Rado 2016.

Study characteristics
Methods	Randomisation: randomised, computer‐generated Blinding: double‐blind, no other details Duration: 24 weeks Country: no details
Participants	Diagnosis: schizophrenia, schizoaffective disorder, bipolar disorder, major depression with psychotic features N = 25 Age: ≥ 18 years Sex: male and female Setting: in‐ and outpatients Excluded: patients with history of diabetes mellitus or a baseline fasting blood glucose level > 126 mg/dL, or 2 random blood glucose levels > 200 mg/dL, or a blood glucose level > 200 mg/dL on a 2‐h oral glucose tolerance test (OGTT); haemoglobin A1c (HbA1c) > 7.0%; baseline serum glutamic oxaloacetic transaminase, serum glutamic pyruvic transaminase, or alkaline phosphatase > 3 times the normal; alcohol dependence or abuse; abnormal kidney function as measured by the Modification of Diet in Renal Disease < 60 mL/1.73 m2 (as an estimation of glomerular filtration rate); unstable medical problems in the opinion of the primary investigators; QTc prolongation > 430 ms on baseline ECG; history of lactic acidosis or hypoglycaemia; current treatment with antidiabetic agents; treatment with antihyperlipidaemic agents within 3 months of randomisation (to better assess the impact of metformin on lipid profile); concurrent treatment with an antipsychotic other than olanzapine; administration of oral corticosteroids; current treatment with topiramate, phentermine, sibutramine, orlistat, or other over‐the‐counter weight‐loss agent; or patients with active homicidal or suicidal ideation; urine pregnancy test was used to exclude pregnant women; patients on lithium or thyroid replacement therapy or with documented thyroid disease underwent serum thyroid‐stimulating hormone testing ‐ those with abnormal values were excluded.
Interventions	Olanzapine + metformin (metformin extended release was titrated to 2000 mg daily as tolerated); N = 12 Olanzapine + placebo; N = 13
Outcomes	Able to use: Primary outcomes Weight measures Changes in body weight Changes in BMI Percent weight change within and between groups Physiological: laboratory measures HOMA‐IR Secondary outcomes Weight measures Change in waist circumference Physiological: laboratory measures Fasting blood glucose HbA1c Fasting lipid profile
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were randomly assigned (without any restriction or stratification) through a computer‐based algorithm to 1 of 2 treatments (olanzapine plus metformin or olanzapine plus placebo)
Allocation concealment (selection bias)	Low risk	Quote: "Metformin and placebo medications were identical in appearance and were provided in coded containers by a separate research pharmacy for each patient according to their randomisation assignment."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants were randomised in double‐blind fashion
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Outcome unlikely to be biased by blinding
Incomplete outcome data (attrition bias) All outcomes	Low risk	In the olanzapine/metformin group, 1 dropout was for drowsiness, and the other was for insomnia. No dropouts occurred in the olanzapine/placebo group
Selective reporting (reporting bias)	Low risk	Measures relevant to the primary outcome were reported
Other bias	Low risk	No obvious bias