Methods	Allocation: random, no further details. Blindness: double, identical capsules. Duration: 14 weeks. Design: parallel. Location: multicentre.
Participants	Diagnosis: (DSM-IV) chronic schizophrenia (n = 135) or schizoaffective disorder (n = 22), suboptimal response to previous treatment, PANSS of 60 or more. N = 167. Sex: 133 M, 24 F (of intent-to-treat population). Age: 18-60 years (mean = 40.8 years) (of intent-to-treat population). History: duration ill mean = 19.5 years (of intent-to-treat population), age at onset not reported. Setting: inpatient.
Interventions	Clozapine: flexible dose. Allowed dose range: 200-800 mg/day. Mean dose: 526.6 mg/day (at the end of the last 6 weeks). N = 40. Haloperidol: flexible dose. Allowed dose range: 10-30 mg/day. Mean dose: 25.7 mg/day (at the end of the last 6 weeks). N = 37. Olanzapine: flexible dose. Allowed dose range: 10-40 mg/day. Mean dose: 30.4 mg/day (at the end of the last 6 weeks). N = 39. Risperidone: flexible dose. Allowed dose range: 4-16 mg/day. Mean dose: 11.6 mg/day (at the end of the last 6 weeks). N = 41
Outcomes	Leaving the study early. any reason, adverse events, inefficacy. Mental State: PANSS total score, PANSS positive subscore, PANSS negative subscore. Quality of life: Quality of life scale, Nurses’observation scale for inpatient evaluation. Cognitive Functioning: Global Neurocognitive Score. Adverse effects: EPS (Use of antiparkinson medication, ESRS), seizures, weight gain, laboratory (cholesterol, glucose, prolactin, white blood cell count) Unable to use: Quality of life scale: no data.
Notes
Risk of bias
Bias	Authors’ judgement	Support for judgement
Adequate sequence generation?	Unclear risk	Random, no further details.
Allocation concealment?	Unclear risk	No further details.
Blinding? subjective outcomes	Unclear risk	Double, identical capsules. Whether blinding was successful has not been examined, but the compounds differ quite substantially in side effects. This can be a problem for blinding
Blinding? objective outcomes	Low risk	Objective outcomes such as laboratory measures or death are unlikely to have been much affected by problems of blinding
Incomplete outcome data addressed? All outcomes	High risk	The overall attrition was high (41.7%). The LOCF method was used to account for people leaving the study early. It assumes that a participant who discontinued the study would not have had a change of his condition if he had remained in the study. This assumption can obviously be wrong
Free of selective reporting?	High risk	Some outcomes were reported on subgroup from the entire sample. Quality of life data were not presented
Free of other bias?	High risk	Quote: “The olanzapine arm was added in November 1997 and required a modified randomization procedure”…It entails the potential for a bias that could be manifested as a cohort effect.”