Article Text

Download PDFPDF

Metformin for weight loss in schizophrenia: safe but not a panacea
  1. Cenk Tek1,
  2. Lydia Chwastiak2
  1. 1 Department of Psychiatry, Yale University School of Medicine, New Haven, Connecticut, USA; cenk.tek{at}
  2. 2 University of Washington, Seattle, Washington, USA; lchwast{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

What is already known on this topic?

Obesity and diabetes rates are two to three times higher among patients with schizophrenia than they are for the general population, as reflected by higher morbidity and mortality rates. Weight gain induced by antipsychotic medication is implicated. Metformin is shown to delay diabetes conversion and produce modest weight loss among prediabetic obese patients in the general population.1 It has also been shown to attenuate olanzapine-induced weight gain in patients with first-episode schizophrenia.2

What does this paper add?

  • Jarskog and colleagues conducted the largest double-blind, placebo-controlled trial of metformin for weight loss in overweight, non-diabetic schizophrenia patients.

  • Metformin combined with lifestyle and diet intervention (LS) produced an average 3 kg weight loss, compared to 1 kg in the placebo plus LS group. Of the patients on metformin, 17% lost clinically significant weight, compared with 10% in the placebo group.

  • Metformin was well tolerated by patients, while HbA1c levels (but not insulin or fasting glucose levels) improved more in the metformin group.


  • Thirty six per cent of patients received aripiprazole, which is a relatively low-weight-producing antipsychotic. It is unclear when patients were switched; they may already have been on a weight-reduction trajectory.

  • The placebo plus LS arm lost 1 kg on average, which is strikingly lower than other LS intervention trials with this population.

  • Although respectable, a 16-week trial may not be sufficiently long to capture clinical outcomes for weight loss, and is certainly not sufficient to observe any diabetes-protective activity of the drug.

What next in research?

Metformin proved to be moderately useful, but was no panacea for antipsychotic-induced weight gain. Weight loss produced was similar to population studies—however small—compared with weight gain induced by antipsychotics.

The good news is that everything which works for weight loss in the general population appears to work for schizophrenia patients. On the other hand, more research is necessary to find agents which counteract the specific appetite-increasing effects of antipsychotics, which are likely mediated by the central nervous system receptor activity of these medications. In experimental models histamine-1, serotonin-2c and dopamine-2 receptor blockades result in appetite increase. Of these, the dopamine-2 receptor blockade appears to be necessary for antipsychotic action.

Could these results change your practice and why?

This study demonstrated the safety and utility of metformin in this population. Since metformin is well tolerated, it can be utilised for moderate weight loss and diabetes prevention, as long as it is added to a lifestyle intervention and diet programme. The best strategy for the prevention of overweight and obesity in this population still seems to involve the rational choice and dosing of antipsychotic medication, combined with lifestyle recommendations made as soon as possible after diagnosis.3



ABSTRACT FROM: Jarskog LF, Hamer RM, Catellier DJ, et al. Metformin for weight loss and metabolic control in overweight outpatients with schizophrenia and schizoaffective disorder. Am J Psychiatry 2013;170:1032–40.

Patients/participants In total 146 individuals (aged 18–65-years old) with schizophrenia or schizoaffective disorder (DSM-IV), who were receiving outpatient antipsychotic treatment and had a body mass index (BMI) of 27 or greater. Participants were required to have been diagnosed at least 1 year previously, and to have had no change to their antipsychotic medication in the past 2 months, with no dosage change for the past month.

Setting Seventeen academic, Veterans Affairs and private clinics in the USA, March 2009–February 2010.

Intervention Metformin or placebo for 16 weeks. Metformin (n=75) was initiated at 500 mg daily and increased to a maximum of 2 g daily depending on individual tolerance. All individuals also received a weight-reduction behavioural intervention, designed for people with mental health illnesses, consisting of eight 20–30 min lessons and seven phone calls to reinforce the lessons.

Comparison Placebo identical in appearance to metformin (n=71).

Patient follow-up Seventy-nine percent at 16 weeks; no significant differences in dropout between groups.

Allocation: Concealed.

Blinding: Double-blind.


Change in body measures at week 16

People taking metformin lost significantly more body weight than those taking placebo (the primary outcome), and also had significantly greater decrease in BMI. Metformin also caused significantly greater reduction in fasting triglycerides and glycated haemoglobin (HbA1c) than placebo, but had no effect on several other metabolic markers (see webextra table for all outcomes).

Adverse events

Ten serious adverse events were reported (three in the metformin group, seven in the placebo group) none of which were attributed to metformin. Of all adverse events, diarrhoea was significantly more frequent in the metformin group (affecting 33% of patients vs 17% of the placebo group). There were no differences in the rate of other adverse events between groups. A total of 11 individuals in the metformin group (15%) and 8 in the placebo group (11%) discontinued treatment due to adverse events.

Treatment adherence

There were no between-group differences. Adherence to study medication was high in both groups (94.9% in the metformin group and 96% in the placebo group) for either all (81–100%) or most (61% to 80%) of the time. Adherence to the weight-reduction behavioural intervention was reasonable-to-excellent (84.9% in the metformin group and 89.7% in the placebo group).


  • Competing interests None.