Tohen 2002.

Methods	Randomised double‐blind trial
Participants	Number of total participants/type of patients: 251 participants randomised. Inpatient study Inclusion criteria: Sex: male and female Age: 18 ‐ 75 years Diagnosis: diagnosis of bipolar I disorder, manic or mixed episode, with or without psychotic features. Clinical diagnoses were confirmed with the Structured Clinical Interview for DSM‐IV. A score of ≥ 20 on the YMRS was required at both the screening visit and on the day of random assignment to study groups (baseline) Exclusion criteria: Serious and unstable medical illness, DSM‐IV substance dependence within the past 30 days (except nicotine or caffeine), documented history of intolerance to olanzapine or valproate, and treatment with lithium, an anticonvulsant, or an antipsychotic medication within 24 hours of random assignment to study groups
Interventions	Location: USA, 48 sites (cities not stated) Study duration: Not stated Treatment groups: 1. Olanzapine: 5 – 20 mg/day. The initial daily dose was 15 mg/day of olanzapine consistent with the recommendations of the manufacturers. Investigators made dose adjustments primarily on the basis of clinical response but also on plasma levels and adverse events. Participants who did not tolerate the minimum dose level for treatment (5 mg/day olanzapine or 500 mg/day valproate) were discontinued from participation in the study. Mean modal doses for olanzapine were 17.4 mg/day 2. Valproate: 500 – 2500 mg/day. The initial daily doses was 750 mg/day of valproate, consistent with the recommendations of the manufacturers. Investigators made dose adjustments primarily on the basis of clinical response but also on plasma levels (valproate levels of 50 ‐ 125 ug/ml were aimed for) and adverse events. Participants who did not tolerate the minimum dose level for treatment (5 mg/day olanzapine or 500 mg/day valproate) were discontinued from participation in the study. Mean modal dose of 1401.2 mg/day. Concomitant medication: Concomitant lorazepam use was restricted to a maximum dose of 2 mg/day, and administration was not allowed within 8 hours of the administration of a symptom rating scale. Benztropine was permitted to treat extrapyramidal symptoms up to a maximum of 2 mg/day throughout the course of the study. Benztropine was not allowed as prophylaxis for extrapyramidal symptoms. Length of study: 21 days Randomisation: 126 randomised to valproate 125 randomised to olanzapine
Outcomes	Primary outcome: Changes on YMRS from baseline to endpoint. In addition, clinical response was defined as ≥ 50% improvement in the YMRS score at endpoint Secondary outcome: HDRS Safety was assessed using adverse events measures (using Abnormal Involuntary Movement Scale, Barnes Akaesthesia Rating Scale, Simpson Angus Rating Scale) and by monitoring other laboratory test values (e.g. ECG results, weight changes) Study withdrawals were reported
Funding	"Sponsored by Lilly Research Laboratories"
Conflict of interest	Not stated
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Randomly assigned" Comment: We contacted the lead author, who confirmed that random numbers were computer‐generated.
Allocation concealment (selection bias)	Unclear risk	Comment: Not mentioned in text. We contacted the lead author who confirmed that the random allocation was determined centrally (not on site). However, it is still unclear in what way the randomisation sequence was concealed.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Double‐blind" Quote: "To maintain the blind all patients randomly assigned to receive olanzapine had blood drawn and sham 'divalproex' plasma level results were reported. [...] All investigators at the clinical sites and at Lily Research Laboratories remained blind to subjects' treatment assignment" Quote: "Blood was [...] shipped to an independent reference laboratory"
Blinding of outcome assessment (detection bias) Efficacy	Low risk	Comment: There is no mention in text of who conducted outcome assessments and whether the assessors were blind. However, we contacted the lead author who confirmed that the rater was blind.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comment: Dropout rates in valproate group were 35.7% and in the olanzapine group 31.2%. All groups were over 30% dropout rates, but dropout rates were comparable (less than two‐fold difference) and overall dropout rate below 75%.
Selective reporting (reporting bias)	Unclear risk	Comment: No pre‐published protocol found. We are therefore unable to assess the bias.
Other bias	Low risk	Comment: None identified