Arman 2008.

Study characteristics
Methods	Randomisation: randomised, no other details Blinding: double‐blind Duration: 12‐weeks Country: Iran
Participants	Diagnosis: schizophrenia or schizoaffective disorder N = 49 Age: 10.09 years (< 20 years) Sex: male and female Setting: inpatients History: < 20 years of age, taking risperidone (2‐6 mg/d according to their responses to treatment), creatinine level < 1.4 mg/dL, normal liver function test Excluded: treatment with antipsychotic earlier, current substance abuse or significant medical illness, untreated hypertension, history of intolerance to metformin, receiving weight loss agents, glaucoma, heart disease, abnormal ECG, asthma, combination of antipsychotics, or treatment with anti‐migraine agents containing serotonin agonists
Interventions	Metformin (1000 mg/d; started as 500 mg tablet for week 1, with an increase to 500 mg tablets twice a day) in combination with risperidone (2‐6 mg/d); N = 16 Placebo (once/d during week 1, with an increase to twice/d after week 1) in combination with risperidone (2‐6 mg/d); N = 16
Outcomes	Able to use: Primary outcome Weight measures Weight BMI Unable to use: Secondary outcome Physiological: laboratory measures Fasting blood glucose (data not available) Complete blood count (data not available) Creatinine (data not available) Prolactin level (data not available) Liver function tests (data not available)
Notes	61.4% study completers
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "After completing baseline assessments, the subjects were randomly assigned to metformin and placebo." Pg 1131 Comment: randomisation methods are unavailable
Allocation concealment (selection bias)	High risk	Comment: information is unavailable. Combined with the lack of details about randomisation and complete absence of information about allocation concealment, the risk of bias is quite high.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "...Identical appearing placebo pills..." Pg 1131 Comment 1: the study indicates the presence of identical placebo pills, however, it is unclear which personnel were blinded in the study. Comment 2: various adverse effects were reported in the treatment group, which may have broken blinding.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Information is unavailable
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "17 were excluded due to incomplete use of drugs or side‐effects of the drugs. 2 of these patients have experienced diarrhoea at the second week. 3 patients had nausea and vomiting in metformin group, which were excluded from study too." Pg 1132 Comment: excluded participants were not included in the analysis, hence only ~62% of the study population was analysed.
Selective reporting (reporting bias)	Unclear risk	Study protocol is unavailable
Other bias	Low risk	No obvious bias