O´Brien 1974.

Methods	Randomisation: implied randomisation. Blinding: double‐blind. Duration: 3 weeks. Design: parallel. Location: Hospital of the University of Pennsylvania, Philadelphia Naval Hospital (two‐centre). Setting: inpatients.
Participants	Diagnosis (N = 24): paranoid schizophrenia (N = 20), paranoid personality (N = 2), paranoid state (N = 1), explosive personality (N = 1). N = 30. Gender: 22M, 2F. Age: mean 29.5 years. History: duration stable: not indicated, duration of illness: not indicated, number of previous hospitalisations: not indicated, age at onset: not indicated, severity of illness: not indicated, baseline antipsychotic dose: not indicated.
Interventions	1. Haloperidol: flexible dose, dose range: maximum dose 20 mg/day, mean dose: not indicated. N = 15. 2. Trifluoperazine: flexible dose, dose range: maximum dose 48 mg/day, mean dose: not indicated. N = 15. Rescue medication: “benztropine was given only when necessary to control extra pyramidal reactions.” “If parenteral medication was required, intramuscular sodium amytal was given.”
Outcomes	Examined: Clinically important response to treatment: Global clinical outcome rating. Leaving the study early due to any reason. Adverse effects: use of antiparkinson medication. Unable to use: Mental state general: Brief Psychiatric Rating Scale (BPRS) (no raw data available; only the average improvement was provided, no SDs available). Mental state specific: Hamilton Depression Rating Scale (HAM‐D) (no raw data available). Mental state specific: Global Hostility Scale (GH) (no raw data available; only the average improvement was provided, no SDs available). Mental state specific: Global Paranoia Scale (GP) (no raw data available; only the average improvement was provided, no SDs available).
Notes	Study participants were characterised as “hostile, suspicious, uncooperative patients” and a “population of hostile suspicious patients”. “Starting sample was 30 patients but 6 …. were dropped from study”.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomisation mentioned in the publication, but the trial was described as “double‐blind.” Thus it was implied that the study was randomised.
Allocation concealment (selection bias)	Unclear risk	No further details.
Blinding (performance bias and detection bias) (performance bias)	Low risk	“Double‐blind fashion”. “identically appearing pink capsules.”
Blinding (performance bias and detection bias) (detection bias)	Unclear risk	“Double‐blind fashion”. No further details.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The attrition was moderate (3 of 15 participants (20%) in both study arms). Completers‐only analyses were used in the study.
Selective reporting (reporting bias)	High risk	The outcome data were not fully addressed (no raw data for the BPRS, GH, GP, and HAM‐D). Adverse effects were not fully addressed.
Other bias	High risk	“Milieu therapy, group therapy, and individual sessions with a psychiatrist” were allowed during the trial.