Rowan 2005

Methods	Randomised, double‐blind, parallel‐group trial conducted in 18 Veterans Affairs Medical Centres in the USA 3 treatment arms: LTG, CBZ and GBP
Participants	Adults > 60 years with newly diagnosed seizures, untreated or treated with sub‐therapeutic AED levels, with at least 1 seizure in the previous 3 months. Number randomised: CBZ = 198, GBP = 195, LTG = 200 570 male participants (96%) 446 participants with partial epilepsy (75%) Not stated how many participants had received previous AED treatment Mean age (years): CBZ = 71.9, GBP = 72.9, LTG = 71.9. Range not stated
Interventions	Monotherapy with CBZ, GBP, LTG 6‐week escalation phase leading to CBZ = 600 mg/d, GBP = 1500 mg/d, LTG = 150 mg/d Trial duration: 12 months. Range of follow‐up: not stated
Outcomes	Retention in the trial for 12 months Seizure freedom at 12 months Time to first, second, fifth and tenth seizure (time to seizures) Drug toxicity (incidence of systemic and neurologic toxicities) Serum drug levels and compliance Seizure‐free retention rates
Notes	IPD requested from trial sponsor, the Department of Veterans Affairs, USA. At the time of review, IPD has not been received. Aggregate data extracted from graphs in the publication
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Block randomisation (varying sizes) performed by site via a computer‐generated list
Allocation concealment (selection bias)	Low risk	Telephone randomisation used and pharmacy dispensed a prescription of the allocated drug (part of a blinded drug kit) to participants
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double blind achieved with double dummy tablets, doses of both increased and decreased simultaneously
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information provided
Incomplete outcome data (attrition bias) All outcomes	Low risk	Attrition rates reported. Most of the randomised participants included in analysis, 3 excluded due to site closure (not related to treatment)
Selective reporting (reporting bias)	Low risk	No protocol available but case report forms of data collected provided by the sponsor. Seizure outcomes and adverse events well reported
Other bias	Low risk	None identified