Perphenazine 1963.

Study characteristics
Methods	Randomisation: arbitrarily allocated. Allocation: procedure not described. Blinding: double‐blind (only the pharmacist knew which bottles were active. Participants were asked whether they were aware of the medication, but only one realised a change in taste. Nurses were also asked, but did not guess the right medication better than by chance alone. Duration: 10 weeks. Design: parallel. Location: two centres. Setting: inpatient.
Participants	Diagnosis: chronic schizophrenia (clinical diagnosis by at least two psychiatrists), all with paranoid condition, two additionally catatonic tendencies and one hebephrenic features, six were leucotomised. N = 26. Gender: 26 men. Age: mean 50.7 years. History: duration stable‐ n.i., but all had been receiving maintenance doses of perphenazine for a mean of 16 months, duration ill‐ n.i., but mean duration of current hospitalisation 16.5 years, number of previous hospitalisations‐ n.i., age at onset‐ n.i., severity of illness‐ n.i., baseline antipsychotic dose‐ two 12 mg three times, one 20 mg three times per day, all other 8 mg three times per day and most less.
Interventions	1. Drug: perphenazine liquid. Fixed dose (same dose as before the start of the study) two 12 mg three times, one 20 mg three times, all other 8mg three times and most less. Mean dose: see above. N = 13. 2. Placebo. duration of taper: 0 days. N = 13. 3. No medication*. duration of taper: 0 days. N = 13. Rescue medication: not allowed apart from antiparkinson medication.
Outcomes	Examined Relapse: "major relapse" = replaced on active medication. Violent/aggressive behaviour. Unable to use/Not included Mental state: self‐developed psychiatric rating scale ‐ unpublished scale (no predefined outcome of interest). Behaviour: Fergus Falls Behaviour Rating Scale (no predefined outcome of interest).
Notes	*This group was not used for the review.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Arbitrarily allocated.
Allocation concealment (selection bias)	Unclear risk	Procedure not described.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind ‐ only the pharmacist knew which bottles were active. Participants were asked whether they were aware of the medication, but only one realised a change in taste. Nurses were also asked, but did not guess the correct medication better than by chance alone.
Blinding (performance bias and detection bias) Subjective outcomes	Low risk	Double‐blind ‐ only the pharmacist knew which bottles were active. Participants were asked whether they were aware of the medication, but only one realised a change in taste. Nurses were also asked, but did not guess the correct medication better than by chance alone.
Blinding (performance bias and detection bias) Objective outcomes	Low risk	Double‐blind ‐ only the pharmacist knew which bottles were active. Participants were asked whether they were aware of the medication, but only one realised a change in taste. Nurses were also asked, but did not guess the correct medication better than by chance alone.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Dropouts were not reported. It is not clear, whether there really no dropouts.
Selective reporting (reporting bias)	Low risk	No evidence for selective reporting.
Other bias	Low risk	No clear evidence for other bias.