Dissociating the effects of nitrous oxide on brain electrical activity using fixed order time series modeling.

A number of commonly used anesthetics, including nitrous oxide (N2O), are poorly detected by current electroencephalography (EEG)-based measures of anesthetic depth such as the bispectral index. Based on a previously elaborated theory of electrocortical rhythmogenesis we developed a physiologically inspired method of EEG analysis that was hypothesized to be more sensitive in detecting and characterizing N2O effect than the bispectral index, through its combined EEG estimates of cortical input and cortical state. By evaluating sevoflurane-induced loss of consciousness in the presence of low brain concentrations of N2O in 38 elective surgical patients, N2O was associated with a statistically significant reduction in the input the frontal cortex received from other cortical and subcortical areas. In contrast the bispectral index responded only to low, but not to high, concentrations of N2O.