Gerlach 1985.

Methods	Randomisation: “randomly assigned”. Blinding: double‐blind. Duration: 12 weeks (first period of the cross‐over‐trial containing altogether two 12‐week periods). Design: cross‐over. Location: not indicated. Setting: inpatients.
Participants	Diagnosis: schizophrenia (based on the criteria of Feighner 1972). N = 28. Gender (N = 20): 17M, 3F. Age (N = 20): mean 34 years. History: duration stable: not indicated, duration of illness: mean 9 years, number of previous hospitalisations: not indicated, age at onset: not indicated, severity of illness: not indicated, baseline antipsychotic dose: not indicated.
Interventions	1. Haloperidol: fixed/flexible dose: not indicated, dose range: 6 mg to 24 mg/day, mean final dose: 12 mg/day. N = 10. 2. Sulpiride: fixed/flexible dose: not indicated, dose range: 800 mg to 3200 mg/day, mean final dose: 2000 mg/day. N = 10.   “During the first four to eight weeks, doses were gradually increased until an optimal therapeutic effect was attained. The optimal dose was maintained until the end of the 12‐week treatment period.” Rescue medication: allowed were “biperiden in case of extrapyramidal side‐effects, and diazepam or levomepromazine when sedation was required”
Outcomes	Examined: Mental state general: Brief Psychiatric Rating Scale (BPRS) (no SDs available). Unable to use: Leaving the study early due to adverse effects (number of participants randomised to the groups was not indicated). Adverse effects: special checklist (only data regarding the whole trial duration available).
Notes	Only the 20 study completers were included in the analyses. “Following a wash‐out period of 1‐6 weeks (until clear treatment‐demanding symptoms had developed), patients were randomly assigned to either sulpiride or haloperidol, and treated for 12 weeks.” The mean duration of the neuroleptic treatment was 6 years.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	“Randomly assigned”. According to e‐mail correspondence with the first author: “Referring to a random number table.”
Allocation concealment (selection bias)	Unclear risk	No further details.
Blinding (performance bias and detection bias) (performance bias)	Low risk	“Double‐blind”. “identically looking capsules”.
Blinding (performance bias and detection bias) (detection bias)	Low risk	“Double‐blind”. According to e‐mail correspondence with the first author the raters were blinded to treatments.
Incomplete outcome data (attrition bias) All outcomes	High risk	Regarding the whole trial duration, the overall attrition can be considered as being high (8 of 26 randomised participants). There were no information available, how many participants dropped out during the first phase of the trial. “Relatively small number of patients and the consequent Type II error.” The analyses were based on completers‐only data.
Selective reporting (reporting bias)	High risk	The outcome data were not fully addressed (no SDs for the BPRS). The numbers of participants randomised to each treatment group (haloperidol or sulpiride) were not indicated. Concerning the adverse effects reporting there were only data for the whole trial duration (both 12‐week periods) provided but not separately for the first phase which was of interest for this systematic review.
Other bias	High risk	“The study was carried out as a double‐blind cross‐over trial.” ”Dampening effect of the relatively high doses [of sulpiride] employed in this study, and the sample of chronic, long‐term hospitalised patients which was studied.”