Study of the Relationship Between Preferences for High Fat Food Items and Ketogenic and Modified Atkins Diet Efficacy Among Children with Seizure Disorders

Principal Investigator: Adrianna Amari

Although highly effective for the treatment of intractable epilepsy, the ketogenic diet is not always included in the treatment option hierarchy presented to families, in part due to perceptions that children will find the high-fat/low-carbohydrate regimen unpalatable. However, in a recent study, it was found that children with seizures exhibit preferences for foods compatible with the ketogenic diet (Amari et al., 2007). Specifically, children with seizures (ages 2-17 years) exhibited higher preferences for high fat items and lower preferences for high carbohydrate items in comparison to controls.  This detection of unique preferences compatible with the high-fat/low-carbohydrate diet in this clinical population warrants further research on underlying metabolic and neurophysiological differences as they are related to both seizure etiology and treatment. Perhaps more importantly, these preferences may have implications for treatment matching. 

In the proposed study, systematic food preference assessments will be conducted with children with seizures who are being started on the ketogenic or modified Atkins diets. Results will be correlated with diet efficacy (i.e., seizure control) at 1, 3, and 6 months. It is hypothesized that strong preferences for high fat foods and lower preferences for high carbohydrate foods will be positively correlated with diet efficacy. If true, this food preference assessment would have predictive value prior to using the ketogenic or modified Atkins diets.  Some clinicians and researchers have proposed that these diets be considered earlier in the treatment hierarchy, rather than as a “last resort”, and have advocated for research to identify variables that might predict the best responders to these diets. A positive food preference assessment performed early in the course of epilepsy in an individual child might identify these responders. Future research could then focus on modifying the assessment procedure (e.g., shortening in length) to be as clinically efficient as possible while maintaining good predictive validity.