An assessment of the validity of spectral entropy as a measure of sedation state in mechanically ventilated critically ill patients

To assess whether the Entropy Module (GE Healthcare, Helsinki, Finland), a device to measure hypnosis in anesthesia, is a valid measure of sedation state in critically ill patients by comparing clinically assessed sedation state with Spectral Entropy Prospective observational study. Teaching hospital general ICU. 30 intubated, mechanically ventilated patients without primary neurological diagnoses or drug overdose receiving continuous sedation. Monitoring of EEG and fEMG activity via forehead electrodes for up to 72h and assessments of conscious level using a modified Ramsay Sedation Scale. 475 trained observer assessments were made and compared with concurrent Entropy numbers. Median State (SE) and Response (RE) Entropy values decreased as Ramsay score increased, but wide variation occurred, especially in Ramsay 4-6 categories. Discrimination between different sedation scores [mean (SEM) P(K) value: RE 0.713 (0.019); SE 0.710 (0.019)] and between lighter (Ramsay 1-3) vs.deeper (Ramsay 4-6) sedation ranges was inadequate [P(K): RE 0.750 (0.025); SE 0.748 (0.025)]. fEMG power decreased with increasing Ramsay score but was often significant even at Ramsay 4-6 states. Frequent "on-off" effects occurred for both RE and SE, which were associated with fEMG activity. Values switched from low to high values even in deeply sedated patients. High Entropy values during deeper sedation were strongly associated with simultaneous high relative fEMG powers. Entropy of the frontal EEG does not discriminate sedation state adequately for clinical use in ICU patients. Facial EMG is a major confounder in clinical sedation ranges.