Types of studies

We included randomized controlled trials (RCTs) of topiramate versus placebo or other antiepileptic drug (AED) treatment for people with juvenile myoclonic epilepsy (JME). We included randomized cross‐over studies in meta‐analyses using the results from paired analyses (Elbourne 2002).

Types of participants

Types of interventions

• Experimental intervention: topiramate (monotherapy and add‐on studies)

• Control intervention: placebo or other AED treatment (different control groups were separately analyzed)

Types of outcome measures

We collected and analyzed the outcomes of all the participants initially randomized by intention‐to‐treat (ITT).

Electronic searches

Searches were run for the original review in July 2012 and subsequent searches were run on 2 November 2015, 21 February 2017, and 10 July 2018. For the latest update we searched the Cochrane Register of Studies (CRS Web) on 26 August 2021, and MEDLINE (Ovid; 1946 to 26 August 2021). See Appendix 1 and Appendix 2 for the respective search strategies. 

CRS Web includes randomized or quasi‐randomized, controlled trials from PubMed, Embase, ClinicalTrials.gov, the World Health Organization International Clinical Trials Registry Platform (ICTRP), the Cochrane Central Register of Controlled Trials (CENTRAL), and the Specialized Registers of Cochrane Review Groups, including Cochrane Epilepsy.

Previously we also searched Embase (1 July 2015), using the search strategy set out in Appendix 3. It is no longer necessary to search Embase, because randomized and quasi‐randomized controlled trials from Embase are now included in CRS Web.

Searching other resources

We also:

• searched reference lists of articles obtained from any source;

• searched conference proceedings likely to contain trials relevant to the review;

• contacted researchers, pharmaceutical companies and relevant trial authors to seek information about unpublished or incomplete trials.

We did not impose any language restrictions on our searches, and we attempted to obtain translations of articles where necessary.

Selection of studies

Two review authors (LJ, WL) independently evaluated titles and abstracts of identified trials to determine if they fulfilled the inclusion criteria. We obtained all potentially relevant studies for further consideration. We listed the excluded studies and reported the reasons for exclusion. We resolved any disagreements by discussion or by involving an independent party if necessary.

Data extraction and management

Two review authors (LJ, WL) independently extracted eligible data from the published reports onto standardized forms, and cross‐checked them for accuracy. We resolved disagreements regarding data extraction by discussion between the review authors.

We used checklists to record, independently, details of the following.

• Study design

• Total study duration

• Methods of generating randomization schedule

• Method of concealment of allocation

• Blinding

• Use of an ITT analysis (all participants initially randomized were included in the analyses as allocated to groups)

• AEs and dropouts for all reasons

• Participants (country, number of participants, age, gender, inclusion and exclusion criteria)

• Comparison (details of the intervention in treatment and control groups, details of co intervention(s) in both groups, duration of treatment)

• Outcomes and time points of measures (number of participants in each group and outcome, regardless of compliance)

• Factors for heterogeneity (sample size, missing participants, confidence interval (CI) and P value in measurement, subgroup analyses)

Assessment of risk of bias in included studies

Two review authors (LJ, WL) independently assessed the risk of bias for each trial using the Cochrane 'Risk of bias' tool, as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We assessed six specific domains including: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting and other bias. We judged the result in each domain as low, high, or unclear risk of bias.

Measures of treatment effect

We tried to evaluate the data from all the participants initially randomized as far as practically possible. We expressed continuous outcomes as mean difference (MD) with 95% CI. For outcomes commonly reported dichotomously, such as the proportion of responders, we used risk ratio (RR) with 95% CI to express the effect size. If a trial (or group within a trial) reported no AEs or dropouts, we calculated the risk difference (RD) instead of RR, with 95% CI.

Unit of analysis issues

We dealt with any unit of analysis issues using the methods outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011).

Dealing with missing data

We attempted to contact the principal investigator for further details if any data were missing. According to ITT analysis, all randomized participants were included. We planned to assign zero improvement of dichotomous outcomes for the withdrawals. We planned to consider different scenarios (best‐case and worst‐case) to account for missing data.

Assessment of heterogeneity

We planned to use the standard Chi² statistic and I² statistic (Higgins 2003) to measure heterogeneity (Deeks 2011), and make a judgement, along with visual inspection of forest plots. For the Chi² test, we would have rejected the hypothesis of tolerability if the P value was less than 0.10 and an I² greater than 50% would represent significant heterogeneity. In this case, we would have tried to explore factors for heterogeneity.

Assessment of reporting biases

We planned to assess the publication bias by funnel plot if we had found more than ten studies. However, we only included three studies in our review.

Data synthesis

If we found neither clinical nor statistical heterogeneity, we planned to pool results using a fixed‐effect model. We would have analyzed different controls separately. For statistical heterogeneity, we planned to incorporate the results into a random‐effects model. For heterogeneity that could not be readily explained, we would not have pooled the data but only given a description of the results.

Subgroup analysis and investigation of heterogeneity

We did not perform subgroup analyses due to the limited available data. We planned to analyze subgroups of studies categorized according to different doses and durations of topiramate. As a formal method of comparing subgroups, we planned to use the Chi² test (to test for significant differences between subgroups of participants). For all statistical analyses, we used Review Manager 5 (Review Manager 2014). 

Sensitivity analysis

We were unable to perform any sensitivity analyses due to the limited available data. We planned to use best‐case and worst‐case scenarios for taking into account missing data. We also planned to undertake the following sensitivity analyses to investigate unexplained heterogeneity and to test the robustness of results.

• Exclusion of cross‐over trials from the analysis

• Exclusion studies at high risk of bias, with inadequate allocation concealment or lack of blinded outcome assessor

• Comparison of fixed‐effect versus random‐effects models

Summary of findings and assessment of the certainty of the evidence

We have presented two 'Summary of findings' tables; one for each comparison (summary of findings Table 1; summary of findings Table 2). We reported all outcomes in the tables but made a general statement about the summary of findings for individual adverse events and graded the evidence based on consideration of the evidence of all of the adverse events (Schünemann 2011).

We determined the certainty of the evidence using the GRADE approach and downgraded evidence due to the presence of high risk of bias in at least one study, indirectness of the evidence, unexplained heterogeneity or inconsistency, imprecision of results, high probability of publication bias. We downgraded evidence by one level if we considered the limitation to be serious and by two levels if very serious (Schünemann 2013).