Bennett 1961.

Methods	Allocation: randomised. Blindness: unclear. Duration: 12 weeks, preceded by 2 weeks wash‐out. Setting: hospital, single‐centre.
Participants	Diagnosis: all except two had chronic schizophrenia. N = 30 (n = 25 relevant). Age: median ˜ 37 years. Sex: 30 M. History: chronic; 'ill for more than two years'; median number of years hospitalised 2.6; Multidimensional Scale for Rating Psychiatric Patients (MSRPP) showed patients to be more 'perceptually confused, paranoid and mentally agitated than the usual male patients confined to Veterans Administration hospitals'; n = 22 were previously receiving phenothiazine derivatives without sustained improvement; n = 10 received courses of electric shock therapy; n = 7 had been given insulin coma therapy. Included: not stated. Excluded: history of central nervous system disease, seizures, lobotomy, somatic illnesses, which included duodenal ulcer, liver disease, blood dyscrasia, or those psychotic as a result of infection or toxic factors. Consent: not stated.
Interventions	1. Perphenazine: 16 mg/day gradually increased until by the 5th week 48 mg/day, then until the 12th week 96 mg/day, N = 5. 2. Chlorpromazine: 200 mg/day gradually increased until by the 5th week 600 mg/day, then until the 12th week 1200 mg/day, N = 5. 3. Prochlorperazine: 25 mg/day gradually increased until by the 5th week 75 mg/day, then until the 12th week 150 mg/day, N = 5. 4. Triflupromazine: 50 mg/day gradually increased until by the 5th week 150 mg/day, then until the 12th week 300 mg/day, N = 5. 5. Mepazine: 50 mg/day gradually increased until by the 5th week 150 mg/day, then until the 12th week 300 mg/day, N = 5. [6. Phenobarbital: 32 mg/day gradually increased until by the 5th week 100 mg/day, then until the 12th week 200 mg/day, N = 5 ‐ not included in data and analysis, not an antipsychotic].
Outcomes	Leaving the study early. Adverse events: EEG results. Unable to use ‐ Global state: no better or worse (no data for individual groups). Mental state: MSRPP, CEPS (no data). Adverse events (no data). Laboratory tests (no data). Cognitive functioning: Porteus Maze (no data).
Notes	Prior to treatment, participants were given inert identical‐appearing placebo for 14 days; "judicious use of conventional short‐acting sedatives was allowed during first 4 weeks to control behaviour/ insomnia".
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	No details.
Blinding (performance bias and detection bias) All outcomes	High risk	Single‐blind (implied) ‐ all participants were given identical‐looking placebo 14 days prior to treatment; there is no mention of blinding at treatment commencing.
Incomplete outcome data (attrition bias) All outcomes	High risk	Follow‐up 90% ‐ n = 2 people left early (n = 1 from the chlorpromazine group; n = 1 from the prochlorperazine), no explanation provided; not clear which numbers are completed in full analysis and by which group.
Selective reporting (reporting bias)	High risk	Non‐reported data by group include: improvement; behaviour (MSRPP); mental state (CEPS); side effects; laboratory testing.
Other bias	Unclear risk	Funding: no details provided. Raters: not stated to be independent of treatment.