Factors determining response to antiepileptic drugs in randomized controlled trials. A systematic review and...

Because of the lack of head-to-head adjunctive-therapy trials of antiepileptic drugs (AEDs) in refractory partial epilepsy, meta-analyses of placebo-controlled randomized controlled trials (RCTs) represent a potentially important source of evidence to guide treatment decisions. However, such indirect comparisons raise various methodologic issues that may hamper their relevance.

All RCTs in adult refractory partial epilepsy were analyzed to assess whether efficacy outcomes are influenced by: characteristics of patients and trials ; use of last observation carried forward (LOCF) analysis; evaluation period (entire period versus maintenance period); and year of publication. A meta-analysis of these AEDs was then performed taking these factors into consideration.

Sixty-three RCTs evaluating 20 AEDs were included. The following variables influenced efficacy estimates: (1) responder rates correlated positively with duration of the entire treatment period (p = 0.038); (2) response to placebo was significantly greater in the maintenance period than in the entire treatment period (p = 0.005); (3) responder rates increased over the years both for AEDs (p < 0.001) and for placebo (p = 0.001); (4) LOCF analysis overestimated responder rates for AEDs (p < 0.001) and for placebo (p = 0.001) compared with completer-based analysis, and the overestimation correlated positively with withdrawal rates (p < 0.001). A meta-analysis of available data showed large differences in efficacy ranking in relation to dose selection and type of analysis, but these were mostly nonsignificant due to statistical power limitations.

Several methodologic issues hamper the relevance of indirect comparisons of AEDs in the adjunctive-therapy of refractory partial epilepsy. Some of these issues could be overcome by improved standardization in the reporting of efficacy outcomes.