Klapper 1997.

Methods	Prospective, randomised, double‐blind, parallel trial. Five half‐lives washout period. Total duration then 16 weeks. 4‐week single‐blind, placebo‐only baseline period. 12‐week treatment period Discontinuation rate: dropout: 25% of randomised participants Compliance (adherence) data: compliance data not reported Rule for use of acute medication: fewer than 3 times per week, but no information about permitted medications Methodological quality score: 4
Participants	Inclusion: migraine according to ICHD‐I; migraine onset at least 6 months prior to study; at least 2 attacks per month for previous 3 months Exclusion: daily headaches and analgesic overuse headaches were adequately excluded. No information on the exclusion of secondary headaches is reported. Other exclusions: more than 15 days of non‐migrainous headache per month, migraine symptoms without headache, pregnancy, inadequate contraception for women, previous treatment with valproate, failure of more than 2 other migraine prophylactic agents, significant medical or psychiatric disorder, many concomitant medications Setting: multi‐centre Country: USA and Canada 176 migraine patients were randomised; 91% had had attacks without aura and 40% with aura. 157 females and 19 males; age range 17 to 76
Interventions	Divalproex sodium 500 mg versus 1000 mg versus 1500 mg versus placebo (12 weeks). Dosage titrated for 4 weeks, then maintained for 8 weeks
Outcomes	Number of headache days per 28 days. Number of migraine attacks per 28 days. Frequency of non‐migraine headaches Time point(s) considered in the review: entire 3‐month treatment phase
Notes	Funders of the trial: Abbott Laboratories
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Description of method for random sequence generation (1:1:1:1 ratio within each study centre) is lacking
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Single‐blind baseline phase followed by double‐blind experimental phase. Placebo tablets were matched to verum
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention‐to‐treat analysis includes all randomised participants who provided headache data during the double‐blind phase
Selective reporting (reporting bias)	High risk	Dose comparisons not possible, as insufficient data were provided