Methods	Randomised double-blind PCB-controlled parallel multicentre (71 in the USA and 5 in Canada) trial. 5 treatment arms: 1 PCB, 4 PGB. Patients randomised in blocks of 5, each allocated unique ID number. Capsule sizes varied (no further information available). Duration of baseline period: 8 weeks. There was no titration; 12-week treatment period
Participants	Patients 12 years and above (range 12 to 75 years, mean 38.4 years), 48.1% male, all with treatment-resistant partial epilepsy. Patients were on 1 to 4 baseline AEDs. 586 patients screened, 455 patients randomised: 100 patients to PCB (mean baseline seizure frequency: 22.3); 88 patients to 50 mg PGB (mean baseline seizure frequency: 27.4); 88 patients to 150 mg PGB (mean baseline 28-day seizure frequency: 23.1); 90 patients to 300 mg PGB (mean baseline 28-day seizure frequency: 19.1) and 89 patients to 600 mg PGB (mean baseline 28-day seizure frequency: 18.6)
Interventions	Group 1: PCB Group 2: 50 mg PGB daily Group 3: 150 mg PGB Group 4: 300 mg PGB Group 5: 600 mg PGB daily
Outcomes	Primary outcome: reduction in seizure frequency compared to baseline (response ratio) Secondary outcomes: responder rate, pairwise comparisons with PCB, adverse events
Notes	Blinding broken for interim analysis (data obtained were only known to committee who were not involved in further running of study) and for 1 patients who developed visual field defect. 2 patients were excluded from ITT analysis (1 withdrew consent, 1 had AEDs changed during baseline period). Seizure frequency and responder rate were calculated from data collected from seizure diaries and mean calculated over a 4-week period
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Used a computer-generated randomised schedule using block sizes of 5
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Study medication presented in identical capsules
Incomplete outcome data (attrition bias) All outcomes	Low risk	Attrition rates reported. ITT analysis employed
Selective reporting (reporting bias)	Unclear risk	All outcomes stated in methods section of paper were reported in the results; however, there was no protocol available to check a priori outcomes
Other bias	High risk	Possibility of the inclusion of patients with primary generalised epilepsy Sponsored by same pharmaceutical company as with all the other included trials