Electromyographic Activity Falsely Elevates the Bispectral Index

Case 1

A 35-yr-old healthy male volunteer participated in a study evaluating the pharmacodynamic interaction of remifentanil and propofol. The EEG was recorded continuously with a bipolar montage Fp1-A1,Fp2-A2 (international 10-20 system of electrode placement), using Zipprep electrodes (Aspect). Electrode impedance was kept less than 2 kΩ. Electroencephalographic recordings were performed with a A-1000 EEG monitor (Version 3.22, Aspect). The raw EEG signal was filtered with a 30 Hz low-pass filter and a 0.5 Hz high-pass filter. In addition to the processed EEG variables, the raw EEG signal was stored on a computer hard disk for further off-line analysis with a sampling rate of 256 Hz (Datalogger, Aspect). Propofol and remifentanil concentrations were ramped up–down to predetermined concentrations using a computer controlled infusion pump (STANPUMP). Before changing target concentrations the sedation–anesthetic depth was assessed by applying the following stimuli until the first response: calling his name, shaking the volunteer by the shoulder, painful electrical stimulus, insertion of a laryngeal mask, and direct laryngoscopy.

After having completed the initial assessment of anesthetic depth at a propofol concentration of 1 μg/ml (predicted), remifentanil was started with an effect compartment concentration of 3 ng/ml. Ten minutes later the volunteer was assessed and responded to soft voice at a BIS of 80. At 9:43 (fig. 1), the propofol effect compartment concentration was increased to 2 μg/ml. At 9:57, the volunteer did not respond to loud voice and shaking his shoulder but responded to the insertion of a laryngeal mask at a BIS of 54. At 10:02 the propofol effect compartment concentration was increased to 3 μg/ml. The bispectral index paradoxically increased (fig. 1). At 10:12, the patient underwent direct laryngoscopy without a response at a BIS of 84. At 10:17, the infusion pumps with propofol and remifentanil were stopped and the bispectral index paradoxically decreased.

A 36-yr-old healthy female volunteer participated in the same study. The EEG montage and the EEG monitor settings were the same as in case 1. After having completed the initial assessment of anesthetic depth at a remifentanil concentration of 1 ng/ml (predicted), propofol was started with an effect compartment concentration of 1 μg/ml. At 9:15 (fig. 2) the remifentanil effect compartment concentration was increased to 2 ng/ml. The bispectral index paradoxically increased. After a first 1 mg vecuronium bolus at 9:05 to minimize opioid-induced muscle rigidity, we gave a second 1 mg vecuronium bolus at 9:25 to prove the hypothesis whether a decrease of the bispectral index value can be noted when a muscle relaxant was given. After administration of vecuronium the BIS decreased to 69 (fig. 2). The conventional EEG parameters spectral edge frequency 95 and median EEG frequency filtered at 30 Hz again decreased appropriately with increasing anesthetic drug concentration.

The bispectral index is a multivariate measure, whose weighting coefficients are unpublished and proprietary. 1,3Traditionally the EEG is low-pass filtered at 30 to 40 Hz. 1To calculate the bispectral index requires inclusion of frequencies above 40 Hz, approaching frequencies generated by muscle activity (EMG).

Figures 3 and 4show the bispectral index, the EMG and spectral edge frequency 95 calculated by the A-1000monitor for cases 1 and 2 respectively. In each case the anomalous behavior is caused by EMG activity artifactually increasing the bispectral index. The spectral edge frequency 95 was not influenced by the EMG for two reasons. First, differential filtering is used inside the Aspect monitor before calculating the BIS (0–47 Hz) and before calculating spectral edge frequency (0–30 Hz).

Second, the EMG related power in the remaining higher frequencies is quite small in relation to the total power (≈0.5%), therefore EMG related changes are not well reflected by spectral edge frequency 95. In contrast the bispectral index includes a feature especially for these higher frequencies: the relative β ratio (log(P^{30-47 Hz}/P^{11-20 Hz})). 1 

Whereas increasing frontalis muscle activity is mostly associated with light anesthesia, 4in the described two cases the EMG activity increased after increasing drug concentrations. The appearance of opiate-induced muscle rigidity is a possible explanation.

Interestingly, the high-frequency components caused by the electrical muscle activity could easily be identified visually in the stored raw EEG, but not in the monitor raw EEG (fig. 1).

It has been suggested that fluctuations of the bispectral index may not be artifacts but rather changes in the clinical state of the subject to which the other EEG measures are not sensitive. 5In these two cases the bispectral index misclassified the anesthetic depth. In case 1 the volunteer underwent direct laryngoscopy without a response at a BIS of 84, while he responded before to a stimulus of lower intensity at a BIS of 54. In case 2 the BIS rapidly decreased from 85 to 58 after administration of vecuronium, despite constant target concentrations of propofol and remifentanil.

We conclude that EMG activity can falsely increase the bispectral index. This is not a problem when the patients are fully paralyzed, but should be considered when the bispectral monitor is used in the absence of muscle relaxants.