Abstracts

Rationale: Use of the invasive Intracarotid Amobarbital Procedure (IAP) is decreasing in epilepsy centers due to potential morbidity and difficulty in demonstrating its validity. Attempts to demonstrate predictive validity are hampered by the low base rate of postsurgical amnesia. In addition, less invasive neuropsychological testing (NP) can predict postsurgical material-specific memory decline. A more fruitful approach may be to examine the discriminant and convergent validity of the IAP using multi-trait, multi-method correlational analysis. The current study is the first known to us to use this approach to IAP validity. Methods: Participants were 102 temporal lobe epilepsy patients (left onset = 56, right onset = 45, onset undetermined in 1), with mean age of 34, mean education of 13 years. Included were 62 females and 94 right-handers. Participants underwent NP of left and right hemisphere function and a standard IAP utilizing unilateral injection of 125 mg sodium amobarbital (drug effect verified by EEG and hemiparesis) with 40-minute washout between hemispheres. Results: Left and right hemisphere functioning were considered “traits.” Inter-correlations between tests of the same trait reflect convergent validity while correlations with tests of different traits reflect discriminant validity. Statistically significant inter-correlations (range .29 to .97) were found between nearly all neuropsychological tests regardless of trait measured. In contrast, the IAP correlated significantly only with left hemisphere measures of language and memory following right injection and visuospatial memory following left injection. Conclusions: Correlations associated with the IAP are lower, but the IAP has better discriminant and convergent validity than NP. Interpretation of NP results can therefore be ambiguous. While the IAP may soon be replaced by less invasive functional neuroimaging, these procedures will need to demonstrate better convergent and discriminant validity than the IAP.