McElroy 1996.

Methods	Randomised single‐blind trial
Participants	Number of total participants/type of patients: 36 participants randomised. Inpatient study Inclusion criteria: Sex: male and female Age: 18 ‐ 65 years Diagnosis: DSM‐III‐R diagnosed bipolar disorder, manic or mixed episode with psychotic features Exclusion criteria: Those unable to provide informed consent Prior treatment with valproate Unstable medical condition History of seizures or neurological disorders Psychoactive substance dependence
Interventions	Location: USA, hospital site in Cincinnati Study duration: Not stated Treatment groups: 1. Valproate oral loading dose of 20 mg/day adjusted to achieve serum level of 50 micrograms/L 2. Haloperidol: 0.2 mg/kg/day in divided doses Concomitant medications: in participants receiving psychotropic medications prior to admission, these medications (other than lorazepam up to 4 mg/day) were discontinued upon admission to the unit Benztropine was administered as needed for treatment of extrapyramidal side effects Length of study: 6 days Randomisation: 21 patients randomised to valproate 15 patients randomised to haloperidol
Outcomes	Primary outcome: Changes on the YMRS Secondary outcomes: Changes in the global scores of the Scale for Assessment of Positive Symptoms (SAPS) and SAPS subscale scores, total dose of adjunctive lorazepam received per participant and length of hospital stay Adverse events were reported Study withdrawals were reported
Funding	"Support in part by grants from Abbott Laboratories and the Theorode and Vada Stanley Foundation"
Conflict of interest	Not stated
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Comment: No comment on randomisation procedure in text. We contacted the authors but received no reply.
Allocation concealment (selection bias)	Unclear risk	Comment: No comment on randomisation concealment in text. We contacted the authors but received no reply.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "Although the rater was blind to treatment, patients were not."
Blinding of outcome assessment (detection bias) Efficacy	High risk	Quote: "Although the rater was blind to treatment, patients were not." Quote: "The substantially higher rate of extrapyramidal side effects in the haloperidol treated patients may have compromised the raters blindness." Comment: 53.3% (n = 8) of the haloperidol participants had extrapyramidal side effects compared to none in the other group. This is likely to have compromised the rater blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comment: All participants completed the study.
Selective reporting (reporting bias)	Unclear risk	Comment: No pre‐published protocol found. We are therefore unable to assess the bias.
Other bias	High risk	Comment: To deal with extrapyramidal side effects, participants in the haloperidol group received benztropine. No participants in the valproate group received this medication. Differences between the 2 groups may potentially have been affected by this difference