Medication versus placebo for anorexia nervosa: antidepressants, antipsychotics and hormonal therapy


Anorexia nervosa (AN) is an eating disorder that is characterised by a body weight that is at least 15% below what would be expected, accompanied by deliberate weight loss (e.g. through excessive exercise or avoidance of high calorie foods) and a distorted body image with a fear of weight gain.

Whilst no single line of causality has been established, it is generally accepted that a combination of sociocultural and biological factors have a part to play, whilst some personality traits may also be a risk factor (Word Health Organisation, 1992).

AN has a prevalence of between 0.5 – 1.0% (Hudson, Hiripi, Pope and Kessler, 2007) and has typically been associated with females and adolescents, with recent research suggesting that more than 1 in 20 female adolescents are affected (Smink, van Hoeken, Oldehinkel & Hoek, 2014).

AN may be treated in a number of ways in primary care, inpatient and outpatient settings. At present, psychological therapies are considered crucial in the treatment of eating disorders, whilst pharmacotherapy may be an adjunctive treatment (NICE, 2004).  One rationale for the use of pharmacotherapy is the comorbidity between eating disorders and other mental health conditions such as depression and anxiety. A second rationale rises from taking a neurobiological approach to eating disorders and considering the biological drives towards appetite.

A number of types of medication may be prescribed, namely antipsychotics, antidepressants and hormonal therapy. However, the evidence base for these treatments is lacking. In particular, it is currently unclear to what extent any improvements born from pharmacotherapy may actually be a result of the placebo effect.

A recent meta-analysis therefore aimed to draw together existing research on the use of antipsychotics, antidepressants and hormonal medication versus a placebo in the treatment of eating disorders (de Vos et al, 2014).

Anorexia is thought to affect more than 1 in 20 teenage girls.

Anorexia is thought to affect more than 1 in 20 teenage girls.


  • This meta-analysis formed part of a wider meta-analysis into eating disorders
  • The authors conducted a search for published randomised controlled trials on PubMed, PsychINFO, Embase and Cochrane Library
  • Criteria for inclusion were:
    • The study was a randomised controlled trial
    • The study compared pharmacotherapy to a placebo condition
    • Participants within the study had a diagnosis of Anorexia Nervosa and were at least 12 years old
    • Weight change was the outcome measure
  • Only studies that were written in English or Dutch were included
  • The authors conducted a separate meta-analysis for each type of medication (antipsychotic, antidepressant or hormonal medication)
  • Subgroup analyses were conducted for setting (inpatient, outpatient or other) and whether or not pharmacotherapy was the sole form of treatment
Outcomes were measured in terms of changes in weight.

Outcomes were measured in terms of changes in weight.


Study characteristics

18 studies with 869 participants were included in the analysis.  Of these:

  • 4 studies investigated antidepressants
  • 6 studies looked at antipsychotics
  • 8 studies investigated hormonal medication (e.g. recombinant human insulin-like growth factor or oral contraceptives)

Overall, the quality of the studies included was poor; based on the Cochrane domains, 13 of the 18 studies included were judged as having a high risk of bias.

The included studies differed in several notable ways:

  • The number of participants in the experimental condition ranged from 7 to 55
  • There was variation in the treatment setting:
    • Inpatient n = 6
    • Outpatient n = 10
    • Day care programme n = 1
    • Unknown n = 1)
  • There was variation in the age of participants
    • Adult n = 12
    • Adolescent n = 2
    • Adult and adolescent n = 3
  • There was variation in the recruitment population
    • Clinical n = 14
    • Community and clinical n = 4
  • There was less variation in the type of AN presentation
    • Restriction and binging/purging type n = 16
    • Restricting only n = 1
    • Unknown n = 1
  • Furthermore, there was great variation in adjunctive treatments such as individual, group or family therapy, CBT or caloric repletion

Results of meta-analyses

When the three types of pharmacotherapy were combined, the difference in weight gain compared to placebo was significant, with an effect size of 0.33.

No significant difference in weight gain was found for antidepressants or antipsychotics versus placebos (effect sizes of 0.26 and 0.25, respectively). However, the effect size for antidepressants did increase when adjusted for publication bias, giving a significant effect size of 0.35. It was not possible to conduct subgroup analyses for either antidepressants or antipsychotics due to the small sample sizes.

The difference in weight gain for hormonal medication versus placebo was significant, with an effect size of 0.42. Subgroup analyses were conducted and no significant results were found. Meta-regression analyses were conducted for weeks of treatment (slope = -0.008, p = 0.04), suggesting that shorter term hormonal treatment has a significantly greater effect than longer term treatment. However, it should be noted that heterogeneity for this condition was found to be high.

Mood stabilizers, antipsychotic drugs and hormone therapy may be beneficial in the prevention of postnatal psychotic episodes in women at high risk

This study found some evidence for the efficacy of hormonal medication, but not for antidepressants or antipsychotics.


The authors concluded that:

Hormonal therapy has a significantly larger effect on weight compared to placebo in the treatment of AN. However, for these significant results heterogeneity was high, which means that these significant results have to be regarded with caution.

Antidepressants and antipsychotics are the most commonly used medication treatments for anorexia nervosa in the Netherlands. Yet in both cases we failed to reveal efficacy when compared to placebo.

Strengths and limitations

One of the main weaknesses of this paper is both the quantity and quality of studies included. It is also important to note that several of the studies included in the analysis did not report where their funding came from, which is particularly important in pharmacotherapy research.

Like many meta-analyses, language was part of the inclusion criteria and only English and Dutch papers were included. Furthermore, the inclusion of only published research may have created a bias due to the tendency towards not publishing non-significant results.

The key significant finding in this study was in relation to hormonal medication. However, as the authors highlight, heterogeneity was high for this condition, thus limiting our confidence in the results. They reported no evidence of publication bias in the hormonal therapy research.


All in all, this meta-analysis did a good job at drawing together the existing research in this field, but (as is often the case in eating disorders research) its value is limited by the lack of quality studies available.

I was also disappointed to learn that the only outcomes investigated in this study were related to weight. Whilst weight gain is undoubtedly one treatment goal, other aspects of the condition such as preoccupation with weight and body image distortion are also vital and cannot be ignored in clinical practice.

This study has been useful in highlighting the lack of quality research in this area, and begins to raise questions over the use of pharmacotherapy in clinical practice. However, more research is needed before firm conclusions can be drawn about the efficacy of pharmacotherapy in the treatment of AN.

Better quality research is needed to support the prescribing of medication for anorexia nervosa.

Better quality research is needed to support the prescribing of medication for anorexia nervosa.


Primary study

de Vos J, Houtzager L, Katsaragaki G, van de Berg E, Cuijpers P, Dekker J. (2014). Meta analysis on the efficacy of pharmacotherapy versus placebo on anorexia nervosa (PDF). Journal of Eating Disorders 2014 Oct 30;2(1):27. doi: 10.1186/s40337-014-0027-x. eCollection 2014.

Other references

Hudson, J. I., Hiripi, E., Pope, H. G. Jr & Kessler, R. C. (2007). The prevalence and correlates of eating disorders in the National Comorbidity Survey Replication. Biological Psychiatry, 61 (3) 348-258.

National Institute for Health and Clinical Excellence (2004). CG9 Eating Disorders: Core interventions in the treatment and management of anorexia nervosa, bulima nervosa and related eating disorders (PDF). London: National Institute of Health and Clinical Excellence.

Smink FR, van Hoeken D, Oldehinkel AJ, Hoek HW. (2014) Prevalence and severity of DSM-5 eating disorders in a community cohort of adolescents. The International Journal of Eating Disorders. 2014; 47(6): 610-619. [PubMed abstract]

Word Health Organisation (1992). The ICD-10 classification of mental and behavioural disorders: Clinical descriptions and diagnostic guidelines. Switzerland: World Health Organisation

Share on Facebook Tweet this on Twitter Share on LinkedIn Share on Google+