Evid Based Mental Health 16:18 doi:10.1136/eb-2012-101072
  • Therapeutics

Non-pharmacological interventions reduce antipsychotic-associated weight gain in outpatients


Question: What are the effects of non-pharmacological interventions on antipsychotic-associated weight gain and metabolic abnormalities?

Outcomes: Primary outcomes: body weight and body mass index (BMI). Secondary outcomes: waist circumference, body fat percentage, total cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglycerides, systolic sitting blood pressure and all-cause discontinuation.


Design: Systematic review and meta-analysis.

Data sources: PsycInfo, MEDLINE, PubMed, CINAHL and the Cochrane Library (search date not reported). Reference lists of relevant articles were also searched for additional studies.

Study selection and analysis: Randomised controlled trials (RCTs) of non-pharmacological interventions which aimed to prevent or reduce antipsychotic-associated weight gain were included. All data were extracted by one author and verified by a second author. Data were pooled using random effects models in Review Manager 5.0.24 software. Study heterogeneity was measured using the I2 statistic. Sensitivity analyses were carried out to assess the effects of potential moderators.

Main results

Of the 17 RCTs meeting inclusion criteria, 7 trialled cognitive behavioural therapy and 10 assessed nutritional and/or exercise interventions. The intervention duration ranged from 8 to 72 weeks. In total, the RCTs included 810 participants with the mean age of 38.1 years in the intervention groups and 37.2 years in the control groups. The non-pharmacological interventions significantly reduced weight gain and BMI compared with body weight change in controls: 14 RCTs, weighted mean difference (WMD) −3.12 kg, 95% CI −4.03 to −2.21, p<0.0001, I2=42%; BMI change: 16 RCTs, WMD −0.94 kg/m2, 95% CI −1.45 to −0.43, p=0.0003, I2=75%). Sensitivity analyses showed that improvements in weight and BMI were only achieved in outpatient populations (p<0.0001), and not in inpatient or mixed samples (p values 0.09–0.96). The studies with follow-up data showed that the weight change benefit of the intervention was maintained at between 2 and 12 months after the intervention (5 RCTs, n=220, WMD −3.48 kg, 95% CI −6.37 to −0.58, p=0.02), but the BMI benefit was not maintained (5 RCTs, n=211, WMD −0.72 kg/m2, 95% CI −2.36 to +0.58, p=0.40). The intervention significantly reduced waist circumference and body fat percentage compared with controls (waist circumference: 6 RCTs, n=349, WMD −3.58 cm, 95% CI −5.51 to −1.66, p=0.03; body fat percentage: 3 RCTs, n=83, WMD −2.83%, 95% CI −5.35% to −0.30%, p=0.03). Compared to controls, the intervention also reduced total cholesterol (WMD −20.98 mg/dl, 95% CI −33.78 to −8.19, p=0.0001), LDL (WMD −22.06 mg/dl, 95% CI −37.80 to −6.32, p=0.006) and triglycerides (WMD −61.68 mg/dl, 95% CI −92.77 to −30.59, p=0.0001).


Non-pharmacological interventions were effective at reducing antipsychotic-associated weight and BMI gain and metabolic abnormalities. The effects on weight and BMI appear to be restricted to outpatients.

Abstracted from

[CrossRef][Medline] Search Google Scholar


  • Correspondence to: Lawrence Maayan, New York University School of Medicine Child Study Center, New York, NY, USA;

  • Sources of funding Zucker Hillside Hospital National Institute of Mental Health (NIMH) Advanced Center for Intervention and Services Research for the Study of Schizophrenia, and Stanley Medical Research Institute Award.


  1. John Lally
  1. National Psychosis Service, South London and Maudsley NHS Foundation Trust, London, SE5 8AF, UK

The huge reduction in life expectancy which comes with a diagnosis of schizophrenia is largely explained by cardiovascular disease.1 While smoking, poor diet and limited exercise are important modifiable risk factors, clinicians, patients and families are all too aware of the increase in weight which occurs once antipsychotic medications start.2 In this meta-analysis of 17 randomised controlled trials, Caemmerer and colleagues confirm that non-pharmacological interventions (NPIs) reduce antipsychotic-induced weight gain and improve certain cardiometabolic outcomes. This benefit is independent of treatment modality and duration; whether the treatment is group or individual, or designed to prevent or reverse cardiovascular risk.

The effectiveness of NPIs in outpatients is encouraging, as outpatients have discretion over their diet and exercise habits. Inpatients benefitted less, perhaps related to an inadequately powered analysis. However, given that many people start antipsychotics as an inpatient, further research is needed on prevention of weight gain at this critical and difficult time. Likewise, this analysis highlights the need for evidence on the optimum duration of NPIs to allow persistence of cardiovascular benefits.

The similar numbers needed to treat for NPIs and pharmacological interventions is good news, as the possibility of emergent cardiovascular risk does not automatically mean more tablets. Nevertheless, NPIs may not suit everyone prescribed with antipsychotics and there is limited information on which clinical and motivational factors predict entry to NPI trials. Such knowledge may help guide when pharmacological or combined NPI and pharmacological approaches are preferable to NPIs alone.

Formal cognitive behavioural therapies had no advantage over nutritional interventions. This ratifies the use of less formal interventions in routine practice, which may be easier to staff and indeed cheaper to implement.

This meta-analysis, the largest to date, shows that NPIs work in the prevention and reduction of weight gain and cardiometabolic risk associated with antipsychotic treatment. The urgency now is to integrate NPIs into widespread clinical practice.


  • Competing interests FG has a family member with professional links to GSK and Lilly, and has received funding from various different pharmaceutical companies.