Abstracts

RATIONALE: False-positive (FP) errors during recognition may confound memory testing during the intracarotid amobarbital procedure (IAP). Previous research found these errors were more frequent following left hemisphere injection and were associated with performance on neuropsychological indices of memory performance. The purpose of this study is to determine which clinical and neuropsychological variables predict the occurrence of these errors.
METHODS: 56 patients with diagnoses of partial epilepsy confirmed through continuous video-EEG monitoring (29 left frontotemporal onset; 27 right frontotemporal onset) completed the IAP as part of a comprehensive presurgical work-up. Memory for 8 objects and 4 line-drawings was evaluated through yes/no recognition testing. Additional memory testing with the California Verbal Learning Test (CVLT) was performed as part of a neuropsychological battery administered within a mean of 22 weeks of the procedure. Predictor variables were taken from demographic factors (Age, Sex, Education, & FSIQ), seizure variables (Focus & Age of Onset), IAP variables (Side of Injection, Speech Changes, & Level of Consciousness) and memory indices from the CVLT. Exploratory analyses were performed with a series of independent t-tests. Subsequent analyses included linear regression and computation of odds-ratios.
RESULTS: Nineteen patients (34%) had two or more FP errors during the IAP. These patients were significantly older than those with one or less FP errors. There was also a trend for lower education level and lower IQ. There was no relationship between FP errors and any of the IAP variables. The high FP error group exhibited significantly lower scores on learning and recall indices from the CVLT. They also showed a higher number of recall intrusion errors during learning trials and more FP errors on CVLT recognition testing. The results of a linear regression model found FP errors from the CVLT to be the single most significant predictor of FP errors during the IAP (F = 9.38, p [lt] .01). The correlation between these two sources of FP errors was in the moderate range (Pearson r = .384, p [gt] .01). Patients with less than 2 FP errors on the CVLT were at reduced risk for having a high number of FP errors on the IAP (Odds Ratio = 1.85, 95% CI = 1.21, 2.83).
CONCLUSIONS: The occurrence of FP errors appears to be more related to an enduring response-bias[dsquote] rather than to any demographic index, epilepsy variable, or factor associated with the procedure itself. Our findings indicate that recognition memory scores from neuropsychological testing provide a valid prediction of FP errors during the IAP.