Network meta-analyses of antiepileptic drug efficacy and tolerability in drug-resistant focal epilepsies: a clinical perspective

Network meta-analysis (NMA) is a new technique that allows multiple treatment comparisons and provides estimates of effect sizes for all possible pair-wise comparisons. Several NMAs of antiepileptic drug (AED) efficacy and tolerability in individuals with refractory focal epilepsy, however, came to non-specific and, in some cases, divergent conclusions. We review some clinical factors that may be responsible for these inconsistent findings. A major issue is the small number of individuals included in the meta-analyses with consequent wide confidence intervals and lack of ability to achieve significant results. Further issues are lack of robustness of the measured efficacy outcome-the responder ratio (the percentage of individuals with a > 50 % improvement in seizure frequency); the selection of randomized studies (RCTs) included, i.e., the inclusion of studies with heterogeneous populations (children and adults); and inclusion of people treated with different doses of the experimental drug. Some methods of analysing data from RCTs, such as the last observation carried forward (LOCF) analysis, the choice of different phases of the study to compare to baseline, and the year in which the trial was conducted, selectively affect measurement of efficacy outcomes. Titration speed and other methodological aspects selectively affect tolerability. Several factors restrict the analysis of clinically useful estimates of the comparative efficacy of AEDs, while analysis of tolerability may be easier to accomplish.