Abstracts

Rationale: Epilepsy patients' valuation of their current health status is not routinely assessed in clinic settings, but is generally presumed based on measures such as seizure frequency, despite a growing national trend toward patient-centered care. Using two preference-based quality of life (QOL) metrics that allow quantification of patients' values for their current health status (Petitti 2000), we examined preference-based QOL as a function of seizure frequency, with the primary goal to assess whether preference-based QOL is linearly related to seizure frequency. Methods: One hundred eighty two adult patients with chronic epilepsy were consecutively sampled between 2006 and 2008 at Columbia Epilepsy Center for an in-person interview using visual analog (Bond MR 1966) and standard reference gamble metrics (Torrance 1987). Information regarding patients' epilepsy and mood were also obtained. Non-parametric test and correlation coefficients were used to examine the relationship between preference-based QOL versus ordinal categories of seizure frequency. As a sensitivity analysis, we performed a Monte-Carlo computer simulation to determine how often ordinal categories of seizure frequency failed to relate to QOL. We tested the reliability of the two metrics in a subset of patients who returned for a re-test, and also examined the validity against other proxy measures of disease. Finally, we tested significant predictors of preference-based QOL using multivariate logistic and linear regression analyses. Results: The relationship between preference-based QOL and seizure frequency was binary, in that only the patients who had been seizure free (for >1 year) had significantly higher QOL than those of any patients with recurrent seizure regardless of seizure frequency (p=0.003). Among patients with persistent seizures, QOL and seizure frequency were not monotonically linearly related, as patients with most frequent seizures did not report lowest QOL. Among patients with similar seizure frequency, QOL varied substantially. In addition, there were large overlaps in QOL across different seizure frequency categories. In a Monte Carlo simulation using QOL values from patients, we found that seizure frequency was poorly predictive of QOL about a third of the time. In multivariate regression models to determine whether patient factors accounted for the variation in QOL, the presence of depressive symptoms (β=-2.48, p=.000 for visual analog; β= -.161, p=.02 for standard reference gamble), and not seizure frequency, was the only independent predictor of quality of life, accounting for 25% to 34% of the variation of quality of life assessed through visual analog and standard reference gamble. Conclusions: Our findings underscore the singular importance of attaining complete seizure freedom. Given the lack of linear relation between QOL and seizure frequency, and a wide variation of QOL among patients experiencing similar seizure frequency, direct assessment of patients' valuation of epilepsy health status is needed in the clinic setting.