Behavioral treatments of epilepsy: methodological characteristics and problems of published studies

Eleven studies that used operant or respondent conditioning procedures in the treatment of seizure disorders were evaluated primarily with respect to methodological considerations and secondarily with respect to procedure and outcome. The majority of studies used seizure frequency as the main dependent variable, but failed to adequately describe the observational procedure. Blind observers, unaware of experimental conditions and expected outcomes, were never used and measures of interobserver agreement were rarely provided. Ten studies reported data for a single subject. All of the investigations used within-subject analyses, but only three employed an experimental design adequate to demonstrate a functional relation between treatment and changes in seizure activity. A variety of treatments were used, some imprecisely described, and across studies seizure frequency was reduced in 14 of 15 subjects during treatment. However, because of obvious methodological inadequacies, these reductions cannot be conclusively attributed to treatment in the majority of cases. While the studies reviewed suggest that behavioral treatments of seizure disorders are potentially beneficial, further and more carefully controlled research is required to conclusively evaluate the generality and power of such treatments.