Consoli 2012.

Methods	Multicentre, open‐label randomised trial conducted in two centres in Italy 2 treatment arms: CBZ and LEV
Participants	Participants > 18 years with late post‐stroke seizures (2 weeks to 3 years after stroke) seen in the Cerebrovascular Unit between September 2008 and March 2009. No previous AED treatments were allowed except for emergency treatments Number randomised: CBZ = 66, LEV = 62. Number completing the trial: CBZ = 54, LEV = 52 58 male participants (55%) of those completing the trial 74 participants with partial epilepsy (74%) of those completing the trial Mean age of those completing trial (SD): CBZ = 69.7 (13.2), LEV = 74.1 (11.3)
Interventions	Monotherapy with CBZ or LEV 2‐week titration period to CBZ: 600 mg/d or LEV: 1000 mg/d Titration period followed by 52‐week maintenance period. Range of follow‐up not stated
Outcomes	Frequency of seizures during the treatment period Rentention of treatment from the first intake Changes in cognitive measures and quality‐of‐life measures at the end of the treatment period: Mini Mental Scale Examination to evaluate global cognitive functioning Logical Memory from the Wechsler Memory Scale‐Revised Visual Memory was assessed with the Benton visual memory test Digital Span Test for attention and some executive functions Stroop Test to investigate the inhibition process Raven’s Coloured Progressive Matrices Test for nonverbal reasoning Corsi span and supraspan learning test ADL index and the Instrumental‐ ADL (IADL) depression was assessed with the Geriatric Depression Scale Changes in EEG assessments at the end of the treatment period Tolerability of treatment
Notes	Contact made with trial author who provided additional information for one of the trial centres but full IPD dataset unavailable. Aggregate data extracted from graph of time to seizure recurrence in the publication. Trial was terminated early due to financial reasons when 128 out of a target 630 participants had been recruited
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation numbers were sequentially assigned across centres, and a computer‐generated randomisation scheme was used to provide balanced blocks of participants for each treatment group within each centre
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open label trial
Blinding of outcome assessment (detection bias) All outcomes	High risk	Open label trial
Incomplete outcome data (attrition bias) All outcomes	High risk	Attrition rates reported, only those who completed the trial were included in efficacy analyses. This is not an ITT approach
Selective reporting (reporting bias)	Low risk	All efficacy, cognitive and tolerability outcomes specified in the methods sections were reported well in the results section. No protocol was available.
Other bias	High risk	Likely that trial is underpowered from the early termination with 20% of target sample size recruited