results across studies Directness: generalisability of population or outcomes Effect size: based on relative risk or odds ratio

Glossary

Beck Depression Inventory

Standardised scale to assess depression. This instrument consists of 21 items to assess the intensity of depression. Each item is a list of four statements (rated 0, 1, 2, or 3), arranged in increasing severity, about a particular symptom of depression. The range of scores possible are 0 = least severe depression to 63 = most severe depression. It is recommended for people aged 13–80 years. Scores of more than 12 or 13 indicate the presence of depression.

Body mass index

Weight (kg) divided by height (m) squared.

Dietary counselling

Dieticians with experience of eating disorders discuss diet, mood, and daily behaviours.

Inpatient treatment

This has been regarded as the standard approach to the management of anorexia nervosa. One of the key components of inpatient treatment is refeeding, which is achieved through structured, supervised meals. Psychotherapy (of a variety of different types) and pharmacotherapy are included in many programmes.

Morgan Russell scale

A widely used measure of outcome for anorexia nervosa consisting of two scores: an average outcome score and a general outcome score (possible total of 12). The average outcome score is based on the outcome in five areas: nutritional status, menstrual function, mental state, sexual adjustment, and socioeconomic status.

Psychotherapy

Different types of psychological treatments given individually, in groups, or within the family are included here. These use psychodynamic, cognitive behavioural, or supportive techniques, or combinations of these. Family therapy includes members of the family of origin or the constituted family, and addresses the eating disorder as a problem of family life. Non-Specific Supportive Clinical Management (NSSCM) is currently known as Specialist Supportive Clinical Management (SSCM). It is a form of supportive treatment which uses some motivational elements to increase engagement. The focus is to return to normal weight and eating in the usual environment. It consists of three phases: orientation, agree target symptoms and goals; monitoring and support to achieve goals; and work on ending therapy and relationship.

Very low-quality evidence

Any estimate of effect is very uncertain.

Disclaimer

The information contained in this publication is intended for medical professionals. Categories presented in Clinical Evidence indicate a judgement about the strength of the evidence available to our contributors prior to publication and the relevant importance of benefit and harms. We rely on our contributors to confirm the accuracy of the information presented and to adhere to describe accepted practices. Readers should be aware that professionals in the field may have different opinions. Because of this and regular advances in medical research we strongly recommend that readers' independently verify specified treatments and drugs including manufacturers' guidance. Also, the categories do not indicate whether a particular treatment is generally appropriate or whether it is suitable for a particular individual. Ultimately it is the readers' responsibility to make their own professional judgements, so to appropriately advise and treat their patients.To the fullest extent permitted by law, BMJ Publishing Group Limited and its editors are not responsible for any losses, injury or damage caused to any person or property (including under contract, by negligence, products liability or otherwise) whether they be direct or indirect, special, incidental or consequential, resulting from the application of the information in this publication.

Contributor Information

James Lock, Stanford University, Stanford, USA.

Kathleen Kara Fitzpatrick, Stanford University, Stanford, USA.

References

  • 1.American Psychiatric Association. Diagnostic and statistical manual of mental disorders (DSM-IV), 4th ed. Washington, DC: American Psychiatric Association, 1994. [Google Scholar]
  • 2.Currin L, Schmidt U, Treasure J, et al. Time trends in eating disorder incidence. Br J Psychiatry 2005;186:132–135. [DOI] [PubMed] [Google Scholar]
  • 3.Wittchen HU,.Jacobi F. Size and burden of mental disorders in Europe — a critical review and appraisal of 27 studies. Eur Neuropsychopharmacol 2005;15:357–376. [DOI] [PubMed] [Google Scholar]
  • 4.Nicado EG, Hong S, Takeuchi DT. Prevalence and correlates of eating disorders among Asian Americans: Results from the National Latino and Asian American Study. Int J Eating Disord 2007;40:S22–S26. [DOI] [PubMed] [Google Scholar]
  • 5.Alegria M, Woo M, Cao Z, et al. Prevalence and correlates of eating disorders in Latinos in the United States. Int J Eating Disord 2007;40:S15–S21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Taylor JY, Howard Caldwell C, Baser RE, et al. Prevalence of eating disorders among Blacks in the National Survey of American Life. Int J Eating Disord 2007;40:S10–S14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Strober M, Freeman R, Lampert C, et al. Controlled family study of anorexia nervosa and bulimia nervosa: evidence of shared liability and transmission of partial syndromes. Am J Psychiatry 2000;157:393–401. [DOI] [PubMed] [Google Scholar]
  • 8.Lilenfeld LR, Kaye WH, Greeno CG, et al. A controlled family study of anorexia nervosa and bulimia nervosa: psychiatric disorders in first-degree relatives and effects of proband comorbidity. Arch Gen Psychiatry 1998;55:603–610. [DOI] [PubMed] [Google Scholar]
  • 9.Wade TD, Bulik CM, Neale M, et al. Anorexia nervosa and major depression: shared genetic and environmental risk factors. Am J Psychiatry 2000;157:469–471. [DOI] [PubMed] [Google Scholar]
  • 10.Fairburn CG, Cooper Z, Doll HA, et al. Risk factors for anorexia nervosa: three integrated case-control comparisons. Arch Gen Psychiatry 1999;56:468–476. [DOI] [PubMed] [Google Scholar]
  • 11.Cnattingius S, Hultman CM, Dahl M, et al. Very preterm birth, birth trauma, and the risk of anorexia nervosa among girls. Arch Gen Psychiatry 1999;56:634–638. [DOI] [PubMed] [Google Scholar]
  • 12.Wentz E, Gillberg C, Gillberg IC, et al. Ten-year follow-up of adolescent-onset anorexia nervosa: psychiatric disorders and overall functioning scales. J Child Psychol Psychiatry 2001;42:613–622. [PubMed] [Google Scholar]
  • 13.Eisler I, Simic M, Russell GF, et al. A randomised controlled treatment trial of two forms of family therapy in adolescent anorexia nervosa: a five-year follow-up. J Child Psychology Psychiatry Allied Disciplines 2007;48:552–560. [DOI] [PubMed] [Google Scholar]
  • 14.Steinhausen HC. The outcome of anorexia nervosa in the 20th century. J Psychiatry 2002;159:1284–1293. [DOI] [PubMed] [Google Scholar]
  • 15.Lock J, Couturier J, Bryson S, et al. Predictors of dropout and remission in family therapy for adolescent anorexia nervosa in a randomized clinical trial. Int J Eating Disord 2006;39:639–647. [DOI] [PubMed] [Google Scholar]
  • 16.Berkman ND, Lohr KN, Bulik CM. Outcomes of eating disorders: a systematic review of the literature. Int J Eating Disord 2007;40:293–309. [DOI] [PubMed] [Google Scholar]
  • 17.Harri, EC, Barraclough B. Excess mortality of mental disorder. Br J Psychiatry 1998;173:11–53. [DOI] [PubMed] [Google Scholar]
  • 18.Nielsen S, Møller-Madsen S, Isager T, et al. Standardized mortality in eating disorders: a quantitative summary of previously published and new evidence. J Psychosom Res 1998;44:413–434. [DOI] [PubMed] [Google Scholar]
  • 19.Franko DebraL, Keel PK. Suicidality in eating disorders: Occurrence, correlates, and clinical implications. Clin Psychol Rev 2006;26:769–782. [DOI] [PubMed] [Google Scholar]
  • 20.Holm-Denoma JM, Witte TK, Gordon KH, et al. Deaths by suicide among individuals with anorexia as arbiters between competing explanations of the anorexia-suicide link. J Affect Disord 2008;107:231–236. [DOI] [PubMed] [Google Scholar]
  • 21.Lucas A, Melton L, Crowson C, et al. Long term fracture risk among women with anorexia nervosa: a population-based cohort study. Mayo Clin Proc 1999;74:972–977. [DOI] [PubMed] [Google Scholar]
  • 22.Morgan HG, Russell GF. Value of family background and clinical features as predictors of long-term outcome in anorexia nervosa: four-year follow-up study of 41 patients. Psychol Med 1975;5:355–371. [DOI] [PubMed] [Google Scholar]
  • 23.Cooper Z, Fairburn CG. The Eating Disorders Examination. A semi-structured interview for the assessment of the specific psychopathology of eating disorders. Int J Eat Disord 1987;6:1–8. [Google Scholar]
  • 24.Garner DM. Eating Disorder Inventory-2 (EDI-2): professional manual. Odessa, FL: Psychological Assessment Resources Inc, 1991. [Google Scholar]
  • 25.Garner DM, Garfinkel PE. The eating attitudes test: an index of the symptoms of anorexia nervosa. Psychol Med 1979;9:273–279. [DOI] [PubMed] [Google Scholar]
  • 26.Henderson M, Freeman CPL. A self-rating scale for bulimia: the ‘BITE'. Br J Psychiatry 1987;150:18–24. [DOI] [PubMed] [Google Scholar]
  • 27.Couturier J, Lock J. What is remission in adolescent anorexia nervosa? A review of various conceptualizations and quantitative analysis. Int J Eat Disord 2006;39:175–183. [DOI] [PubMed] [Google Scholar]
  • 28.Touyz SW, Beumont PJ, Glaun D, et al. A comparison of lenient and strict operant conditioning programmes in refeeding patients with anorexia nervosa. Br.J Psychiatry 1984;144:517–520. [DOI] [PubMed] [Google Scholar]
  • 29.Touyz SW, Beaumont PJ, Dunn SM. Behaviour therapy in the management of patients with anorexia nervosa. A lenient, flexible approach. Psychother Psychosom 1987;48:151–156. [DOI] [PubMed] [Google Scholar]
  • 30.Robb AS, Silber TJ, Orrell-Valente JK, et al. Supplemental nocturnal nasogastric refeeding for better short-term outcome in hospitalized adolescent girls with anorexia nervosa. Am J Psychiatry 2002;159:1347–1353. [DOI] [PubMed] [Google Scholar]
  • 31.Berkman ND, Bulik CM, Brownley KA, et al. Management of eating disorders. Evid Rep Technol Assess (Full Rep) 2006;135:1–166. [PMC free article] [PubMed] [Google Scholar]
  • 32.Bulik CM, Berkman ND, Brownley KA, et al. Anorexia nervosa treatment: a systematic review of randomized controlled trials. Int J Eat Disord 2007;40:310–320. [DOI] [PubMed] [Google Scholar]
  • 33.Birmingham CL, Goldner EM, Bakan R. Controlled trial of zinc supplementation in anorexia nervosa. Int J Eat Disord 1994;15:251–255. [PubMed] [Google Scholar]
  • 34.Nakahara T, Nagai N, Tanaka M, et al. The effects of bone therapy on tibial bone loss in young women with anorexia nervosa. Int J Eat Disord 2006;39:20–26. [DOI] [PubMed] [Google Scholar]
  • 35.Gowers S, Jackson A, Richardson K, et al. CR130. Guidelines for the nutritional management of anorexia nervosa. London: Royal College of Psychiatrists, 2005. http://www.rcpsych.ac.uk/publications/collegereports/cr/cr130.aspx (last accessed 20 January 2009). [Google Scholar]
  • 36.Goldberg SC, Halmi KA, Eckert RC, et al. Cyproheptadine in anorexia nervosa. Br J Psychiatry 1979;134:67–70. [DOI] [PubMed] [Google Scholar]
  • 37.Halmi KA, Eckert E, LaDu TJ, et al. Anorexia nervosa. Treatment efficacy of cyproheptadine and amitriptyline. Arch Gen Psychiatry 1986;43:177–181. [DOI] [PubMed] [Google Scholar]
  • 38.West Midlands Development and Evaluation Service. In-patient versus out-patient care for eating disorders. DPHE 1999 Report No 17. Birmingham: University of Birmingham, 1999. Search date 1999; primary sources Medline, Psychlit, The Cochrane Library, variety of internet sites, and hand searches of relevant editions of relevant journals and references from identified articles. [Google Scholar]
  • 39.Crisp AH, Norton K, Gowers S, et al. A controlled study of the effect of therapies aimed at adolescent and family psychopathology in anorexia nervosa. Br J Psychiatry 1991;159:325–333. [DOI] [PubMed] [Google Scholar]
  • 40.Gowers S, Norton K, Halek C, et al. Outcome of outpatient psychotherapy in a random allocation treatment study of anorexia nervosa. Int J Eat Disord 1994;15:165–177. [DOI] [PubMed] [Google Scholar]
  • 41.Kåchele H for the study group MZ-ESS. Eine multizentrische studie zu aufwand und erfolg bei psychodynamischer therapie von eβstörungen. Psychother Med Psychol (Stuttg) 1999;49:100–108. [PubMed] [Google Scholar]
  • 42.le Grange D, Lock J. The dearth of psychological treatment studies for anorexia nervosa. Int J Eat Disord 2005;37:79–91. [DOI] [PubMed] [Google Scholar]
  • 43.Hay P, Bacaltchuk J, Claudino A, et al. Individual psychotherapy in the outpatient treatment of adults with anorexia nervosa. In: The Cochrane Library, Issue 3, 2007. Chichester, UK: John Wiley & Sons, Ltd. Search date 2005; primary sources Medline, Extramed, Embase, Psychlit, Current Contents, hand searches of Int J Eat Disord, and correspondence with researchers in the field. [Google Scholar]
  • 44.Dare C, Eisler I, Russell G, et al. Psychological therapies for adult patients with anorexia nervosa: a randomised controlled trial of outpatient treatments. Br J Psychiatry 2001;178:216–221. [DOI] [PubMed] [Google Scholar]
  • 45.Hall A, Crisp AH. Brief psychotherapy in the treatment of anorexia nervosa. Outcome at one year. Br J Psychiatry 1987;151:185–191. [DOI] [PubMed] [Google Scholar]
  • 46.Serfaty MA. Cognitive therapy versus dietary counselling in the outpatient treatment of anorexia nervosa: effects of the treatment phase. Eur Eat Disord Rev 1999;7:334–350. [Google Scholar]
  • 47.Pike KM, Walsh BT, Vitousek K, et al. Cognitive behavioural therapy in the post-hospital treatment of anorexia nervosa. Am J Psychiatry 2003;160:2046–2049. [DOI] [PubMed] [Google Scholar]
  • 48.Halmi KA, Agras WS, Crow S, et al. Predictors of treatment acceptance and completion in anorexia nervosa: implications for future study designs. Arch Gen Psychiatry 2005;62:776–781. [DOI] [PubMed] [Google Scholar]
  • 49.Walsh BT, Kaplan AS, Attia E, et al. Fluoxetine after weight restoration in anorexia nervosa: a randomized controlled trial. JAMA 2006;295:2605–2612. [DOI] [PubMed] [Google Scholar]
  • 50.Eisler I, Dare C, Hodes M, et al. Family therapy for adolescent anorexia nervosa: the results of a controlled comparison of two family interventions. J Child Psychol Psychiatry 2000;41:727–736. [PubMed] [Google Scholar]
  • 51.Robin AL, Siegel PT, Moye AW, et al. A controlled comparison of family versus individual therapy for adolescents with anorexia nervosa. J Am Acad Child Adolesc Psychiatry 1999;38:1482–1489. [DOI] [PubMed] [Google Scholar]
  • 52.Treasure JL, Todd G, Brolly M, et al. A pilot study of a randomized trial of cognitive analytical therapy vs educational behavioral therapy for adult anorexia nervosa. Behav Res Ther 1995;33:363–367. [DOI] [PubMed] [Google Scholar]
  • 53.Russell GFM, Szmukler G, Dare C, et al. An evaluation of family therapy in anorexia nervosa and bulimia nervosa. Arch Gen Psychiatry 1987;44:1047–1056. [DOI] [PubMed] [Google Scholar]
  • 54.Lock J, Agras WS, Bryson S, Kraemer HC. A comparison of short- and long-term family therapy for adolescent anorexia nervosa. J Am Acad Child Adolesc Psychiatry 2005;44:632–639. [DOI] [PubMed] [Google Scholar]
  • 55.McIntosh VV, Jordan J, Carter FA, et al. Three psychotherapies for anorexia nervosa: a randomized, controlled trial. Am J Psychiatry 2005;162:741–747. [DOI] [PubMed] [Google Scholar]
  • 56.Eisler I, Dare C, Russell GFM, et al. Family and individual therapy in anorexia nervosa. A 5-year follow-up. Arch Gen Psychiatry 1997;54:1025–1030. [DOI] [PubMed] [Google Scholar]
  • 57.Wallin U, Kronvall P, Majewski ML. Body awareness therapy in teenage anorexia nervosa: outcome after 2 years. Eur Eat Disord Rev 2000;8:19–30. [Google Scholar]
  • 58.Bell LE-M, Bell L:. What can we learn from consumer studies and qualitative research in the treatment of eating disorders? Eat Weight Disord 2003;8:181–187. [DOI] [PubMed] [Google Scholar]
  • 59.Claudino AM, Hay P, Lima MS, et al. Antidepressants for anorexia nervosa. In: The Cochrane Library, Issue 3, 2007. Chichester, UK: John Wiley & Sons, Ltd. Search date 2005. 16437485 [Google Scholar]
  • 60.Attia E, Haiman C, Walsh BT, et al. Does fluoxetine augment the inpatient treatment of anorexia nervosa? Am J Psychiatry 1998;155:548–551. [DOI] [PubMed] [Google Scholar]
  • 61.Kaye WH, Nagata T, Weltzin TE, et al. Double-blind placebo-controlled administration of fluoxetine in restricting- and restricting–purging-type anorexia nervosa. Biol Psychiatry 2001;49:644–652. [DOI] [PubMed] [Google Scholar]
  • 62.Fassino S, Leombruni P, Daga G, et al. Efficacy of citalopram in anorexia nervosa: a pilot study. Eur Neuropsychopharmacol 2002;12:453–459. [DOI] [PubMed] [Google Scholar]
  • 63.Ackerman MJ. The long QT syndrome: ion channel diseases of the heart. Mayo Clin Proc 1998;73:250–269. [DOI] [PubMed] [Google Scholar]
  • 64.Becker A, Grinspoon SK, Klibanski A, et al. Current concepts: eating disorders. N Engl J Med 1999;340:1092–1098. [DOI] [PubMed] [Google Scholar]
  • 65.Yap Y, Camm J. Risk of torsades de pointes with non-cardiac drugs: doctors need to be aware that many drugs can cause QT prolongation. BMJ 2000;320:1158–1159. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66.Sheridan DJ. Drug-induced proarrhythmic effects: assessment of changes in QT interval. Br J Clin Pharmacol 2000;50:297–302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Reilly JG, Ayis SA, Ferrier IN, et al. QTc interval abnormalities and psychotropic drug therapy in psychiatric patients. Lancet 2000;355:1048–1052. [DOI] [PubMed] [Google Scholar]
  • 68.Brambilla F, Garcia CS, Fassino S, et al. Olanzapine therapy in anorexia nervosa: psychobiological effects. Int Clin Psychopharmacol 2007;22:197–204. [DOI] [PubMed] [Google Scholar]
  • 69.Dunican KC, DelDotto D. The role of olanzapine in the treatment of anorexia nervosa. Ann Pharmacother 2007;41:111–115. [DOI] [PubMed] [Google Scholar]
  • 70.Mondraty N, Birmingham CL, Touyz S, et al. Randomized controlled trial of olanzapine in the treatment of cognitions in anorexia nervosa. Australasian Psychiatry 2005;13:72–75. [DOI] [PubMed] [Google Scholar]
  • 71.Newman-Toker J. Risperidone in anorexia nervosa. J Am Acad Child Adolesc Psychiatry 2000;39:941–942. [DOI] [PubMed] [Google Scholar]
  • 72.Powers PS, Bannon Y, Eubanks R, et al. Quetiapine in anorexia nervosa patients: An open label outpatient pilot study. Int J Eat Disord 2007;40:21–26. [DOI] [PubMed] [Google Scholar]
  • 73.Bosanac P, Kurlender S, Norman T, et al. An open-label study of quetiapine in anorexia nervosa. Human Psychopharmacol 2007;22:223–230. [DOI] [PubMed] [Google Scholar]
  • 74.Biederman J, Herzog DB, Rivinus TM, et al. Amitriptyline in the treatment of anorexia nervosa: a double-blind, placebo-controlled study. J Clin Psychopharmacol 1985;5:10–16. [PubMed] [Google Scholar]
  • 75.Klibanski A, Biller BMK, Schoenfeld DA, et al. The effects of estrogen administration on trabecular bone loss in young women with anorexia nervosa. J Clin Endocrinol Metab 1995;80:898–904. [DOI] [PubMed] [Google Scholar]
  • 76.Grinspoon S, Thomas L, Miller K, et al. Effects of recombinant human IGF-I and oral contraceptive administration on bone density in anorexia nervosa. J Clin Endocrinol Metab 2002;87:2883–2891. [DOI] [PubMed] [Google Scholar]
  • 77.Strokosch GR, Friedman AJ, Wu SC, et al. Effects of an oral contraceptive (norgestimate/ethinyl estradiol) on bone mineral density in adolescent females with anorexia nervosa: a double-blind, placebo-controlled study. J Adolesc Health 2006;39:819–827 [DOI] [PubMed] [Google Scholar]
  • 78.American Psychiatric Association. Practice guideline for the treatment of patients with eating disorders (revision). Am J Psychiatry 2000;157(suppl 1):1–39. [PubMed] [Google Scholar]
BMJ Clin Evid. 2009 Mar 10;2009:1011.

Refeeding

Summary

SYMPTOM IMPROVEMENT Zinc compared with placebo: Zinc may be more effective at increasing the daily rate of weight gain in women with anorexia nervosa who have normal zinc levels before treatment ( very low-quality evidence ). NOTE We found no RCTs of refeeding in people who have anorexia nervosa; an RCT is unlikley to be conducted. A number of complications can occur during refeeding, in particular with nasogastric or intravenous methods. The so-called “refeeding syndrome” includes a range of electrolyte disturbances or gastrointestinal problems, such as acute gastric dilatation, or both. The most common, rapidly occurring problem is hypophosphataemia, which can cause cardiac and respiratory failure, delirium, and fits.

Benefits

Different approaches to encourage refeeding:

We found no RCTs of refeeding in people who have anorexia nervosa; an RCT is unlikely to be conducted. We found one quasi-experimental study, which found that a lenient, as opposed to strict, behavioural approach was more acceptable to staff and patients, and equally effective in encouraging refeeding. However, this study (65 consecutive inpatient admissions due to anorexia) was not randomised, and the finding that this approach was more acceptable was by consensus and not tested statistically, so any conclusions drawn should be made with extreme caution.

Nasogastric feeding:

We found one controlled study (100 girls with anorexia nervosa, mean age 15.5 years) comparing nasogastric versus oral refeeding, which found that nasogastric feeding increased weight gain and time taken to regain weight compared with oral refeeding.However, the study was based on a retrospective chart review, and participants were analysed according to the treatment that they had already received, so any conclusions drawn should be made with extreme caution.

Total parenteral nutrition:

We found no observational studies of sufficient quality.

Zinc versus placebo:

We found two systematic reviews (search date 2005), which found one RCT.The identified RCT (54 women aged over 15 years, mean BMI 15.8 kg/m2, mean duration of anorexia nervosa 3.7 years, admitted to 2 eating disorder units) compared 100 mg zinc gluconate versus placebo. All but three of the women had normal zinc levels before treatment. Treatment was continued until each woman had gained 10% of her weight on admission on two consecutive weeks. A total of 10 people in the zinc group and nine in the placebo group did not complete the study (35%). The RCT found limited evidence from an analysis of people completeting the study that zinc significantly increased the daily rate of gain in BMI compared with placebo (0.079 kg/m2 with zinc v 0.039 kg/m2 with placebo; P = 0.03).

Calcium and associated supplements:

We found no RCTs or observational studies of sufficient quality (see comment below).

Harms

A number of complications can occur during refeeding, particularly with nasogastric or intravenous methods. The so-called “refeeding syndrome” includes a range of electrolyte disturbances or gastrointestinal problems, such as acute gastric dilatation, or both. The most common, rapidly occurring problem is hypophosphataemia, which can cause cardiac and respiratory failure, delirium, and fits.

Different approaches to encourage refeeding:

The study did not report on harms.

Nasogastric feeding:

The retrospective chart review reported adverse effects only for the nocturnal nasogastric refeeding group. Of this group, 2/52 (4%) people required an anti-anxiety drug to relieve anxiety about tube placement, 3/52 (6%) removed their own tube, 6/52 (11%) had epistaxis, 15/52 (29%) had nasal irritation, and there were no reported cases of refeeding syndrome or aspiration pneumonia.

Total parenteral nutrition:

We found no studies.

Zinc versus placebo:

The RCT found that none of the people taking zinc had adverse effects.

Calcium and associated supplements:

We found no RCTs.

Comment

In this option, we have excluded evidence on any form of forced refeeding. We found one RCT comparing calcium plus vitamin D and etidronate with placebo.However, for the comparison of interest (calcium plus vitamin D versus placebo), it only included 27 people. This is below Clinical Evidence reporting criteria for this review of at least 30 people, so we have not reported it further. People who have anorexia nervosa are at increased risk of bone fractures in later life, owing to decreased bone density. Restoration of weight alone does not appear to reverse bone loss, particularly when not accompanied by a return of menses. This has led to the exploration of supplement use to improve bone density, focusing on calcium and vitamin D, but also on biphosphates associated with bone growth.

Clinical guide:

Refeeding is a necessary component of treatment but is not sufficient alone. The aim is for the person to increase the amount and variety of food consumed and to reintroduce meals as a source of shared, positive emotions. This can be achieved in the outpatient setting if there is a good alliance, especially in adolescent cases where parents can have support in order to provide the requisite care (the so-called Maudsley Model). However, a proportion of severe cases require more intensive care, such as that provided in inpatient settings. Guidelines on the nutritional management of anorexia nervosa have been produced. In some centres, the calorie deficit is made up with normal, albeit larger, meals or snacks, plus vitamin and mineral supplements. In other centres, liquid foods can be used to supplement or replace some or all of the meals. All of the studies on nutritional approaches are quasi-experimental and underpowered. Rarely is it necessary to resort to nasogastric feeding or percutaneous endoscopic gastrostomy. The legal and ethical considerations of these procedures and assisted feeding by a skilled nurse for severe treatment-resistant anorexia nervosa need to be addressed. To avoid hypophosphataemia, it is advisable to cover an increase in calorie intake with a general vitamin and mineral supplement and to ensure that there is adequate phosphate. The calorie content of the diet should start at approximately 50 kcal/kg and gradually increase.The use of supplements appears to vary widely and clinical consensus on the use of vitamin and mineral supplementation to assist refeeding and manage the sequelae of malnourishment has not yet been established due to limited evidence.

Substantive changes

Refeeding Two systematic reviews added, which both identified the same RCT of zinc previously reported in this review. No new data added. Categorisation unchanged (Likely to be beneficial).

BMJ Clin Evid. 2009 Mar 10;2009:1011.

Cyproheptadine

Summary

SYMPTOM IMPROVEMENT Compared with placebo: We don't know whether cyproheptadine is more effective at improving weight gain in women with anorexia nervosa who are in specialised inpatient units ( very low-quality evidence ).

Benefits

Cyproheptadine versus placebo:

We found no systematic review but found two small RCTs. The first RCT (81 women in 3 specialised inpatient units, no withdrawals reported) compared cyproheptadine versus placebo, and behavioural therapy versus no behavioural therapy. The RCT found no significant difference in weight gain over 10 weeks between cyproheptadine and placebo (mean 5.11 kg with cyproheptadine v 4.32 kg with placebo; reported as non-significant, CI not reported). The effect of behavioural therapy was not reported. The second RCT (72 women, mean age 20.6 years, mean 77% of target weight, mean duration of anorexia nervosa 2.9 years, at 2 specialised inpatient units) compared three interventions: cyproheptadine (up to a maximum of 32 mg), amitriptyline, and placebo for 6 weeks. It found that cyproheptadine significantly increased the rate of weight gain compared with placebo (mean days to target weight: 36 with cyproheptadine v 45 with placebo; P less than 0.05). Women were withdrawn from the trial if they failed to gain at least 2 kg after 6 weeks; 10 women were withdrawn for this reason but the total withdrawal rate is unclear. The authors adjusted for withdrawals by assigning a standard time to attainment of target weight of 120 days to all women who withdrew. It is not clear if people also received psychotherapy.

Harms

Cyproheptadine versus placebo:

The first RCT gave no information on adverse effects. In the second RCT, on both day 7 and day 21, people taking cyproheptadine had fewer physical adverse events rated moderate or severe compared with people taking placebo or amitriptyline. There were no withdrawals from the trial because of adverse effects.

Comment

Both RCTs were of short duration.

Substantive changes

No new evidence

BMJ Clin Evid. 2009 Mar 10;2009:1011.

Inpatient versus outpatient treatment setting

Summary

SYMPTOM IMPROVEMENT Inpatient compared with outpatient treatment setting: We don't know whether inpatient treatment, outpatient treatment (individual and group psychotherapy), outpatient group therapy, and assessment interview only, differ in their effectiveness at improving mean weight gain or symptoms (measured by Morgan Russell scale global scores) at 1, 2, and 5 years, in women with anorexia nervosa who do not require an emergency intervention ( very low-quality evidence ).

Benefits

Inpatient versus outpatient treatment setting:

We found three systematic reviews (search date 1999;search date 2005), which included RCTs comparing inpatient treatment versus outpatient care. All three reviews identified the same single RCT, which had a 5-year follow-up.The RCT (90 women referred with anorexia nervosa, mean age 22 years, weight loss 26% of matched population mean weight, mean duration of anorexia nervosa 3.2 years, and not requiring emergency intervention) compared four interventions: inpatient treatment, outpatient treatment (individual and group psychotherapy), outpatient group therapy, and assessment interview only. Assessors were not blind to treatment allocation. Difficulties with acceptance of randomised treatment allocation led to more people being assigned to inpatient treatment than to the other interventions, which may have compromised the randomisation of the trial. Adherence to allocated treatment (defined as accepting allocation and at least 1 attendance at a group or individual treatment session) was significantly higher with outpatient than with inpatient treatment (adherence rates: inpatient treatment 18/30 [60%], outpatient treatment [individual and family therapy] 18/20 [90%], outpatient treatment [group psychotherapy] 17/20 [85%], and assessment interview only 20/20 [100%]; RR for failure to adhere with inpatient v outpatient treatment 1.5, 95% CI 1.1 to 2.0). Average continuation with treatment also varied among groups (mean attendance: 20 weeks' inpatient treatment, 9 outpatient sessions, and 5 group sessions). In the assessment interview only group, six people had no treatment of any kind in the first year and the others had treatment elsewhere (6 had inpatient treatment, 5 had outpatient hospital treatment, and 3 had at least weekly contact with their general practitioners). Six people in this group spent almost the entire year in treatment. The RCT found similar mean weight gain and Morgan Russell scale global scores in all four groups at 1, 2, and 5 years; the proportion of people with a good outcome after inpatient treatment was 5/29 (17%) at 2 years and 9/27 (33%) at 5 years; with outpatient treatment (individual and family therapy) this was 4/20 (20%) at 2 years and 8/17 (47%) at 5 years; with the outpatient group psychotherapy it was 5/19 (26%) at 2 years and 10/19 (53%) at 5 years; and with assessment interview only it was 2/20 (10%) at 2 years and 6/19 (32%) at 5 years. The RCT did not assess the significance of the difference among groups. One systematic review also identified numerous case series, from which it was unable to draw meaningful conclusions because participant characteristics, treatments, mortality, and outcomes varied widely. For example, people admitted for inpatient treatment in the case series had a lower mean weight than those treated as outpatients.

Harms

Inpatient versus outpatient treatment setting:

One person died from anorexia nervosa between the assessment and the start of outpatient group treatment, and one of the people allocated to inpatient treatment died from anorexia nervosa within 5 years.

Comment

Clinical guide:

One observational study (355 people with anorexia nervosa, 169 of whom had bulimic-type anorexia nervosa; mean age 25 years; mean duration of illness 5.7 years; 75% available for 2.5-year follow-up) found that people with a longer duration of illness had a better chance of good outcome with longer duration of inpatient treatment. People with a shorter duration of illness had a higher likelihood of good outcome with briefer inpatient treatment. The median duration of inpatient treatment was 11.6 weeks for anorexia nervosa and 10.6 weeks for bulimic-type anorexia nervosa.

Substantive changes

Inpatient versus outpatient treatment setting Two systematic reviews added which both identified the same RCT previously reported in this review. No new data added. Categorisation unchanged (Unknown effectiveness).

BMJ Clin Evid. 2009 Mar 10;2009:1011.

Psychotherapy

Summary

SYMPTOM IMPROVEMENT Psychotherapy compared with treatment as usual or dietary counselling: We don't know whether psychotherapy is more effective at improving outcomes (including "significant improvement" or "good outcome" on Morgan Russell scale score, mean body weight, "full recovery" or resumption of menses) in women with either late-age onset anorexia nervosa and/or a long duration of illness ( very low-quality evidence ). Psychotherapy compared with fluoxetine or combined psychotherapy plus fluoxetine: We don't know whether CBT plus placebo is more effective than CBT plus fluoxetine at improving symptoms or increasing time to relapse in women with anorexia nervosa who had previously successfully completed treatment in an inpatient or day programme setting (very low-quality evidence). Different types of psychotherapy versus each other: We don't know whether any one type of psychotherapy is more effective at improving outcomes in women with anorexia nervosa (very low-quality evidence).

Benefits

Psychotherapy versus treatment as usual or dietary counselling:

We found four systematic reviews (search date 2003;search date 2005), which between them identified six, mainly small, RCTs (263 people). The reviews did not conduct a meta-analysis, and only four of the identified RCTs met our inclusion criteria. The RCTs were of limited quality, and compared different types of psychotherapy versus treatment as usual or versus dietary counselling (see table 1 ). All RCTs were carried out in an outpatient setting in people with either late age-onset anorexia nervosa or long duration of illness, or both. The first RCT found significant improvements in weight gain and for the proportion of people classified as recovered with focal analytical therapy and family treatment compared with treatment as usual.Treatment as usual involved a 30-minute session (weekly or every 2 weeks) with a trainee psychiatrist (supervised by an eating-disorder expert) explaining the consequences of anorexia nervosa, encouraging a regular, sustainable healthy diet, and monitoring weight and physical status at each session. The second RCT found no significant difference in outcomes between 12 sessions of psychotherapy and 12 sessions of dietary counselling. The third RCT comparing CBT versus dietary counselling found a significant improvement from baseline for CBT but did not report a between-group analysis. All people treated with dietary counselling either did not take up or withdrew from treatment and refused release of their results, making it impossible to compare the two groups. The fourth RCT found that significantly more women receiving CBT than receiving dietary advice had a "good outcome" on the Morgan Russell scale after 1 year, but found no significant difference between groups in women making a "full recovery" (see table 1 ; see comment below).

Table 1.

Effects of psychotherapies for anorexia nervosa (only studies with at least 30 participants included).

Ref People Participant characteristics Intervention Outcome Comment
Psychotherapy versus treatment as usual or dietary counselling
  84 People with BMI less than 12 kg/m2 excluded. Mean age 26.3 years; 2 males; mean BMI 15.4 kg/m2; mean duration of AN 6.3 years FAP or CAT or FT or TAU for 1 year Number of people recovered or "significantly improved" at 1 year on Morgan Russell scale: FAP 7/21 (33%), FT 8/22 (36%), CAT 6/22 (27%), TAU 1/19 (5%). FAP and FT were significantly better than TAU. CAT was not significantly better than TAU. Weight gain at 1 year: FAP v TAU, P = 0.03; FT v TAU, P = 0.05; absolute values not reported 4/84 (5%) failed to take up treatment. 54/84 (64%) completed full course of treatment (FAP, 12; FT, 16; CAT, 13; TAU, 13). 12/84 (14%) people required admission during the course of treatment (FAP, 2; FT, 3; CAT, 2; TAU, 5) with no differences between groups. One person died in the TAU group. TAU involved a 30-minute general training session on consequences of anorexia and supportive encouragement with trainee psychiatrist
  30 Mean age 19.5 years; mean 27% below predicted weight; mean duration of AN 2.25 years 12 sessions of PT or DA No significant difference in outcome at 1 year between groups in mean body weight, Morgan Russell scale, or resumption of menses (mean weight: 45.1 kg with PT v 46.0 kg with DA; mean global Morgan Russell score 8.8 with PT; reported as non-significant for all outcomes, P value not reported) No one failed to take up treatment. 14/15 (93%) in PT group completed treatment (1 finished early with good outcome), and 11/15 (73%) in DA group completed treatment. 3/30 (10%) were admitted. As soon as treatment had ended, 1 person in PT was admitted because of very low weight; in the DA group, there were two admissions, and four people attended 68 outpatient sessions
  35 Mean age 20 years; 2 males; mean BMI 16.5 kg/m2 ; mean duration of AN or atypical AN 4.2 years 20 sessions of CT or DA at a ratio of 3:1 No between-group comparison possible; see comment 0/25 (0%) in cognitive group therapy and 3/10 (30%) in dietary counselling group failed to take up treatment. At 6 months, 23/25 (92%) allocated to CT were still engaged in treatment, whereas all 10 of the people allocated to DA had withdrawn and refused further contact. DA involved encouragement towards healthy diet plus regular monitoring of weight plus physical status
  33 Mean age 25 years; mean BMI 15.5 kg/m2 ; mean duration of AN 7 years 50 sessions of CBT or DA over 1 year Significantly more women receiving CBT than receiving DA had a “good outcome” on Morgan Russell scale at 1 year (8/18 [44%] with CBT v 1/15 [7%] with DA; P less than 0.02). More women receiving CBT than receiving DA had a “full recovery” but difference between groups not significant (3/18 [17%] with CBT v 0/15 [0%] with DA; P less than 0.10) “Full recovery” defined as “good outcome” on the Morgan Russell scale plus comparable eating attitudes to those of women with AN and no binging or purging
Psychotherapy versus fluoxetine or versus combined psychotherapy plus fluoxetine
  122 Mean age 25 years (89% over 18 years) with BMI 17.8 kg/m2; all had fulfilled DSM AN criteria within the past year, and 33% had received prior inpatient treatment. At entry to trial, all participants were within 75% of a target weight established by standardised growth charts 37 sessions of CBT v drug treatment (fluoxetine 60 mg) v CBT plus drug treatment. All participants had 18 15-minute sessions of medical management Medical management involved questioning the participant about adverse effects of drugs, difficulties with compliance, and their general health, as well as providing information about their treatment RCT not able to report the planned primary outcomes because withdrawal from treatment was high Of the 122 patients who received randomised treatments, 77 (63%) did not complete the trial; 33 withdrew from treatment in the first 5 weeks, and an additional 45 withdrew later 21/122 (17%) of the withdrawals from the study were due to treatment failure (weight below 70% of their target weight) More people receiving CBT or combined treatment than receiving drug treatment alone completed treatment (18/42 [43%] with CBT v 16/39 [41%] with combination treatment v 11/41 [27%] with drug treatment alone) The RCT gave no information on adverse effects owing to high withdrawal rates
93 Mean age 23 years; all had fulfilled DSM AN criteria within the past year, recruited from inpatient treatment. At entry to RCT, all participants had a BMI of at least 19.0 CBT plus placebo v fluoxetine plus CBT; all participants had 52 sessions of CBT. The RCT reported no significant differences between groups in BMI or psychological variables at time of termination (reported as no significant difference, P values not reported). No significant difference between groups in time to relapse (HR 1.12, 95% CI 0.65 to 2.01; P = 0.64) Withdrawal from treatment was high. Of 93 people randomised, 53 (57%) stopped treatment early ( 57% with fluoxetine plus CBT v 57% with CBT plus placebo; P = 0.98). The study evaluated time to relapse as well as feasibility of medication treatment. This study suggests that fluoxetine did not increase time to relapse, although difficulty in adherence and study completion make outcome difficult to evaluate
Different types of psychotherapy versus each other
  40 Mean age 15.5 years; 1 male; mean weight 74% of predicted; mean duration of AN 1.1 years CFT or SFT for 12 months No significant difference in weight gain (mean 6.4 kg gained with CFT v 9.8 kg with SFT; P = 0.09) or Morgan Russell rating scale (average 3.1 improvement with CFT v 2.5 with SFT; P = 0.23) between groups after 1 years' treatment Three of the initial eligible families refused to participate, four withdrew within the first 3 months, 29 completed 12 months' treatment. Four people were admitted to hospital (1 SFT, 3 CFT) for 2.0–5.5 months. A mean of 12 sessions of treatment was given. 4/40 (10%) people admitted for inpatient treatment
  41 Mean age 14.1 years; mean BMI 15.9 kg/m2; duration of AN less than 1 year BFST or EOIT (with collateral parental sessions) for 12–18 months No significant differences in weight gain, proportion menstruating, eating attitudes, depression, or eating-related family conflict between groups at 2.5 years' follow-up (proportion at target weight: 80% with BFST v 69% with EOIT; proportion menstruating: 93% with BFST v 80% with EOIT; absolute results not reported for other outcomes; difference between groups reported as non-significant for all outcomes; P values not reported) 7/41 (17%) refused to participate in 2.5-year follow-up. Number of cases admitted for inpatient treatment 16/41 (39%)
  33 Mean age 14 years; mean weight 73% of expected; mean duration 1.1 year BAT plus 25 sessions of FT or 25 sessions of FT alone Similar global clinical score at 2 year follow-up (8.5 with BAT plus FT v 8.9 with FT alone; P value not reported) 7 families refused to participate. Number of treatment sessions taken varied between 8 and 157. Number of cases admitted for inpatient treatment 8/33 (24%)
  57 People were stratified into subgroups based on clinical parameters Subgroup 1 (AN with age at onset less than 18 years and duration less than 3 years): mean age 16.6 years; mean weight at admission 66% of ABW; mean duration 1.2 years Subgroup 2 (age at onset 18 years or below and duration greater than 3 years): mean age 20.6 years; mean weight at admission 66% of ABW; mean duration 5.9 years Subgroup 3 (age at onset greater than 19 years): mean age 27.7 years; mean weight at admission 63% of ABW; mean duration 3 years FT or IT for 1 year after inpatient weight restoration In subgroup I, FT superior to IT on Morgan and Russell general outcome scales at 1 year (good outcome: FT 6/10 [60%] v IT 1/11 [9%]; RR 7.2, 95% CI 1.03 to 50.3) and at 5 years (FT 9/10 [90%] v IT 4/11 [36%]; RR 2.25, 95% CI 1.02 to 4.94). No significant differences in other subgroups Withdrawals: subgroup 1, 1/10 (10%) FT v 7/11 (64%) IT; subgroup 2, 2/10 (20%) FT v 3/9 (33%) IT; subgroup 3, 3/7 (43%) FT v 0/7 (0%) IT Subgroups were defined before randomisation. Subgroup 1 benefited more from FT than from IT for relapse prevention. There was no information about the number of people assessed who refused inpatient treatment
  84 People with BMI less than 12 kg/m2 excluded. Mean age 26.3 years; 2 males; mean BMI 15.4 kg/m2; mean duration of AN 6.3 years FAP, CAT, FT, or TAU for 1 year Number of people recovered or "significantly improved" at 1 year on Morgan Russell scale: FAP 7/21 (33%), FT 8/22 (36%), CAT 6/22 (27%); no significant difference among groups (reported as non-significant, P value not reported) 4/84 (5%) failed to take up treatment. 54/84 (64%) completed full course of treatment (FAP, 12; FT, 16; CAT, 13; TAU, 13). 12/84 (14%) of people required admission during the course of treatment (FAP, 2; FT, 3; CAT, 2; TAU, 5) with no differences between groups
  30 Mean age 25.1 years; 1 male; mean BMI 15.2 kg/m2; mean duration of AN 4.7 years. People with severe weight loss were excluded and offered inpatient treatment 20 sessions of EBT or CAT No significant differences between groups in terms of weight gain at 1 year, or on the Morgan Russell scale (mean weight gain: 6.7 kg with EBT v 6.9 kg with CAT; average Morgan Russell score: 6.4 with EBT v 7.3 with CAT; reported as non-significant; P values not reported). No one was admitted to hospital 1/31 (3%) refused randomisation. 10/30 (33%) withdrew from treatment. 10 people in each group completed treatment. No one was admitted to hospital
  86 Age 12–18 years; mean age 15 years; mean duration of AN 12 months. BMI 17 kg/m2 at randomisation and 16 kg/m2 when first identified; 30% of the participants were hospitalised prior to randomisation with mean BMI of 16 kg/m2All participants offered inpatient treatment as needed and all received approximately 35 short (10 to 15-minute) check-ups with a paediatrician during the year Short term FT (10 sessions over 6 months) v long term (20 sessions over 12 months) FT in both groups based upon the Maudsley model No significant difference in BMI or weight at 1 year between short- and long-term treatment. BMI and weight same in both groups: BMI: 19.0 kg/m2 at 6 months and 19.5 kg/m2 at 12 months; weight: 51 kg at 6 months and 53 kg at 12 months (reported as non-significant, P value not reported) 22% of participants were admitted to hospital during the intervention 10% did not complete 80% of the sessions
  56 Age 17–40; mean BMI 17.1 kg/m2, mean duration of AN less than 5 years, including people with EDNOS-AN subtype (i.e. people did not fulfil amenorrhoea criteria or have a BMI less than 17.5 kg/m2) People with severe illness excluded IPT v CBT v NCSSM for 20 weekly sessions following a treatment manual More people having NSSCM than having IPT had a good outcome at 1 year, and a similar number of people having NSSCM and CBT had a good outcome at 20 weeks (proportion of people with score of 1 or 2 on a non-validated scale from 1–4: 9/16 [56%] with NSSCM v 2/21 [10%] with IPT v 6/19 [32%] with CBT; reported as significant/non-significant for each comparison; P values for paired comparisons not reported) Similar BMI and weight at 1 year (BMI: 18.8 kg/m2 with NSSCM v 18.1 kg/m2 with IPT v 18.1 kg/m2 with CBT; weight: 50.4 kg with NSSCM v 49.0 kg with IPT v 48.6 kg with CBT) No significant difference among groups in the proportion of people who completed treatment (69% with NSSCM v 57%, with IPT v 63% with CBT; reported as non-significant; P value not reported). 4/56 [7%] people were hospitalised during treatment (1 having 1 NSSCM, 3 having IPT). One person died
Open in a new tab

ABW, average body weight; AN, anorexia nervosa; BAT, body-awareness therapy; BFST, behavioural family systems therapy; BMI, body mass index; CAT, cognitive analytical therapy; CBT, cognitive behavioural therapy; CFT, conjoint family therapy; CT, cognitive therapy; DA, dietary advice; DSM, Diagnostic and Statistical Manual of Mental Disorders; EBT, educational behavioural therapy; EDNOS, Eating disorders not otherwise specified; EOIT, ego-orientated individual therapy; FAP, focal analytical therapy; FT, family therapy; IPT, interpersonal therapy; IT, individual therapy; NSSCM, non-specific supportive clinical management, now known as Specialist Supportive Clinical Care (SSCM); PT, psychotherapy; SFT, separated family therapy; TAU, treatment as usual.

Psychotherapy versus fluoxetine or versus combined psychotherapy plus fluoxetine:

We found two RCTs. The first RCT (122 people) compared CBT alone, fluoxetine alone, and combined treatment. The RCT was unable to assess clinical outcomes owing to high withdrawal rates, particularly in people receiving drug treatment alone. It therefore assessed the acceptability of treatment (see table 1 ). The second RCT (93 women) compared fluoxetine versus placebo in women also receiving individual CBT after weight restoration (see table 1 ). The women in the RCT had previously successfully completed treatment in an inpatient or day programme setting. Outcome was assessed by looking at time to relapse and the proportion of participants completing 1 year of treatment. The RCT found that fluoxetine plus CBT did not improve time to relapse following weight restoration compared with placebo plus CBT (see table 1 ). As with other studies in adults who have anorexia nervosa, attrition rates were high (53/93 [57%] of those randomised terminated treatment prematurely), which makes drawing definitive conclusions problematic. The authors concluded that the evidence suggests the need to explore other types of treatment both to achieve and maintain weight restoration.

Different types of psychotherapy versus each other:

The reviews identified 11 RCTs (published in 10 papers) comparing different types of psychotherapy. Only seven of the identified RCTs met our inclusion criteria. We also found one additional RCT. Five of these RCTs were undertaken in an outpatient setting in people with an early age (mean less than 18 years) of onset and short illness duration (mean 1 year or more). Two of the RCTs were carried out in an outpatient setting in people with a later age (mean more than 18 years) of onset and/or a longer duration of illness (mean between 2 and 6 years). One RCT, published in two papers, included people with early- and late-onset anorexia nervosa and with long and short duration of illness. None of the RCTs found an overall significant difference between different psychotherapies (see table 1 ).

Harms

The acceptability of the treatment varied among RCTs. Failure to take up treatment ranged from 0–30%, and withdrawal from treatment ranged from 0–70% among RCTs, but this may have been caused by different methods of case ascertainment (see table 1 ). The proportion of people admitted for inpatient treatment also varied among RCTs, ranging from 0–36%. One death was attributed to anorexia nervosa in the control group in one outpatient RCT with a 1-year follow-up. Three deaths attributed to anorexia nervosa occurred in the 5-year follow-up period of one inpatient-based RCT.

Comment

All of the RCTs were small and had limited power to detect clinically important differences between groups. The amount of therapeutic input varied considerably among and within the RCTs. There was variation in methods of recruitment, reporting of key results (e.g. withdrawal rates), and the description of participants' characteristics and selection. We found one systematic review (search date 2002, 2 small RCTs, 32 people) comparing psychotherapies versus each other, although it used a narrower definition of psychotherapy than we have used in this review. Neither RCT identified by the review met our inclusion criteria. One RCT comparing short- versus long-term family therapy found that the severity of obsessive-compulsive features and family structure affected treatment outcome. These findings suggest that people with more severe eating related obsessive-compulsive thinking and those from non-intact families may respond better to long-term treatment. Studies of adult participant reactions to treatment indicate that outpatient psychotherapy is a preferred means of treatment, although those focusing specifically on weight gain were viewed negatively by participants.

Substantive changes

Psychotherapy Two systematic reviews added, which identified RCTs previously reported in this review. No new data added. One already included systematic review updated with new seach date of 2005. No new data added. One RCT comparing CBT plus placebo versus fluoxetine plus CBT added. This study had a high withdrawal rate, which limits the conclusions that can be drawn. Categorisation unchanged (Unknown effectiveness).

BMJ Clin Evid. 2009 Mar 10;2009:1011.

SSRIs

Summary

SYMPTOM IMPROVEMENT Fluoxetine or citalopram compared with placebo: Fluoxetine may be no more effective than placebo at improving weight gain, eating symptoms, or depressive symptoms in women inpatients with anorexia nervosa who also have individual and group psychotherapy. We don't know whether citalopram is more effective than waiting list control at improving weight gain or self-reported depressive symptoms in adults with moderately severe restricting-type anorexia nervosa ( very low-quality evidence ). Fluoxetine plus psychotherapy compared with psychotherapy or fluoxetine: We don't know whether fluoxetine plus CBT is more effective than placebo plus CBT at improving symptoms or increasing time to relapse in women with anorexia nervosa who have previously successfully completed treatment in an inpatient or day programme setting (very low-quality evidence).

Benefits

Fluoxetine or citalopram versus placebo:

We found three systematic reviews (search date 2005). The reviews did not pool data. They found three small RCTs of sufficient quality. The first RCT (33 women, mean age 26.2 years, mean BMI 15.0 kg/m2, mean duration of anorexia nervosa 8.0 years) compared fluoxetine 60 mg versus placebo for the duration (mean 36 days) of inpatient treatment, which included individual and group psychotherapy. Two people withdrew early from the fluoxetine group. The RCT found no significant difference in weight gain, eating symptoms, or depressive symptoms between fluoxetine and placebo (reported as non-significant; P value not reported). The second RCT (39 women, binge–purge subtype of anorexia nervosa excluded, mean age about 22 years, mean duration of anorexia nervosa 4–7 years) compared fluoxetine (starting dose 20 mg/day) versus placebo for 1 year. All women had been discharged from hospital after weight gain (minimum weight restoration was to 75% of average body weight). Women were allowed additional psychotherapy. Women who had substantial and incapacitating symptoms were encouraged to withdraw from the study. Withdrawal rates were too high to draw reliable conclusions about effects, although the withdrawal rate was significantly lower with fluoxetine compared with placebo (6/16 [38%] with fluoxetine v 16/19 [84%] with placebo; RR 0.45, 95% CI 0.23 to 0.86). The third RCT (52 adults with moderately severe restricting-type anorexia nervosa [BMI 15.8 kg/m2]) compared citalopram (10 mg/day increasing to 20 mg/day) versus waiting list control for 12 weeks before the start of standard integrated dietary and psychiatric treatment. Reliability was limited because withdrawal rates were high (7/26 [27%] with citalopram v 6/26 [23%] with control). The RCT found similar weight gain between citalopram and control (mean 2.99 kg with citalopram v 1.44 kg with control). It found that self-reported depressive symptoms (and some additional measures of comorbidity) improved in the citalopram group only (change in weight from baseline to 12 weeks: from 43.5 kg to 46.5 kg with citalopram v from 42.5 kg to 43.9 kg with control; P value not reported; Beck Depression Inventory: 14.5 to 7.3 with citalopram v 12.7 to 12.3 with control; P value not reported).

Fluoxetine versus psychotherapy:

See benefits of psychotherapy.

Harms

Fluoxetine or citalopram versus placebo:

There have been a number of alerts regarding the use of SSRIs in adults and adolescents (see reviews on Depression in adults: drug and other physical treatments and Generalised anxiety disorder). The RCT comparing citalopram versus control did not report adverse effects or reasons for withdrawal.

Fluoxetine versus psychotherapy:

See harms of psychotherapy.

Comment

Fluoxetine or citalopram versus placebo:

In the second RCT, four further women were excluded from the analysis. Three became aware of the treatment allocation and one stopped taking medication before the end of 30 days.

Substantive changes

SSRIs Three systematic reviews added, which did not pool data, and which identified three RCTs comparing fluoxetine or citalopram versus placebo which were previously reported in this review. No new data added from the reviews. One RCT comparing fluoxetine plus CBT versus placebo plus CBT added which had a high withdrawal rate which limited any conclusions that could be drawn. Categorisation unchanged (Unknown effectiveness).

BMJ Clin Evid. 2009 Mar 10;2009:1011.

Anxiolytic drugs (benzodiazepines, older-generation antipsychotics, atypical antipsychotics)

Summary

We found no direct information from RCTs about anxiolytic drugs in the treatment of people who have anorexia nervosa. NOTE The QT interval may be prolonged in people with anorexia nervosa, and many neuroleptic drugs (haloperidol, pimozide, sertindole, thioridazine, chlorpromazine, and others) also increase the QT interval. Prolongation of the QT interval may be associated with increased risk of ventricular tachycardia, torsades de pointes, and sudden death.

Benefits

We found no systematic review or RCTs of sufficient quality on the effects of anxiolytic drugs (benzodiazepines, older-generation antipsychotics, or atypical antipsychotics).

Harms

General harms of neuroleptic drugs are described in the review on schizophrenia (see review on schizophrenia). The QT interval may be prolonged in people who have anorexia nervosa, and many neuroleptic drugs (haloperidol, pimozide, sertindole, thioridazine, chlorpromazine, and others) may also increase the QT interval. One observational study (495 people with mental illness and 101 healthy controls) found an increased risk of prolonged QT interval with high- and very high-dose neuroleptic use, after adjusting for age and other drug use (high dose: adjusted OR 3.4, 95% CI 1.2 to 10.1; very high dose: adjusted OR 5.6, 95% CI 1.6 to 19.3).The FDA has issued a drug safety alert on haloperidol relating to cardiovascular adverse effects and sudden death relating to the risk of QT prolongation and torsades de pointes (www.fda.gov/cder/drug/InfoSheets/HCP/haloperidol.htm).

Comment

Prolongation of the QT interval may be associated with an increased risk of ventricular tachycardia, torsades de pointes, and sudden death.

Clinical guide:

Given the challenges in identifying appropriate pharmacotherapy for anorexia nervosa, several case reports, open trials, and one placebo-controlled trial have been conducted using atypical antipsychotics, most notably olanzapine. These are typically prescribed during acute refeeding to assist with management of agitation, obsessions, and anxiety related to refeeding, but have also been explored for their potential to increase appetite and weight. A review of case studies and open trials indicated preliminary support for the use of olanzapine for reducing agitation and psychological distress during weight restoration. This led to a single placebo-controlled clinical trial. In this trial, 30 female outpatients (18 with restricting-type anorexia and 12 with a binge–purge subtype) completed a short-duration trial of olanzapine compared with placebo. Participants were 18–35 years of age and were given olanzapine or a sugar placebo for 3 months. In addition, participants in both groups completed weekly CBT sessions. Although both groups gained weight, evidence suggested that olanzapine was useful in reducing obsessions and other psychological variables associated with relapse. Although the authors conducted stratified analyses by type of anorexia, the sample sizes were too small to draw valid conclusions by subtype. One small RCT compared olanzapine versus chlorpromazine in 15 adult inpatients with anorexia nervosa. Individualised doses were not reported and the length of treatment varied between groups (mean length of treatment: 46 days in olanzapine group v 53 days in chlorpromazine group). It found no significant difference between groups in average weight gain, although the authors reported a reduction in ruminative thinking with olanzapine. Case reports and open label trials have also been reported assessing risperidone and quetiapine. However, evidence from large, well-conducted RCTs is necessary in order to draw reliable conclusions about effectiveness, dosing, and clinical indications. As with other trials for medication to treat anorexia nervosa, studies of aytpical antipsychotics are currently compromised by small sample sizes and particpant loss due to failure to take up treatment or treatment withdrawal. Some atypical antipsychotics do not seem associated with the same cardiac risks as older-generation antipsychotics. However, further research needs to be done.

Substantive changes

Anxiolytic drugs (benzodiazepines, older-generation antipsychotics, atypical antipsychotics) No new RCTs added to the benefits or harms sections. One small RCT (not meeting Clinical Evidence inclusion criteria for this review), controlled trials, and observational data on atypical antipsychotics added to the comment section as background data.

BMJ Clin Evid. 2009 Mar 10;2009:1011.

Tricyclic antidepressants

Summary

SYMPTOM IMPROVEMENT Tricyclic antidepressants compared with placebo: Amitriptyline may be more effective at increasing the rate of weight gain, but may be no more effective at improving global response (defined as a more than 50% improvement) in people with anorexia nervosa ( very low-quality evidence ). NOTE The QT interval may be prolonged in people who have anorexia nervosa, and tricyclic antidepressants (amitriptyline, protriptyline, nortriptyline, doxepin, and maprotiline) also increase the QT interval. Prolongation of the QT interval may be associated with increased risk of ventricular tachycardia, torsades de pointes, and sudden death.

Benefits

Tricyclic antidepressants versus placebo:

We found three systematic reviews (search date 2005). Two reviews were narrative and did not pool data. The other review pooled data for tricyclic antidepressants; however, there was significant heterogeneity in the analysis, so it presented data separately for the included RCTs.The reviews identified two small RCTs of amitriptyline of sufficient quality, both of which were of short duration. The first RCT (43 people, 5 of them outpatients, with early-onset and short-duration anorexia nervosa, mean age 16.6 years, mean 27% below average weight, mean duration of anorexia nervosa 1.5 years) compared amitriptyline versus placebo for 5 weeks. Participants also received various kinds of psychotherapy. A total of 18 people refused to participate and were used as a third comparison group. The RCT found similar effects at 5 weeks among all groups on all of the outcome scales measured (more than 50% improvement in global response: 1/11 [9%] with amitriptyline v 1/14 [7%] with placebo v 0/18 [0%] with control; CI not reported). The second RCT (72 women, mean age 20.6 years, mean duration 2.9 years) compared amitriptyline (up to a maximum of 160 mg), cyproheptadine, and placebo for 6 weeks. It found that amitriptyline significantly increased the rate of weight gain compared with placebo (mean days to target weight: 32 with amitriptyline v 45 with placebo; P less than 0.05). Women were withdrawn from the trial if they failed to gain at least 2 kg after 6 weeks; 10 women were withdrawn for this reason, but the total withdrawal rate is unclear. The authors adjusted for withdrawals by assigning a standard time to attainment of target weight of 120 days to all women who withdrew. It is not clear if people also received psychotherapy.

Harms

In the first RCT, adverse effects more common with amitriptyline than with placebo included increased perspiration (2/11 [18%] with amitriptyline v 0/14 [0%] with placebo), drowsiness (6/11 [55%] v 0/14 [0%]), dry mouth (4/11 [36%] v 2/14 [14%]), blurred vision (1/11 [9%] v 0/14 [0%]), urinary retention (1/11 [9%] v 0/14 [0%]), hypotension (2/11 [18%] v 0/14 [0%]), and leukopenia (1/11 [9%] v 0/14 [0%]; statistical analysis between groups not reported). Adverse effects more common with placebo than with amitriptyline included palpitations (0/11 [0%] with amitriptyline v 1/14 [7%] with placebo) and dizziness (0/11 [0%] v 2/14 [14%]; statistical analysis between groups not reported). The second RCT found that fewer people taking amitriptyline had adverse physical symptoms (including drowsiness, excitement, tachycardia, and increased motor activity) rated moderate or severe compared with people taking placebo (absolute numbers not reported). The QT interval may be prolonged in people who have anorexia nervosa,and tricyclic antidepressants (amitriptyline, protriptyline, nortriptyline, doxepin, and maprotiline) also increase the QT interval. In an observational study (495 people with mental illness and 101 healthy controls), an increased risk of prolonged QT interval was seen with tricyclic antidepressant use, adjusting for age and other drug use (adjusted OR 2.6, 95% CI 1.2 to 5.6). General harms of tricyclic antidepressants are described in the review on depression in adults: drug and other physical treatments. Prolongation of the QT interval may be associated with an increased risk of ventricular tachycardia, torsades de pointes, and sudden death.

Comment

None.

Substantive changes

Tricyclic antidepressants Three systematic reviews added. The reviews did not pool data, and identified two RCTs comparing amitriptyline versus placebo which were already reported in this review. No new data added. Categorisation unchanged (Likely to be ineffective or harmful).

BMJ Clin Evid. 2009 Mar 10;2009:1011.

Oestrogen treatment

Summary

BONE MINERAL DENSITY Oestrogen treatment compared with control: Oestrogen treatment (as oral contraceptives or HRT) may be no more effective than placebo or no treatment at improving bone mineral density in women who have anorexia nervosa ( very low-quality evidence ). NOTE We found no RCTs that measured the effects of oestrogen treatment directly on fracture rates; all the RCTs we identified measured surrogate outcomes such as changes in bone mineral density.

Benefits

Oestrogen treatment versus control:

We found two systematic reviews (search date 2005), which both identified the same RCT. We also found one additional RCT and one subsequent RCTexamining the effect of oestrogen treatment on bone mineral density (see comment, below). The RCT (48 women, mean age 23.7 years, mean duration of anorexia nervosa 4 years) identified by the reviews compared an oestrogen group, which was comprised of women either taking HRT (conjugated oestrogens 0.625 mg on days 1–25 of each month plus medroxyprogesterone 5 mg on days 16–25) (16 women) or an oral contraceptive containing 35 micrograms ethinyl oestradiol (6 women), depending on whether oral contraception was required, versus a no treatment group (26 women) for 6 months. All women maintained a calcium intake of 1500 mg using oral calcium carbonate, and spinal bone mineral density was measured. A total of 44/48 [92%] women completed the study. The RCT found a similar final bone mineral density at follow-up between the oestrogen group and the no treatment group (mean: 128 mg K2 HPO4/cm3 in the oestrogen group v 132 mg K2 HPO4/cm3 in the no treatment group; P value not reported). The RCT did not separately report results for HRT alone versus no treatment or oral contraceptive alone versus no treatment.The additional RCT (60 women aged 18–38 years, mean weight 44.7 kg, BMI 16.6 kg/m2, duration of anorexia nervosa 2.3 years, and with osteopenia at entry) compared four treatments: oral contraceptive alone (35 micrograms ethinyl oestradiol plus 0.4 mg norethindrone); placebo; recombinant human insulin-like growth factor I alone; and oral contraceptive plus recombinant human insulin-like growth factor I. In addition, all women received calcium 1500 mg daily and vitamin D 400 IU daily. A total of 7/60 people (12%) did not complete the study, with similar rates of withdrawal in each group. The RCT found no significant difference between oral contraceptives and placebo in bone mineral density at 9 months (hip density: P = 0.071; spine density: P = 0.21). The subsequent double-blind RCT examined the effects of a triphasic oral contraceptive containing noregesimate/ethinyl estradiol (NGM/EE) on bone mineral density in adolescent females aged 11–17 years who have anorexia nervosa or eating disorder not otherwise specified (EDNOS), with an age-adjusted BMI less than 10%. It compared noregesimate/ethinyl estradiol (noregesimate 180–250 micrograms and ethinyl estradiol 35 micrograms on days 1–21 and inactive tables on days 22–28) versus placebo for 13 consecutive 28-day cycles. All participants were also given a tablet containing 400 IU vitamin D plus 500 mg of calcium carbonate daily. Of 73 women initially randomised to each group (146 women in total), 23 were omitted as they were screening failures, leaving 123 women, of which 61 took at least one dose of noregesimate/ethinyl estradiol, and 62 took at least one dose of placebo. The RCT found no significant difference in bone mineral density between groups after 13 cycles (bone mineral density change from baseline to cycle 13 measured by dual energy absorptiometry scan [DXA] at lumbar spine [L1-L4] or hip: lumbar spine, NGM/EE v placebo, P = 0.24; hip, NGM/EE v placebo, P = 0.78). Results were based on 112/123 [91%] people who had taken at least one dose of treatment. Interpretation is complicated by treatment withdrawal, with only 40/61 [66%] people in the noregesimate/ethinyl estradiol group and 49/62 [79%] people in the placebo group completing 13 cycles of treatment. The rate of exposure was lower for those in the noregesimate/ethinyl estradiol group compared with placebo (mean duration: 280 days with NGM/EE v 329 days with placebo; P value not reported).

Harms

Oestrogen treatment versus control:

In the RCT identified by the reviews, three women withdrew from oestrogen treatment; two because of adverse effects, and one because she had left the country. One woman in the control group was unwilling to return for further testing. The additional RCT reported that one woman in the oral contraceptive alone group and one women in the placebo group had breast tenderness.In the subsequent RCT, 11 people in the noregesimate/ethinyl estradiol group and 6 people in the placebo group discontinued due to "subject choice" (not further defined; P value not reported). Four people were withdrawn due to adverse effects (3 [5%] with noregesimate/ethinyl estradiol [nausea, weight increase, irregular periods] v 1 [2%] with placebo [headache and nausea]; P value not reported). In particular, one person with noregesimate/ethinyl estradiol was withdrawn due to bone loss (11% loss of lumbar spine bone mineral density at cycle 6). See also harms of oestrogen in review on stress incontinence.

Comment

Improvements in bone mineral density may not reduce fracture risk.

Substantive changes

Oestrogen treatment Two systematic reviews added, which identified the same RCT previously reported in this review. No new data added. One subsequent RCT added, which found no significant difference between a triphasic oral contraceptive and placebo in bone mineral density after 13 cycles.Categorisation unchanged (Unknown effectiveness).

Articles from BMJ Clinical Evidence are provided here courtesy of BMJ Publishing Group

RESOURCES