*Doongaji 1989.

Methods	Allocation: randomly assigned, no further information. Blindness: double. Design: 2 parallel groups, single centre. Duration: 6 weeks (preceded by one placebo week). Analysis: not clearly described in the report, (ITT)?*. Country: India.
Participants	Diagnosis: schizophrenia (RDC, instrument 58 New York State Psychiatric Institute, 1975). N=40. Age: range 18‐45 years, mean ˜30 years. Sex: M 22, F 7. Setting: hospital. History: acute (mean duration of illness 9.7 months). Exclusion: those who totalled 35 points on BPRS initially, or whose total score reduced by 20% at one placebo week; concurrent systemic illness; pregnant women.
Interventions	1. Trifluoperazine: range 15‐22.5 mg/day. N=20. 2. Centbutindole: range 3‐45 mg/day. N=20.
Outcomes	Leaving the study early. Side effects. Mean dose of antiparkinson drug. Unable to use ‐ Global state: (CGI ‐ no SD). Mental state: (BPRS ‐ no SD).
Notes	*ITT unclear.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear