How reliable is the Bispectral Index in critically ill patients? A prospective, comparative, single-blinded observer study*

OBJECTIVE: To establish a correlation between a reliable subjective measure, the Sedation-Agitation Scale (SAS), and an objective tool, the Bispectral Index (BIS), for monitoring critically ill patients with a decreased level of consciousness. DESIGN: Prospective, comparative, single-blinded observer study. SETTING: Surgical and medical intensive care units of the Tufts-New England Medical Center, a 349-bed tertiary care, academic medical center. PATIENTS: A convenience sample of 20 adult, critically ill patients with a decreased level of consciousness. The data from one patient were excluded because the patient did not meet inclusion criteria. MEASUREMENTS AND MAIN RESULTS: Patients were prospectively evaluated by a blinded observer using the SAS to subjectively determine their level of consciousness. Sedation levels varied from unarousable (SAS score of 1), to very sedated (SAS score of 2), to mildly sedated (SAS score of 3). Simultaneously, the patients were continuously monitored for 4-6 hrs with the BIS device. There was wide variability in BIS scores for any given level of consciousness as compared with the SAS. Unarousable patients had BIS scores ranging from 23 to 97, with a median score of 50 and an interquartile range of 24. Very sedated patients had BIS scores ranging from 35 to 98, with a median score of 68 and an interquartile range of 36. Mildly sedated patients had BIS scores ranging from 67 to 91, with a median score of 76 and an interquartile range of 8. Overall, there was a less than satisfactory correlation between BIS values and SAS scores (r =.36, p <.001). However, the correlation improved with subgroup analysis when BIS values associated with excessive muscle movement were excluded (r =.50, p <.001). CONCLUSIONS: The correlation between SAS and BIS scores was suboptimal and inconsistent in a heterogeneous group of critically ill patients. The generation of BIS hardware and software, studied herein, is neither reliable nor valid for routinely monitoring the level of consciousness in the critically ill patient. Excessive muscle movement by the patient is an important and spurious influence on BIS values and seriously undermines BIS reliability.