Leff 1971.

Methods	Randomisation: random, random number table. Allocation: trial medication was held by the unit secretary and dispensed to Julian Leff who gave it to the treating consultant. Only the unit secretary knew which pills were active drug and which were placebo. Blinding: double, no further details. But side‐effects were not troublesome in any patient and therefore doctors concerned probably received no clues about whether a patient was on active drug or not. Duration: one year. Design: parallel. Location: single‐centre. Setting: outpatient.
Participants	Diagnosis: schizophrenia (Present State Examination), recently recovered from an acute episode, 32 florid schizophrenia, 3 delusional psychosis. N = 20. Gender: n.i.. Age: 16‐55 years. History: duration stable ‐ n.i., but stabilised at the pre‐admission level during a 6‐12 weeks outpatient period and recently recovered from an acute episode, duration ill ‐ n.i., number of previous hospitalisations ‐ n.i., age at onset ‐ n.i., severity of illness ‐ n.i., baseline antipsychotic dose ‐ n.i..
Interventions	1. Trifluoperazine: flexible dose, allowed dose range 5‐25 mg/day, mean dose 12.3 mg/day. N = 14. 2. Chlorpromazine: flexible dose, allowed dose range 100‐500 mg/day, mean dose 157.1 mg/day. N = 6. Rescue medication: antiparkinson medication, antidepressants, no antipsychotics (doctors received a letter asking them not to prescribe other medication).
Outcomes	Relapse: physician was sufficiently concerned about the patient’s status to want to be certain that he was on active drug. Unable to use: Response to treatment (incomplete data, no data for drugs separately, only combined). Leaving the study early (incomplete data, no data for drugs separately, only combined).
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Random, no further details.
Allocation concealment (selection bias)	Low risk	Trial medication was held by the unit secretary and dispensed to Julian Leff who gave it to the treating consultant. Only the unit secretary knew which pills were active drug and which were placebo.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double, no further details. Side‐effects were not troublesome in any patient and therefore doctors concerned probably received no clues about whether a patient was on active drug or not.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double, no further details. Side‐effects were not troublesome in any patient and therefore doctors concerned probably received no clues about whether a patient was on active drug or not.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Drop‐outs were not reported separately for trifluoperazine and chlorpromazine, therefore it is unclear whether there were incomplete outcome data.
Selective reporting (reporting bias)	High risk	Data on leaving the study early and response were not reported separately for trifluoperazine and chlorpromazine.
Other bias	Low risk	No clear other bias.