Use of the BIS monitor to detect onset of naturally occurring sleep.

Objectives: Inadvertent sleep episodes are a recognized complication of sleep deprivation. Although such events can be life threatening, no system currently exists to detect and prevent sleep onset. Because sleep shares electroencephalographic similarities with the anesthetized state, we hypothesized that the BIS monitor, a currently available EEG-based monitor of anesthetic depth, would detect the onset of physiologic sleep. To test our hypothesis, we monitored volunteers during the transition from waking to sleep.

Methods: Non-medicated volunteers were asked to lie down in a dark room for 30-minutes and fall asleep while attached to a BIS monitor located outside the room. A laptop computer was used to generate an audio tone inside the room. Speaker volume was adjusted to the lowest level detectable by the awake subject. Testing was begun by activating a computer to play a tone at random intervals. The subject was instructed to click a mouse connected to the computer upon hearing the tone to verify wakefulness. The session was terminated upon loss of response to three consecutive tones or after 30 minutes. Subjects were questioned afterwards regarding their perceptions of sleep during testing.

Results: 11 out of 28 self-described good sleepers could not sleep under testing conditions. BIS values for the remaining 17 fell from 96.4 +/- 2.1 to 86.5 +/- 0.79 (p < 0.01) upon sleep onset. All subjects responded to audio stimuli at BIS values >90, and were asleep either by subjective or objective report at BIS values <80. Three subjects retained the mouse-click response despite low BIS scores and subjective descriptions of sleep.

Conclusion: Although variability in the BIS value marking sleep onset was noted, the BIS monitor detected all episodes of sleep onset in our testing regimen. We conclude that a threshold BIS value can be defined to allow the BIS monitor to detect sleep onset.