Kang 2007.
| Methods | Randomised, open‐label, observer‐blinded, parallel group trial; 28 weeks treatment phase | |
| Participants | 112 participants with clinical and EEG findings compatible with benign rolandic epilepsy from 12 centres in Korea, ages five to 15 years old | |
| Interventions | Topiramate (minimum dose 50‐75 mg/day, maximum 4 mg/kg/day) versus Carbamazepine (minimum dose 20 mg/kg/day, maximum 30 mg/kg/day) monotherapy | |
| Outcomes | 1) Outcomes and time‐points considered in the review: effects on seizure remission at 28 weeks, adverse effects of medication, effects on cognitive functioning. 2) Outcomes and time‐points measured (reported) in the study: changes on neuropsychological test battery after 28 weeks maintenance treatment; percentage of patients seizure free, treatment‐emergent adverse events | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | Quote: "Each study center received a separate and independent randomization plan". Comment: The method of randomisation was not specified. |
| Allocation concealment (selection bias) | Unclear risk | Comment: The method of allocation concealment was not described. |
| Incomplete outcome data (attrition bias) Objectively‐measured outcomes (e.g. seizure remission, cognition) | High risk | Quote: "the statistical analysis was based on the intent‐to‐treat population...the last observation...carried forward". Comment: Even though seizure freedom was analysed on an intention‐to‐treat basis, there is a risk of bias with carrying forward the last observation. The significant proportion of missing outcomes compared to observed event risk would be enough to induce clinically relevant bias in measurement of seizure remission. See sensitivity analysis Table 1. Quote: "none of the 25 patients that dropped out had cognitive complaints". Comment: The proportion of patients who had missing cognitive outcome measures is significant and although these patients had no complaints, cognitive testing detects some of the subtle domains of cognitive function. |
| Incomplete outcome data (attrition bias) Subjectively‐measured outcomes (e.g. adverse effects) | Low risk | The risk of bias for adverse effects is low having been accounted for in those who withdrew. |
| Blinding of participants and personnel (performance bias) Objectively‐measured outcomes (e.g. seizure remission, cognition) | High risk | Quote: "open‐label, observer‐blinded". Commment: The participants were not blinded. It was not specified if the observers who were blinded were those assessing effects on seizure outcome as well as those performing the neuropsychological tests. There is a high risk of differential behaviour which can impact on outcome. |
| Blinding of participants and personnel (performance bias) Subjectively‐measured outcomes (e.g. adverse effects) | High risk | The participants were not blinded. This is likely to affect their perception of adverse effects. |
| Blinding of outcome assessment (detection bias) Objectively‐measured outcomes (e.g. seizure remission, cognition) | Unclear risk | For seizure remission, the participants/families reporting seizure occurrences were not blinded and it is unclear if the observers assessing this were blinded. For cognition, the observers performing the neuropsychological tests were blinded and there would be a low risk of bias here. |
| Blinding of outcome assessment (detection bias) Subjectively‐measured outcomes (e.g. adverse effects) | High risk | As participants/families who were unblinded were involved in reporting on these outcomes, this is likely to affect outcome. |