Study	Reason for exclusion
Bowden 2008	A 12‐week, randomised trial comparing valproate to lithium. Was unblinded
Campos 2010	Single‐site, parallel‐group, randomised, outcome assessor‐blinded trial. BD I patients according to the DSM‐IV‐TR, in depressive, manic/hypomanic or mixed episode, aged 18 to 35 years are eligible. Comparing lithium + carbamazepine vs lithium + valproate. However included all stages of bipolar, not specific to acute mania
Clothier 1992	Double‐blind, randomised comparison of valproate and lithium. Mania was not acute
Emrich 1992	Double‐blind, placebo‐controlled study of valproate in acute mania. Used an ABA design. Diagnostic criteria used were ICD 9
Findling 2002	A prospective study that looked at prophylactic dosing of high‐risk youths ‐ no acute mania
Jahangard 2012	A trial where patients on valproate prior to having ECT either had their valproate stopped or continued on valproate during ECT in a blinded manner. In line with our protocol, we exclude trials where all participants are on open‐label valproate prior to discontinuation randomisation
Keck 2005	Post hoc analysis of results obtained in a 12‐month, double‐blind, placebo‐controlled trial of valproate and lithium involving 372 participants. Original trial also ineligible as acutely manic phase was treated in an non‐randomised manner
Müller‐Oerlinghausen 2000	21‐day, randomised, double‐blind, parallel‐group, placebo‐controlled trial comparing valproate add‐on to neuroleptics with placebo add‐on. However they included a large number of participants without bipolar disorder (schizoaffective disorder)
Novartis 2007	A randomised, double‐blind, placebo‐controlled study comparing add‐on carbamazepine to placebo when added to a mood stabiliser. Mood stabiliser allocation was neither randomised nor blinded
Pavuluri 2010	Double‐blind, pseudo‐randomised, outpatient clinical trial with 66 children and adolescents. Participants were allocated to either risperidone or valproate. Randomisation used an alternating schedule, where each successive participant was allocated to the opposite treatment
Pavuluri 2012	6‐week, double‐blind, randomised trial of risperidone plus placebo versus valproate plus placebo MRI and pharmacological study. However participants were not acutely manic
Revicki 2005	Randomised clinical trial, 201 adults hospitalised with bipolar I manic or mixed episodes were randomised to valproate or lithium, in addition to usual psychiatric care, and followed for 1 year. Assessors were not blinded while participants were acutely manic
Sachs 2002	3‐week randomised, double‐blind, placebo‐controlled study included 156 bipolar disorder patients with a current manic or mixed episode who received a mood stabiliser (lithium or valproate) and placebo, risperidone, or haloperidol. Mood stabiliser allocation was not randomised
Sachs 2004	Double‐blind, randomised treatment with quetiapine plus lithium/valproate, or placebo plus lithium/valproate. Mood stabiliser allocation not randomised
Sidana 2012	3‐week, open‐label, randomised,comparative, parallel‐group study. Raters not blinded
Suppes 2007	30 hypomanic patients were randomised to receive oxcarbazepine or valproate as add‐on or monotherapy for 8 weeks. A rater blind to treatment assignment performed all symptom ratings. Treatment was otherwise open‐label
Walkup 2015	Randomised, controlled trial of 379 individuals aged 6 to 15 years. Randomised to lithium, valproate or risperidone. Continuation of TEAM trials that used patients that failed treatment in an earlier study
West 2011	Prospective 6‐week, double‐blind,placebo‐controlled, randomised outpatient medication treatment trial of risperidone versus valproate. Randomisation used an alternating schedule where each successive patient was allocated to the opposite treatment, therefore pseudo‐randomisation
Yatham 2004	Double‐blind, randomised comparison of treatment with quetiapine plus lithium/valproate or placebo plus lithium/valproate. The mood stabiliser allocation was not randomised
Yatham 2007	Double‐blind, randomised, 6 weeks trial of quetiapine (up to 800 mg/day) and lithium/valproate vs placebo and lithium/valproate. The mood stabiliser allocation was not randomised

BD: bipolar disorder; DSM: Diagnostic and Statistical Manual of Mental Disorders; ICD: International Classification of Diseases; mg: milligram