The ketogenic diet, a treatment for intractable epilepsy, is rarely initiated because it requires strict compliance with a diet that is perceived to be unpalatable. In a case study of a 15-year-old girl with uncontrolled epilepsy, we used a stimulus-choice procedure to assess relative preferences of 33 foods from the diet and to develop two treatments based on Premack's principle. The results of a multielement analysis showed that both treatments increased dietary compliance. Compliance was maintained with generalization of treatment across settings, and was associated with a 40% reduction in seizures.DESCRIPTORS: ketogenic diet, epilepsy, diet compliance, Premack's principle, behavioral pediatrics, preferencesThe ketogenic diet can effectively treat some cases of intractable epilepsy, but it is rarely used because physicians and caregivers view the diet as unpalatable and assume that compliance will be low (Kinsman, Vining, Quaskey, Mellits, & Freeman, 1992). Strict adherence to restrictions on types and amounts of food that can be consumed is essential, because even minor deviations (e.g., sugar in toothpaste) can compromise treatment efficacy. In this case study of a 15-year-old girl with uncontrolled epilepsy, we used a stimulus-choice procedure (Fisher et al., 1992) to assess relative preferences of 33 food items from the diet. We then (a) compared the results from the stimulus-choice assessment with predictions about food preferences made by caregivers and (b) developed and compared two treatments based on Premack's principle (Premack, 1962), wherein higher preference ketogenic foods were presented contingent upon consumption of lower preference foods. The case adds to the growing body of literature on using operant methods to improve compliance with medical treatments (Christophersen, 1994).METHOD: Tasha, a 15-year-old female with severe retardation, was hospitalized for the treatment of aggression and a seizure disorder that was not adequately controlled with anticonvulsants. Tasha's caregivers were given a list of 33 ketogenic foods and asked to rank Tasha's preferences for these items. Next, we directly assessed Tasha's preferences for these foods by using the stimulus-choice procedure described by Fisher et al. (1992). During each trial, we presented Tasha with two food items from the diet. She chose one item from each pair and then consumed the selected item. An observer recorded her choice during each trial. We presented each item 32 times, once with every other item, for a total of 528 trials. Items were then ranked according to the percentage of trials in which Tasha selected a particular food. Interobserver agreement data were collected during 57.1% of the trials. Occurrence and nonoccurrence agreement coefficients averaged 91.3% and 94.6%, respectively.Assessment results were used to develop two treatments based on Premack's (1962) principle. In baseline, we provided Tasha with free access to all ketogenic foods presented during a given meal. In Treatment A, we gave her a bite-sized amount of on...