Levisohn 2007

Methods	Pilot, randomized controlled trial
Participants	28 participants with juvenile myoclonic epilepsy
Interventions	Participants were assigned as 2:1 ratio to topiramate (19 participants) or valproate (9 participants) for 26 weeks. The topiramate target dosage was 3‐4 mg/kg/day (maximum 9 mg/kg/day) for participants aged 12‐16 years and 200 mg/day (maximum 600 mg/day) for participants aged > 16 years. Valproate target dosages were 10 mg/kg/day in participants aged 12‐16 years and 750 mg/day in participants aged >16 years (overall maximum 60 mg/kg/day). Medications were titrated at 1‐ to 2‐week intervals according to clinical response and were administered in divided doses
Outcomes	Seizure reduction, evaluation of improvement, systemic toxicity and neurotoxicity scores, adverse events
Notes	The baseline period was 3 months before study entry. A 14‐week titration phase was followed by a 12‐week maintenance phase
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Blinded randomization was achieved by providing study sites with individual envelopes containing medication assignments generated by computer
Allocation concealment (selection bias)	Low risk	Blinded randomization was achieved by providing study sites with individual envelopes containing medication assignments generated by computer
Blinding of participants and personnel (performance bias) All outcomes	Low risk	The participant, investigator and pharmacist remained blinded to medication assignment until screening was completed and the envelope was opened
Blinding of outcome assessment (detection bias) All outcomes	Low risk	The participant, investigator and pharmacist remained blinded to medication assignment until screening was completed and the envelope was opened
Incomplete outcome data (attrition bias) All outcomes	High risk	7 (37%) participants treated with topiramate and 2 (22%) participants treated with valproate discontinued before the endpoint
Selective reporting (reporting bias)	Unclear risk	No pre published protocol
Other bias	Low risk	No other bias was found