Electroencephalography during surgery with cardiopulmonary bypass and hypothermia

After more than 30 yr of use, electroencephalographic (EEG) monitoring during cardiopulmonary bypass has not gained wide clinical acceptance. To assess its utility to predict central nervous system injury, two-channel recordings were made from 78 patients undergoing cardiopulmonary bypass and anesthetized with fentanyl/diazepam/enflurane. The perfusion regimen included the use of high pump flow, a bubble oxygenator, and no arterial tubing filter. Target values were 28-32 degrees C for the minimum rectal temperature, 60-80 mmHg for mean arterial pressure, and 20-25% for hematocrit. Eight descriptors of the Fourier power spectra of the EEG were calculated off-line, and outcome comparisons were made with the results from neuropsychological tests. Among 58 patients yielding complete data of acceptable quality, a statistically significant reduction in total power was observed from prebypass to postbypass, accompanied by an increase in the fractional power in the theta and beta frequency bands and in the spectral edge frequency. The shifts in total and theta power were weakly associated with short-term but not with long-term changes in neuropsychological scores. Nearly 40% of the patients' EEGs were corrupted with electrical noise at some time during bypass. In 15 patients selected for having high-quality recordings and no neuropsychological deficit, an extensive statistical analysis failed to reveal any consistent variation in the EEG descriptors with hypothermia. Under the conditions studied, it appears that for other than gross signal dropout, the strong background variability in the EEG makes it have little value for detecting harbingers of brain injury.