Quantitative electroencephalographic monitoring during myocardial revascularization predicts postoperative disorientation and improves outcome.
We evaluated computerized quantitative electroencephalography for the intraoperative detection of cerebral dysfunction. The quantitative electroencephalogram was recorded continuously during 96 myocardial revascularizations involving hypothermic cardiopulmonary bypass using Cerebrovascular Intraoperative MONitor (CIMON) software. CIMON relies on an adaptive statistical approach to detect subtle, but clinically relevant, changes in electroencephalographic activity indicative of cerebrocortical dysfunction. Relative (percent of total) low-frequency (1.5 to 3.5 Hz) power was chosen as the single quantitative electroencephalographic descriptor because it is an established hallmark of cortical dysfunction and is surprisingly insensitive to moderate changes in body temperature and level of opioid anesthesia. Reference values for this measure were established for each patient after anesthetic induction before sternotomy. The large sample variance often seen in low-frequency power was dramatically decreased by using log-transformed data and allowing each patient to serve as his own control. Quantitative electroencephalographic changes in standard deviation units or z-scores were determined from the individualized reference self-norm. Prolonged (greater than 5 minutes) and statistically significant (greater than 3 standard deviation) focal increases in relative low-frequency power were temperature-corrected to determine a standardized cerebrocortical dysfunction time at 37 degrees C. (CDT37). In phase I (n = 48), this objective quantitative electroencephalogram-based numeric descriptor was used to predict neuropsychologic outcome. These CDT37 greater than 5-minute episodes occurred 38 times in 19 patients. The quantitative electroencephalogram-based descriptor predicted the occurrence of such disorientation (n = 14 or 29%) with a 68% false positive rate but only an 8% false negative rate. Since these intraoperative quantitative electroencephalographic episodes were often (19/38) associated with low (less than 50 mm Hg) pump pressures, phase II (n = 48) sought to correct the quantitative electroencephalographic abnormality and prevent postoperative disorientation by appropriate increases in cerebral perfusion. Although the number of episodes of quantitative electroencephalographic abnormality was similar (n = 31) in phase II, these ischemic events disappeared after prompt elevation of perfusion pressure. The phase II disorientation rate fell significantly (p less than 0.002) to 4%. Thus statistically significant increases in low-frequency electroencephalographic relative power persisting for a temperature-corrected duration of 5 minutes or more are a reliable means of alerting the surgical/anesthesia team to the presence of cerebrocortical dysfunction and provide a rational and objective basis for corrective intervention. This form of electroencephalographic monitoring appears to offer an opportunity for the timely correction of perfusion abnormalities or the administration of cerebroprotectant compounds.