Fifty-four trials (n=11,106 patients), published in 55 articles, met the inclusion criteria. All the trials were placebo-controlled double blinded. Most of the trial used intention-to-treat analysis. Six publications were based on data from children, 46 used data from adults, and three publications involved combined datasets from children and adults.
For patient with refractory epilepsy who received adjunctive anti-epileptic drugs, the seizure-free rate was 8.2% (range 0 to 35%) compared with placebo (2.1%, range 0 to 17%). In terms of weighted risk difference, there was statistically significant seizure-free outcomes with adjunctive anti-epileptic drugs compared with placebo (RD 0.06, 95% CI 0.04 to 0.08; 30 trials; n=6,554 patients), but heterogeneity was significant (I2=74%).
Adjunctive anti-epileptic drugs led to 35% (range 14 to 60%) of patients with a 50% reduction in seizure rate versus 15% (range 0 to 39%) of patients on placebos. In the weighted-pooled analysis, adjunctive anti-epileptic drugs led to a statistically significant 50% reduction in seizure rate compared with placebo (RD 0.21, 95% CI 0.19 to 0.24; 54 trials, n=11,106 patients), but heterogeneity was significant (I2=57%).
Sensitivity analyses did not change the results of seizure-free and 50% seizure reduction. Meta-regression analyses also showed no difference in effect between year of publication (before 2001 versus 2001 and later) on seizure-free (p=0.60) or 50% seizure reduction (p=0.62).
There was evidence of publication bias for trials involving adults only, but not for trials of children only.