Remvig 1987.

Methods	Allocation: randomised. Blindness: double. Duration: 3 to 12 weeks. Setting: inpatient, multi‐centre.
Participants	Diagnosis: acute psychosis or exacerbation of a chronic psychosis. N = 54. Age: 20 to 60 years, mean 38 ˜ years. Sex: 25 F, 15 M (of those having completed at least 3 weeks). History: recent acute episode; inpatient; more than half of participants had received neuroleptic medication previously, prior to admission. Included: no details. Excluded: suffering from a serious somatic illness (including decreased kidney and liver function); organic brain damage; pregnancy; history of abuse; previous good response to a particular neuroleptic drug. Consent: informed consent from participants had to be obtained.
Interventions	1. Perphenazine: initially 16 to 24 mg daily, dose could be increased 2 or 3 times a week if indicated, total average 30 mg daily + zuclopenthixol placebo, n = 27. 2. Zuclopenthixol: initially 20 to 30 mg daily, dose could be increased 2 or 3 times a week if indicated, total average 37 mg daily + perphenazine placebo, n = 27. Additional medication: biperiden or orphenadrine for parkinsonism, benzodiazepines, methotrimeprazine for additional neuroleptic treatment.
Outcomes	Leaving the study early. Global state: no better or worse. Adverse events: UKU‐Scale for side effects. Unable to use ‐ Mental state: CPRS (unpublished scale). CGI score ("slightly modified" version of the scale used).
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomised: no further details.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Double‐blind: quote, "double‐blind, double‐dummy" (p147).
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote, "all patients treated for 3 weeks (21 days) or more were included in the statistical processing of data" (p148). Fourteen participants (n = 5 on zuclopenthixol; n = 9 on perphenazine) had not been treated for the 3‐week minimum, and were excluded ‐ reasons including insufficient effect, undesired effect, or discharge from hospital. Binary data analysed with 74% (ITT); continuous data using CGI scale presented data for n = 25 (46%), no LOCF used.
Selective reporting (reporting bias)	High risk	Quote: "because of the great variation in total scores among the patients, they were stratified post hoc into two subgroups, depending on the total score at baseline" (p149).
Other bias	High risk	Funding: not stated ‐ one author was affiliated with Lundbeck (producer of zuclopenthixol for the study drug). Raters: not stated to be independent of treatment. Quote: "as there was no information available about any causal relationship between symptoms and test treatment for may patients... it was assumed, when the side effects were analyzed, that all side effects were drug‐induced" (p148). Quote: "no additional neuroleptic treatment was allowed" (p148), however, it is later stated that "most patients received benzodiazepines... and 12 patients received an additional neuroleptic in the form of methotrimeprazine" (p149).