EEG Monitoring and Anesthesia

Introduction

• General anesthesia is a drug-induced, reversible alteration of brain state characterized by distinct changes in cortical electrophysiology and network connectivity. Clinically, anesthesia aims to achieve hypnosis/amnesia, antinociception, and immobility. The brain’s cortical response to anesthetic agents can be monitored using EEG.

Classification of Intraoperative EEG Applications

Monitoring Anesthetic Effects

• Anesthetic agents produce dynamic electrophysiological changes in the brain. EEG monitoring is used to assess the “anesthetic brain state” and guide anesthetic dosing. EEG signals are typically recorded from the frontal cortex using a small number of forehead electrodes (2–4 channels). The resulting summated electrical activity (raw EEG) can be decomposed using digital signal processing into quantitative EEG (qEEG) parameters, facilitating interpretation and tracking EEG changes over time (see EEG: Basic Principles and Overview). Commercially available pEEG monitors apply proprietary algorithms to generate a simplified numerical index of hypnotic state, allowing clinicians without formal training in raw EEG and qEEG interpretation to track the anesthetic state of the brain.^1,2 (See below for further details: “pEEG Indices of Anesthetic State”)

Monitoring for Brain Ischemia/Neuronal Injury

• During certain surgical procedures, the brain is at heightened risk for ischemic or other forms of injury (e.g., tumor craniotomy, aneurysm clipping, carotid revascularization). In these cases, a larger number of electrodes and EEG channels are used to detect intraoperative pathophysiologic changes such as ischemia or seizures. Increasing the number of channels improves spatial resolution and enhances the ability to identify focal neuronal injury. Anesthetic effects and homeostatic derangements may interfere with the accurate detection of neuronal injury. Accurate interpretation requires personnel who have both specialized training and credentials in EEG.^1,3,4 This summary will therefore focus primarily on EEG monitoring as it relates to anesthetic brain-state assessment.

General Anesthesia Mechanisms

• Understanding how anesthetic drugs alter brain function provides the foundation for interpreting EEG changes during anesthesia. These agents influence neuronal signaling at several levels.

Molecular Actions

• Common anesthetic drugs enhance inhibitory gamma-aminobutyric acid type A (GABA-A) receptor activity, reduce excitatory transmission through glutamate and nicotinic acetylcholine receptors, and activate potassium channels or alpha-2 adrenergic pathways. These effects shift neuronal populations toward hyperpolarization, decreasing their likelihood of firing. The result is a change in activity across cortical and subcortical structures, including the thalamus and brainstem arousal systems that normally support wakefulness and memory formation.^1,2,5

Network-Level Effects

• Beyond their molecular targets, anesthetics disrupt communication within and between large-scale brain networks. Thalamocortical and frontoparietal circuits are especially sensitive. As connectivity within these systems declines, the brain becomes less capable of integrating information over distance. Functional imaging and EEG studies reveal reduced coupling across networks, including the default mode, salience, and executive networks, accompanied by the emergence of more stereotyped oscillatory patterns. These include prominent frontal alpha activity shaped by thalamocortical loops and a gradual shift toward slower oscillations in the delta range. With higher doses, widespread suppression of cortical activity can progress to burst suppression.^1,2,5
• Taken together, these molecular and circuit-level changes explain the reversible loss of awareness, memory formation, movement, and nociceptive processing that characterize the anesthetized state.

Raw EEG vs qEEG

• Raw EEG records voltage fluctuations over time from electrodes placed on the scalp. Neuronal population activity exhibits rhythmic fluctuations across characteristic frequency ranges (Table 1). Raw EEG signals are typically decomposed using complex mathematical algorithms to derive qEEG parameters, which help clinicians track anesthetic brain state transitions, which are often marked by shifts in power between frequency ranges. Such high-yield qEEG parameters include power spectrum, spectral edge frequency (SEF-95), and spectrograms^1–3 (for more details, see summary: EEG: Basic Principles and Overview. Link).

EEG Manifestations of Anesthesia

EEG provides a real-time window into anesthetic effects:

• Concentration-dependent changes: Increasing anesthetic concentration generally reduces higher-frequency neuronal activity, and lower frequencies become more prominent. With propofol and volatile anesthetic agents the overall EEG power redistributes toward alpha and delta frequency bands (Figure 1). With increasing doses and deepening hypnosis, the SEF-95 and median frequency both decrease, providing continuous numeric indices of cortical depression. At sufficiently high doses, burst suppression or isoelectric EEG can be observed.^1,2,4
• Network connectivity changes: Slow delta oscillations disrupt temporal coordination and synchronization of neuronal activity between distant cortical regions. In contrast, increased power in alpha oscillations in thalamocortical circuits (thalamo-cortical reverberations) can block ascending sensory information from reaching the cortex. “Anteriorization of alpha” is a hallmark feature of loss of consciousness, characterized by the shifting of alpha activity from occipital to frontal regions, mainly due to an increase in thalamocortical reverberations. Likewise, the shifting of alpha activity back to the occipital regions is a marker of the return of consciousness.^1,2,5
• EEG markers of nociceptive states:
  Certain EEG parameters visible on spectrograms may provide additional information about insufficiently controlled nociceptive stimulation. Such parameters have limited clinical utility because experience in their interpretation is not widely available, and they are still under validation.⁵
  • Alpha dropout: When ascending nociceptive stimulation reaches the thalamus, this may disrupt alpha-band thalamocortical reverberations, resulting in alpha-dropout (a decrease of power in the alpha band), a change that can suggest inadequate antinociception.
  • Beta and delta-arousal: Increases in beta or delta power may also occur in response to nociceptive stimulation, indicating insufficient analgesia.

Fingerprints of Specific Anesthetic Agents

Each agent produces characteristic EEG or spectrogram signatures:

• Propofol and volatile anesthetic agents (predominantly GABA-A agonist effects): Prominent frontal alpha and delta activity, with age- and dose-dependent progression to burst suppression at high doses^2,4 (Figure 1).
• Ketamine (N-methyl-D-aspartate [NMDA]-receptor antagonist): Ketamine is unlike other hypnotic agents as it doesn’t act through GABA potentiation and does not uniformly suppress cortical activity. It primarily causes cortical disinhibition, leading to increased beta-gamma power (often extending beyond 50 Hz) and theta activity, which manifests as dissociative anesthesia with potential perceptual phenomena, often described as hallucinations.^2,4
• Dexmedetomidine (centrally acting alpha2-agonist): Produces a sleep-like EEG pattern, with increased power of physiologic spindle-range alpha (around 12-15 Hz) and slow-wave delta oscillations, reflecting its action on endogenous sleep pathways^2,4 (Figure 2).
• Nitrous oxide (predominant NMDA-receptor antagonistic effects): Produces reductions in delta and alpha power at lower concentrations. When added to inhalational agents, a transient increase in delta power is observed. It can modulate patterns of other agents by altering nociceptive input.^2,4
• Opioids (Mu-receptor agonists): Opioids, especially at higher doses, cause an overall shift towards lower frequencies by significantly reducing beta and gamma power. They cause a marked power increase in the delta frequency band, but cannot produce burst suppression. Some opioids have been found to produce epileptic spikes and be useful for seizure focus localization during electrocorticography for seizure focus ablation procedures.⁶
• Barbiturates: (GABA-A receptor agonists): Similar to other GABAergic agents, barbiturates produce progressive neuronal inhibition culminating in burst suppression and electrical silence at higher doses. They have received recognition for their putative role in neuro-protection by providing metabolic suppression and intracranial pressure control in the context of traumatic brain injury. However, the clinical utility of barbiturate coma is limited by increased risk of cardiovascular instability, and current evidence does not support improved clinical outcomes when used in this context. Desired levels of burst suppression can be achieved through continuous raw EEG monitoring, but titrating beyond burst suppression to achieve isoelectric EEG is not recommended. Advanced cardiovascular monitoring is required to minimize complications.^6,7

An understanding of these drug-specific EEG “fingerprints” can help clinicians tailor anesthetic administration, antinociceptive strategies, anticipate patient response, and plan emergence from anesthesia.

pEEG Indices of Anesthetic State

Commercial pEEG monitors (e.g., BIS®, Sedline®) utilize proprietary mathematical algorithms to convert multichannel data into numerical indices designed to correlate with the hypnotic state⁸ (Figure 3). Examples of such indices include:

• Bispectral index (BIS) – integrates spectral power, bispectral phase coupling, and burst suppression ratio. On a scale from 0-100, a range of 40-60 represents optimal hypnosis.
• Patient State Index (PSI) – integrates power, frequency, phase, and coherence features derived from multiple channels. On the scale of 0-100, a range of 25-50 is considered optimal.

Despite their clinical value, pEEG monitors have several important limitations:^1,8,9

1. Drug-specific responses: Different anesthetic agents produce distinct patterns of cortical activity, and generic algorithms often fail to capture these differences accurately. Ketamine or nitrous oxide, for example, can raise index values even when the patient is adequately anesthetized.
2. Population bias: Most proprietary systems were developed and validated in adults receiving primarily GABAergic anesthesia. As a result, their performance in children, older adults, or patients with neurologic disease is inconsistent.
3. Artifacts: Electromyographic activity, electrocautery, and ocular movements can all raise processed index values. These artifacts may give the impression of inadequate hypnosis even when the raw EEG is appropriate for the anesthetic state.
4. Individual variability: Baseline EEG characteristics vary considerably between patients. Two individuals receiving the same drug concentration may show very different spectral patterns or index values.
5. Consciousness versus arousal: Processed indices track cortical synchrony but cannot measure subjective experience. A patient may have dreaming-like activity or, rarely, connected consciousness even when the index appears reassuringly low.

pEEG should therefore be viewed as one component of anesthetic assessment rather than a stand-alone measure. With appropriate training, direct evaluation of the raw EEG and qEEG display often provides more reliable guidance than reliance on a single numerical index.^1,9

pEEG Monitoring and Clinical Utility

• pEEG monitoring is widely used to help guide anesthetic dosing and reduce the risk of awareness, particularly in high-risk settings. Its value is most apparent during total intravenous anesthesia and when neuromuscular blockade is used, and it is generally viewed as an accepted part of clinical practice.⁸
• There is some evidence that processed-index–guided anesthesia may modestly reduce postoperative delirium, particularly in older adults undergoing noncardiac, nonneurosurgical procedures, with reported risk reductions of roughly twenty to forty percent. However, the major randomized trials in cardiac and non-cardiac surgical settings, including ENGAGES and ENGAGES-Canada, respectively, relied on processed index targets rather than raw EEG interpretation. These studies did not demonstrate a reduction in delirium compared with usual care, despite anesthetic exposure and suppression time being lower in the processed-index–guided groups.¹⁰