Ten studies were included for review (n=87). Nine studies were within-subjects design (n=53); six of these studies used combinations of AB, ABA, ABAB and ABCA study designs (A for baseline condition, B for treatment and C for an optional additional treatment). One study had a mixed within-subjects and between-subjects design (n=34); only the within-subject data was used in the meta-analysis. Sample sizes ranged from three to 34. All but one study had a sample size of less than 10. Nine studies had no dropouts and one study had two dropouts from a total of 34 participants.
SMR and slow cortical potentials training significantly reduced the number of seizures post-intervention compared to pre-intervention levels (Hedges g -0.20, SE 0.04; 10 studies, n=87). Removing the largest study did not significantly alter the results (Hedges g -0.19, SE = 0.04; nine studies, n=53). There was no evidence of significant statistical heterogeneity. The results reported were for fixed-effects models; use of a random-effects model did not significantly alter the results.
Inspection of the funnel plot and testing for difference between the effect sizes of studies with small and large standard errors suggested the possibility of publication bias (p=0.06).